url
stringlengths 60
61
| repository_url
stringclasses 1
value | labels_url
stringlengths 74
75
| comments_url
stringlengths 69
70
| events_url
stringlengths 67
68
| html_url
stringlengths 49
51
| id
int64 620M
2.29B
| node_id
stringlengths 18
32
| number
int64 153
6.9k
| title
stringlengths 9
244
| user
dict | labels
listlengths 0
3
| state
stringclasses 1
value | locked
bool 1
class | assignee
dict | assignees
listlengths 0
3
| milestone
dict | comments
listlengths 0
30
| created_at
timestamp[s] | updated_at
timestamp[s] | closed_at
null | author_association
stringclasses 4
values | active_lock_reason
null | draft
bool 2
classes | pull_request
dict | body
stringlengths 10
33.9k
⌀ | reactions
dict | timeline_url
stringlengths 69
70
| performed_via_github_app
null | state_reason
stringclasses 1
value | is_pull_request
bool 2
classes |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://api.github.com/repos/huggingface/datasets/issues/2730
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2730/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2730/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2730/events
|
https://github.com/huggingface/datasets/issues/2730
| 955,987,834
|
MDU6SXNzdWU5NTU5ODc4MzQ=
| 2,730
|
Update CommonVoice with new release
|
{
"login": "yjernite",
"id": 10469459,
"node_id": "MDQ6VXNlcjEwNDY5NDU5",
"avatar_url": "https://avatars.githubusercontent.com/u/10469459?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/yjernite",
"html_url": "https://github.com/yjernite",
"followers_url": "https://api.github.com/users/yjernite/followers",
"following_url": "https://api.github.com/users/yjernite/following{/other_user}",
"gists_url": "https://api.github.com/users/yjernite/gists{/gist_id}",
"starred_url": "https://api.github.com/users/yjernite/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/yjernite/subscriptions",
"organizations_url": "https://api.github.com/users/yjernite/orgs",
"repos_url": "https://api.github.com/users/yjernite/repos",
"events_url": "https://api.github.com/users/yjernite/events{/privacy}",
"received_events_url": "https://api.github.com/users/yjernite/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[
"cc @patrickvonplaten?",
"Does anybody know if there is a bundled link, which would allow direct data download instead of manual? \r\nSomething similar to: `https://voice-prod-bundler-ee1969a6ce8178826482b88e843c335139bd3fb4.s3.amazonaws.com/cv-corpus-6.1-2020-12-11/ab.tar.gz` ? cc @patil-suraj \r\n",
"Also see: https://github.com/common-voice/common-voice-bundler/issues/15"
] | 2021-07-29T15:59:59
| 2021-08-07T16:19:19
| null |
MEMBER
| null | null | null |
## Adding a Dataset
- **Name:** CommonVoice mid-2021 release
- **Description:** more data in CommonVoice: Languages that have increased the most by percentage are Thai (almost 20x growth, from 12 hours to 250 hours), Luganda (almost 9x growth, from 8 to 80), Esperanto (7x growth, from 100 to 840), and Tamil (almost 8x, from 24 to 220).
- **Paper:** https://discourse.mozilla.org/t/common-voice-2021-mid-year-dataset-release/83812
- **Data:** https://commonvoice.mozilla.org/en/datasets
- **Motivation:** More data and more varied. I think we just need to add configs in the existing dataset script.
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2730/reactions",
"total_count": 2,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 2,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2730/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2728
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2728/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2728/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2728/events
|
https://github.com/huggingface/datasets/issues/2728
| 955,892,970
|
MDU6SXNzdWU5NTU4OTI5NzA=
| 2,728
|
Concurrent use of same dataset (already downloaded)
|
{
"login": "PierreColombo",
"id": 22492839,
"node_id": "MDQ6VXNlcjIyNDkyODM5",
"avatar_url": "https://avatars.githubusercontent.com/u/22492839?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/PierreColombo",
"html_url": "https://github.com/PierreColombo",
"followers_url": "https://api.github.com/users/PierreColombo/followers",
"following_url": "https://api.github.com/users/PierreColombo/following{/other_user}",
"gists_url": "https://api.github.com/users/PierreColombo/gists{/gist_id}",
"starred_url": "https://api.github.com/users/PierreColombo/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/PierreColombo/subscriptions",
"organizations_url": "https://api.github.com/users/PierreColombo/orgs",
"repos_url": "https://api.github.com/users/PierreColombo/repos",
"events_url": "https://api.github.com/users/PierreColombo/events{/privacy}",
"received_events_url": "https://api.github.com/users/PierreColombo/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Launching simultaneous job relying on the same datasets try some writing issue. I guess it is unexpected since I only need to load some already downloaded file.",
"If i have two jobs that use the same dataset. I got :\r\n\r\n\r\n File \"compute_measures.py\", line 181, in <module>\r\n train_loader, val_loader, test_loader = get_dataloader(args)\r\n File \"/gpfsdswork/projects/rech/toto/intRAOcular/dataset_utils.py\", line 69, in get_dataloader\r\n dataset_train = load_dataset('paws', \"labeled_final\", split='train', download_mode=\"reuse_cache_if_exists\")\r\n File \"/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/load.py\", line 748, in load_dataset\r\n use_auth_token=use_auth_token,\r\n File \"/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/builder.py\", line 582, in download_and_prepare\r\n self._save_info()\r\n File \"/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/builder.py\", line 690, in _save_info\r\n self.info.write_to_directory(self._cache_dir)\r\n File \"/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/info.py\", line 195, in write_to_directory\r\n with open(os.path.join(dataset_info_dir, config.LICENSE_FILENAME), \"wb\") as f:\r\nFileNotFoundError: [Errno 2] No such file or directory: '/gpfswork/rech/toto/datasets/paws/labeled_final/1.1.0/09d8fae989bb569009a8f5b879ccf2924d3e5cd55bfe2e89e6dab1c0b50ecd34.incomplete/LICENSE'",
"You can probably have a solution much faster than me (first time I use the library). But I suspect some write function are used when loading the dataset from cache.",
"I have the same issue:\r\n```\r\nTraceback (most recent call last):\r\n File \"/dccstor/tslm/envs/anaconda3/envs/trf-a100/lib/python3.9/site-packages/datasets/builder.py\", line 652, in _download_and_prepare\r\n self._prepare_split(split_generator, **prepare_split_kwargs)\r\n File \"/dccstor/tslm/envs/anaconda3/envs/trf-a100/lib/python3.9/site-packages/datasets/builder.py\", line 1040, in _prepare_split\r\n with ArrowWriter(features=self.info.features, path=fpath) as writer:\r\n File \"/dccstor/tslm/envs/anaconda3/envs/trf-a100/lib/python3.9/site-packages/datasets/arrow_writer.py\", line 192, in __init__\r\n self.stream = pa.OSFile(self._path, \"wb\")\r\n File \"pyarrow/io.pxi\", line 829, in pyarrow.lib.OSFile.__cinit__\r\n File \"pyarrow/io.pxi\", line 844, in pyarrow.lib.OSFile._open_writable\r\n File \"pyarrow/error.pxi\", line 122, in pyarrow.lib.pyarrow_internal_check_status\r\n File \"pyarrow/error.pxi\", line 97, in pyarrow.lib.check_status\r\nFileNotFoundError: [Errno 2] Failed to open local file '/dccstor/tslm-gen/.cache/csv/default-387f1f95c084d4df/0.0.0/2dc6629a9ff6b5697d82c25b73731dd440507a69cbce8b425db50b751e8fcfd0.incomplete/csv-validation.arrow'. Detail: [errno 2] No such file or directory\r\nDuring handling of the above exception, another exception occurred:\r\nTraceback (most recent call last):\r\n File \"/dccstor/tslm/elron/tslm-gen/train.py\", line 510, in <module>\r\n main()\r\n File \"/dccstor/tslm/elron/tslm-gen/train.py\", line 246, in main\r\n datasets = prepare_dataset(dataset_args, logger)\r\n File \"/dccstor/tslm/elron/tslm-gen/data.py\", line 157, in prepare_dataset\r\n datasets = load_dataset(extension, data_files=data_files, split=dataset_split, cache_dir=dataset_args.dataset_cache_dir, na_filter=False, download_mode=dataset_args.dataset_generate_mode)\r\n File \"/dccstor/tslm/envs/anaconda3/envs/trf-a100/lib/python3.9/site-packages/datasets/load.py\", line 742, in load_dataset\r\n builder_instance.download_and_prepare(\r\n File \"/dccstor/tslm/envs/anaconda3/envs/trf-a100/lib/python3.9/site-packages/datasets/builder.py\", line 574, in download_and_prepare\r\n self._download_and_prepare(\r\n File \"/dccstor/tslm/envs/anaconda3/envs/trf-a100/lib/python3.9/site-packages/datasets/builder.py\", line 654, in _download_and_prepare\r\n raise OSError(\r\nOSError: Cannot find data file. \r\nOriginal error:\r\n[Errno 2] Failed to open local file '/dccstor/tslm-gen/.cache/csv/default-387f1f95c084d4df/0.0.0/2dc6629a9ff6b5697d82c25b73731dd440507a69cbce8b425db50b751e8fcfd0.incomplete/csv-validation.arrow'. Detail: [errno 2] No such file or directory\r\n```"
] | 2021-07-29T14:18:38
| 2021-08-02T07:25:57
| null |
CONTRIBUTOR
| null | null | null |
## Describe the bug
When launching several jobs at the same time loading the same dataset trigger some errors see (last comments).
## Steps to reproduce the bug
export HF_DATASETS_CACHE=/gpfswork/rech/toto/datasets
for MODEL in "bert-base-uncased" "roberta-base" "distilbert-base-cased"; do # "bert-base-uncased" "bert-large-cased" "roberta-large" "albert-base-v1" "albert-large-v1"; do
for TASK_NAME in "mrpc" "rte" 'imdb' "paws" "mnli"; do
export OUTPUT_DIR=${MODEL}_${TASK_NAME}
sbatch --job-name=${OUTPUT_DIR} \
--gres=gpu:1 \
--no-requeue \
--cpus-per-task=10 \
--hint=nomultithread \
--time=1:00:00 \
--output=jobinfo/${OUTPUT_DIR}_%j.out \
--error=jobinfo/${OUTPUT_DIR}_%j.err \
--qos=qos_gpu-t4 \
--wrap="module purge; module load pytorch-gpu/py3/1.7.0 ; export HF_DATASETS_OFFLINE=1; export HF_DATASETS_CACHE=/gpfswork/rech/toto/datasets; python compute_measures.py --seed=$SEED --saving_path=results --batch_size=$BATCH_SIZE --task_name=$TASK_NAME --model_name=/gpfswork/rech/toto/transformers_models/$MODEL"
done
done
```python
# Sample code to reproduce the bug
dataset_train = load_dataset('imdb', split='train', download_mode="reuse_cache_if_exists")
dataset_train = dataset_train.map(lambda e: tokenizer(e['text'], truncation=True, padding='max_length'),
batched=True).select(list(range(args.filter)))
dataset_val = load_dataset('imdb', split='train', download_mode="reuse_cache_if_exists")
dataset_val = dataset_val.map(lambda e: tokenizer(e['text'], truncation=True, padding='max_length'),
batched=True).select(list(range(args.filter, args.filter + 5000)))
dataset_test = load_dataset('imdb', split='test', download_mode="reuse_cache_if_exists")
dataset_test = dataset_test.map(lambda e: tokenizer(e['text'], truncation=True, padding='max_length'),
batched=True)
```
## Expected results
I believe I am doing something wrong with the objects.
## Actual results
Traceback (most recent call last):
File "/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/builder.py", line 652, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/builder.py", line 983, in _prepare_split
check_duplicates=True,
File "/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/arrow_writer.py", line 192, in __init__
self.stream = pa.OSFile(self._path, "wb")
File "pyarrow/io.pxi", line 829, in pyarrow.lib.OSFile.__cinit__
File "pyarrow/io.pxi", line 844, in pyarrow.lib.OSFile._open_writable
File "pyarrow/error.pxi", line 122, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 97, in pyarrow.lib.check_status
FileNotFoundError: [Errno 2] Failed to open local file '/gpfswork/rech/tts/unm25jp/datasets/paws/labeled_final/1.1.0/09d8fae989bb569009a8f5b879ccf2924d3e5cd55bfe2e89e6dab1c0b50ecd34.incomplete/paws-test.arrow'. Detail: [errno 2] No such file or directory
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "compute_measures.py", line 181, in <module>
train_loader, val_loader, test_loader = get_dataloader(args)
File "/gpfsdswork/projects/rech/toto/intRAOcular/dataset_utils.py", line 69, in get_dataloader
dataset_train = load_dataset('paws', "labeled_final", split='train', download_mode="reuse_cache_if_exists")
File "/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/load.py", line 748, in load_dataset
use_auth_token=use_auth_token,
File "/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/builder.py", line 575, in download_and_prepare
dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
File "/gpfslocalsup/pub/anaconda-py3/2020.02/envs/pytorch-gpu-1.7.0/lib/python3.7/site-packages/datasets/builder.py", line 658, in _download_and_prepare
+ str(e)
OSError: Cannot find data file.
Original error:
[Errno 2] Failed to open local file '/gpfswork/rech/toto/datasets/paws/labeled_final/1.1.0/09d8fae989bb569009a8f5b879ccf2924d3e5cd55bfe2e89e6dab1c0b50ecd34.incomplete/paws-test.arrow'. Detail: [errno 2] No such file or directory
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: datasets==1.8.0
- Platform: linux (jeanzay)
- Python version: pyarrow==2.0.0
- PyArrow version: 3.7.8
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2728/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2728/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2719
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2719/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2719/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2719/events
|
https://github.com/huggingface/datasets/issues/2719
| 953,932,416
|
MDU6SXNzdWU5NTM5MzI0MTY=
| 2,719
|
Use ETag in streaming mode to detect resource updates
|
{
"login": "severo",
"id": 1676121,
"node_id": "MDQ6VXNlcjE2NzYxMjE=",
"avatar_url": "https://avatars.githubusercontent.com/u/1676121?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/severo",
"html_url": "https://github.com/severo",
"followers_url": "https://api.github.com/users/severo/followers",
"following_url": "https://api.github.com/users/severo/following{/other_user}",
"gists_url": "https://api.github.com/users/severo/gists{/gist_id}",
"starred_url": "https://api.github.com/users/severo/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/severo/subscriptions",
"organizations_url": "https://api.github.com/users/severo/orgs",
"repos_url": "https://api.github.com/users/severo/repos",
"events_url": "https://api.github.com/users/severo/events{/privacy}",
"received_events_url": "https://api.github.com/users/severo/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 3470211881,
"node_id": "LA_kwDODunzps7O1zsp",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset-viewer",
"name": "dataset-viewer",
"color": "E5583E",
"default": false,
"description": "Related to the dataset viewer on huggingface.co"
}
] |
open
| false
| null |
[] | null |
[] | 2021-07-27T14:17:09
| 2021-10-22T09:36:08
| null |
CONTRIBUTOR
| null | null | null |
**Is your feature request related to a problem? Please describe.**
I want to cache data I generate from processing a dataset I've loaded in streaming mode, but I've currently no way to know if the remote data has been updated or not, thus I don't know when to invalidate my cache.
**Describe the solution you'd like**
Take the ETag of the data files into account and provide it (directly or through a hash) to give a signal that I can invalidate my cache.
**Describe alternatives you've considered**
None
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2719/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2719/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2714
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2714/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2714/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2714/events
|
https://github.com/huggingface/datasets/issues/2714
| 952,580,820
|
MDU6SXNzdWU5NTI1ODA4MjA=
| 2,714
|
add more precise information for size
|
{
"login": "pennyl67",
"id": 1493902,
"node_id": "MDQ6VXNlcjE0OTM5MDI=",
"avatar_url": "https://avatars.githubusercontent.com/u/1493902?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/pennyl67",
"html_url": "https://github.com/pennyl67",
"followers_url": "https://api.github.com/users/pennyl67/followers",
"following_url": "https://api.github.com/users/pennyl67/following{/other_user}",
"gists_url": "https://api.github.com/users/pennyl67/gists{/gist_id}",
"starred_url": "https://api.github.com/users/pennyl67/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/pennyl67/subscriptions",
"organizations_url": "https://api.github.com/users/pennyl67/orgs",
"repos_url": "https://api.github.com/users/pennyl67/repos",
"events_url": "https://api.github.com/users/pennyl67/events{/privacy}",
"received_events_url": "https://api.github.com/users/pennyl67/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"We already have this information in the dataset_infos.json files of each dataset.\r\nMaybe we can parse these files in the backend to return their content with the endpoint at huggingface.co/api/datasets\r\n\r\nFor now if you want to access this info you have to load the json for each dataset. For example:\r\n- for a dataset on github like `squad` \r\n- https://raw.githubusercontent.com/huggingface/datasets/master/datasets/squad/dataset_infos.json\r\n- for a community dataset on the hub like `lhoestq/squad`:\r\n https://huggingface.co/datasets/lhoestq/squad/resolve/main/dataset_infos.json"
] | 2021-07-26T07:11:03
| 2021-07-26T09:16:25
| null |
NONE
| null | null | null |
For the import into ELG, we would like a more precise description of the size of the dataset, instead of the current size categories. The size can be expressed in bytes, or any other preferred size unit. As suggested in the slack channel, perhaps this could be computed with a regex for existing datasets.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2714/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2714/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2699
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2699/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2699/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2699/events
|
https://github.com/huggingface/datasets/issues/2699
| 950,221,226
|
MDU6SXNzdWU5NTAyMjEyMjY=
| 2,699
|
cannot combine splits merging and streaming?
|
{
"login": "eyaler",
"id": 4436747,
"node_id": "MDQ6VXNlcjQ0MzY3NDc=",
"avatar_url": "https://avatars.githubusercontent.com/u/4436747?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/eyaler",
"html_url": "https://github.com/eyaler",
"followers_url": "https://api.github.com/users/eyaler/followers",
"following_url": "https://api.github.com/users/eyaler/following{/other_user}",
"gists_url": "https://api.github.com/users/eyaler/gists{/gist_id}",
"starred_url": "https://api.github.com/users/eyaler/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/eyaler/subscriptions",
"organizations_url": "https://api.github.com/users/eyaler/orgs",
"repos_url": "https://api.github.com/users/eyaler/repos",
"events_url": "https://api.github.com/users/eyaler/events{/privacy}",
"received_events_url": "https://api.github.com/users/eyaler/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! That's missing indeed. We'll try to implement this for the next version :)\r\n\r\nI guess we just need to implement #2564 first, and then we should be able to add support for splits combinations",
"is there an update on this? ran into the same issue on 2.17.1.\r\n\r\nOn a similar note, the keyword `split=\"all\"` also does not work as intended when `streaming=True`. ",
"No update so far, especially since we haven't implemented an efficient way to query `split=train[50%:]` for example. The addition of two splits should be easy though, since we have `concatenate_datasets()`",
"Can you concatenate_datasets that are being streamed now? I was led to believe concatenation works on non streaming datasets only.",
"Yes `concatenate_datasets` works for datasets loaded in streaming mode as well"
] | 2021-07-22T01:13:25
| 2024-04-08T13:26:46
| null |
NONE
| null | null | null |
this does not work:
`dataset = datasets.load_dataset('mc4','iw',split='train+validation',streaming=True)`
with error:
`ValueError: Bad split: train+validation. Available splits: ['train', 'validation']`
these work:
`dataset = datasets.load_dataset('mc4','iw',split='train+validation')`
`dataset = datasets.load_dataset('mc4','iw',split='train',streaming=True)`
`dataset = datasets.load_dataset('mc4','iw',split='validation',streaming=True)`
i could not find a reference to this in the documentation and the error message is confusing. also would be nice to allow streaming for the merged splits
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2699/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2699/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2670
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2670/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2670/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2670/events
|
https://github.com/huggingface/datasets/issues/2670
| 947,120,709
|
MDU6SXNzdWU5NDcxMjA3MDk=
| 2,670
|
Using sharding to parallelize indexing
|
{
"login": "ggdupont",
"id": 5583410,
"node_id": "MDQ6VXNlcjU1ODM0MTA=",
"avatar_url": "https://avatars.githubusercontent.com/u/5583410?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ggdupont",
"html_url": "https://github.com/ggdupont",
"followers_url": "https://api.github.com/users/ggdupont/followers",
"following_url": "https://api.github.com/users/ggdupont/following{/other_user}",
"gists_url": "https://api.github.com/users/ggdupont/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ggdupont/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ggdupont/subscriptions",
"organizations_url": "https://api.github.com/users/ggdupont/orgs",
"repos_url": "https://api.github.com/users/ggdupont/repos",
"events_url": "https://api.github.com/users/ggdupont/events{/privacy}",
"received_events_url": "https://api.github.com/users/ggdupont/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[] | 2021-07-18T21:26:26
| 2021-10-07T13:33:25
| null |
CONTRIBUTOR
| null | null | null |
**Is your feature request related to a problem? Please describe.**
Creating an elasticsearch index on large dataset could be quite long and cannot be parallelized on shard (the index creation is colliding)
**Describe the solution you'd like**
When working on dataset shards, if an index already exists, its mapping should be checked and if compatible, the indexing process should continue with the shard data.
Additionally, at the end of the process, the `_indexes` dict should be send back to the original dataset object (from which the shards have been created) to allow to use the index for later filtering on the whole dataset.
**Describe alternatives you've considered**
Each dataset shard could created independent partial indices. then on the whole dataset level, indices should be all referred in `_indexes` dict and be used in querying through `get_nearest_examples()`. The drawback is that the scores will be computed independently on the partial indices leading to inconsistent values for most scoring based on corpus level statistics (tf/idf, BM25).
**Additional context**
The objectives is to parallelize the index creation to speed-up the process (ie surcharging the ES server which is fine to handle large load) while later enabling search on the whole dataset.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2670/reactions",
"total_count": 4,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 2,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2670/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2657
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2657/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2657/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2657/events
|
https://github.com/huggingface/datasets/issues/2657
| 945,822,829
|
MDU6SXNzdWU5NDU4MjI4Mjk=
| 2,657
|
`to_json` reporting enhancements
|
{
"login": "stas00",
"id": 10676103,
"node_id": "MDQ6VXNlcjEwNjc2MTAz",
"avatar_url": "https://avatars.githubusercontent.com/u/10676103?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stas00",
"html_url": "https://github.com/stas00",
"followers_url": "https://api.github.com/users/stas00/followers",
"following_url": "https://api.github.com/users/stas00/following{/other_user}",
"gists_url": "https://api.github.com/users/stas00/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stas00/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stas00/subscriptions",
"organizations_url": "https://api.github.com/users/stas00/orgs",
"repos_url": "https://api.github.com/users/stas00/repos",
"events_url": "https://api.github.com/users/stas00/events{/privacy}",
"received_events_url": "https://api.github.com/users/stas00/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[] | 2021-07-15T23:32:18
| 2021-07-15T23:33:53
| null |
CONTRIBUTOR
| null | null | null |
While using `to_json` 2 things came to mind that would have made the experience easier on the user:
1. Could we have a `desc` arg for the tqdm use and a fallback to just `to_json` so that it'd be clear to the user what's happening? Surely, one can just print the description before calling json, but I thought perhaps it'd help to have it self-identify like you did for other progress bars recently.
2. It took me a while to make sense of the reported numbers:
```
22%|██▏ | 1536/7076 [12:30:57<44:09:42, 28.70s/it]
```
So iteration here happens to be 10K samples, and the total is 70M records. But the user does't know that, so the progress bar is perfect, but the numbers it reports are meaningless until one discovers that 1it=10K samples. And one still has to convert these in the head - so it's not quick. Not exactly sure what's the best way to approach this, perhaps it can be part of `desc`? or report M or K, so it'd be built-in if it were to print, e.g.:
```
22%|██▏ | 15360K/70760K [12:30:57<44:09:42, 28.70s/it]
```
or
```
22%|██▏ | 15.36M/70.76M [12:30:57<44:09:42, 28.70s/it]
```
(while of course remaining friendly to small datasets)
I forget if tqdm lets you add a magnitude identifier to the running count.
Thank you!
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2657/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2657/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2654
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2654/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2654/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2654/events
|
https://github.com/huggingface/datasets/issues/2654
| 945,167,231
|
MDU6SXNzdWU5NDUxNjcyMzE=
| 2,654
|
Give a user feedback if the dataset he loads is streamable or not
|
{
"login": "philschmid",
"id": 32632186,
"node_id": "MDQ6VXNlcjMyNjMyMTg2",
"avatar_url": "https://avatars.githubusercontent.com/u/32632186?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/philschmid",
"html_url": "https://github.com/philschmid",
"followers_url": "https://api.github.com/users/philschmid/followers",
"following_url": "https://api.github.com/users/philschmid/following{/other_user}",
"gists_url": "https://api.github.com/users/philschmid/gists{/gist_id}",
"starred_url": "https://api.github.com/users/philschmid/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/philschmid/subscriptions",
"organizations_url": "https://api.github.com/users/philschmid/orgs",
"repos_url": "https://api.github.com/users/philschmid/repos",
"events_url": "https://api.github.com/users/philschmid/events{/privacy}",
"received_events_url": "https://api.github.com/users/philschmid/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
|
{
"login": "mariosasko",
"id": 47462742,
"node_id": "MDQ6VXNlcjQ3NDYyNzQy",
"avatar_url": "https://avatars.githubusercontent.com/u/47462742?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mariosasko",
"html_url": "https://github.com/mariosasko",
"followers_url": "https://api.github.com/users/mariosasko/followers",
"following_url": "https://api.github.com/users/mariosasko/following{/other_user}",
"gists_url": "https://api.github.com/users/mariosasko/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mariosasko/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mariosasko/subscriptions",
"organizations_url": "https://api.github.com/users/mariosasko/orgs",
"repos_url": "https://api.github.com/users/mariosasko/repos",
"events_url": "https://api.github.com/users/mariosasko/events{/privacy}",
"received_events_url": "https://api.github.com/users/mariosasko/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "mariosasko",
"id": 47462742,
"node_id": "MDQ6VXNlcjQ3NDYyNzQy",
"avatar_url": "https://avatars.githubusercontent.com/u/47462742?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mariosasko",
"html_url": "https://github.com/mariosasko",
"followers_url": "https://api.github.com/users/mariosasko/followers",
"following_url": "https://api.github.com/users/mariosasko/following{/other_user}",
"gists_url": "https://api.github.com/users/mariosasko/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mariosasko/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mariosasko/subscriptions",
"organizations_url": "https://api.github.com/users/mariosasko/orgs",
"repos_url": "https://api.github.com/users/mariosasko/repos",
"events_url": "https://api.github.com/users/mariosasko/events{/privacy}",
"received_events_url": "https://api.github.com/users/mariosasko/received_events",
"type": "User",
"site_admin": false
}
] | null |
[
"#self-assign",
"I understand it already raises a `NotImplementedError` exception, eg:\r\n\r\n```\r\n>>> dataset = load_dataset(\"journalists_questions\", name=\"plain_text\", split=\"train\", streaming=True)\r\n\r\n[...]\r\nNotImplementedError: Extraction protocol for file at https://drive.google.com/uc?export=download&id=1CBrh-9OrSpKmPQBxTK_ji6mq6WTN_U9U is not implemented yet\r\n```\r\n"
] | 2021-07-15T09:07:27
| 2021-08-02T11:03:21
| null |
MEMBER
| null | null | null |
**Is your feature request related to a problem? Please describe.**
I would love to know if a `dataset` is with the current implementation streamable or not.
**Describe the solution you'd like**
We could show a warning when a dataset is loaded with `load_dataset('...',streaming=True)` when its lot streamable, e.g. if it is an archive.
**Describe alternatives you've considered**
Add a new metadata tag for "streaming"
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2654/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2654/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2649
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2649/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2649/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2649/events
|
https://github.com/huggingface/datasets/issues/2649
| 944,651,229
|
MDU6SXNzdWU5NDQ2NTEyMjk=
| 2,649
|
adding progress bar / ETA for `load_dataset`
|
{
"login": "stas00",
"id": 10676103,
"node_id": "MDQ6VXNlcjEwNjc2MTAz",
"avatar_url": "https://avatars.githubusercontent.com/u/10676103?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/stas00",
"html_url": "https://github.com/stas00",
"followers_url": "https://api.github.com/users/stas00/followers",
"following_url": "https://api.github.com/users/stas00/following{/other_user}",
"gists_url": "https://api.github.com/users/stas00/gists{/gist_id}",
"starred_url": "https://api.github.com/users/stas00/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/stas00/subscriptions",
"organizations_url": "https://api.github.com/users/stas00/orgs",
"repos_url": "https://api.github.com/users/stas00/repos",
"events_url": "https://api.github.com/users/stas00/events{/privacy}",
"received_events_url": "https://api.github.com/users/stas00/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"Is this done now? I see progress bars when using `load_dataset`.",
"There are progress bars when downloading data and when preparing them as Arrow files.\r\n\r\nThe \"total silence\" part mentioned in OP refer to checksums verifications which have had some changes in the latest release 2.10:\r\n- they're disabled by default (other less costly verifications are still done like checking the generated dataset size)\r\n- they're using a tqdm bar as well"
] | 2021-07-14T17:34:39
| 2023-03-27T10:32:49
| null |
CONTRIBUTOR
| null | null | null |
Please consider:
```
Downloading and preparing dataset oscar/unshuffled_deduplicated_en (download: 462.40 GiB, generated: 1.18 TiB, post-processed: Unknown size, total: 1.63 TiB) to cache/oscar/unshuffled_deduplicated_en/1.0.0/84838bd49d2295f62008383b05620571535451d84545037bb94d6f3501651df2...
HF google storage unreachable. Downloading and preparing it from source
```
and no indication whatsoever of whether things work well or when it'll be done. It's important to have an estimated completion time for when doing slurm jobs since some instances have a cap on run-time.
I think for this particular job it sat for 30min in total silence and then after 30min it started generating:
```
897850 examples [07:24, 10286.71 examples/s]
```
which is already great!
Request:
1. ETA - knowing how many hours to allocate for a slurm job
2. progress bar - helps to know things are working and aren't stuck and where we are at.
Thank you!
@lhoestq
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2649/reactions",
"total_count": 5,
"+1": 5,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2649/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2648
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2648/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2648/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2648/events
|
https://github.com/huggingface/datasets/issues/2648
| 944,484,522
|
MDU6SXNzdWU5NDQ0ODQ1MjI=
| 2,648
|
Add web_split dataset for Paraphase and Rephrase benchmark
|
{
"login": "bhadreshpsavani",
"id": 26653468,
"node_id": "MDQ6VXNlcjI2NjUzNDY4",
"avatar_url": "https://avatars.githubusercontent.com/u/26653468?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bhadreshpsavani",
"html_url": "https://github.com/bhadreshpsavani",
"followers_url": "https://api.github.com/users/bhadreshpsavani/followers",
"following_url": "https://api.github.com/users/bhadreshpsavani/following{/other_user}",
"gists_url": "https://api.github.com/users/bhadreshpsavani/gists{/gist_id}",
"starred_url": "https://api.github.com/users/bhadreshpsavani/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/bhadreshpsavani/subscriptions",
"organizations_url": "https://api.github.com/users/bhadreshpsavani/orgs",
"repos_url": "https://api.github.com/users/bhadreshpsavani/repos",
"events_url": "https://api.github.com/users/bhadreshpsavani/events{/privacy}",
"received_events_url": "https://api.github.com/users/bhadreshpsavani/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
|
{
"login": "bhadreshpsavani",
"id": 26653468,
"node_id": "MDQ6VXNlcjI2NjUzNDY4",
"avatar_url": "https://avatars.githubusercontent.com/u/26653468?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bhadreshpsavani",
"html_url": "https://github.com/bhadreshpsavani",
"followers_url": "https://api.github.com/users/bhadreshpsavani/followers",
"following_url": "https://api.github.com/users/bhadreshpsavani/following{/other_user}",
"gists_url": "https://api.github.com/users/bhadreshpsavani/gists{/gist_id}",
"starred_url": "https://api.github.com/users/bhadreshpsavani/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/bhadreshpsavani/subscriptions",
"organizations_url": "https://api.github.com/users/bhadreshpsavani/orgs",
"repos_url": "https://api.github.com/users/bhadreshpsavani/repos",
"events_url": "https://api.github.com/users/bhadreshpsavani/events{/privacy}",
"received_events_url": "https://api.github.com/users/bhadreshpsavani/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "bhadreshpsavani",
"id": 26653468,
"node_id": "MDQ6VXNlcjI2NjUzNDY4",
"avatar_url": "https://avatars.githubusercontent.com/u/26653468?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/bhadreshpsavani",
"html_url": "https://github.com/bhadreshpsavani",
"followers_url": "https://api.github.com/users/bhadreshpsavani/followers",
"following_url": "https://api.github.com/users/bhadreshpsavani/following{/other_user}",
"gists_url": "https://api.github.com/users/bhadreshpsavani/gists{/gist_id}",
"starred_url": "https://api.github.com/users/bhadreshpsavani/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/bhadreshpsavani/subscriptions",
"organizations_url": "https://api.github.com/users/bhadreshpsavani/orgs",
"repos_url": "https://api.github.com/users/bhadreshpsavani/repos",
"events_url": "https://api.github.com/users/bhadreshpsavani/events{/privacy}",
"received_events_url": "https://api.github.com/users/bhadreshpsavani/received_events",
"type": "User",
"site_admin": false
}
] | null |
[
"#take"
] | 2021-07-14T14:24:36
| 2021-07-14T14:26:12
| null |
CONTRIBUTOR
| null | null | null |
## Describe:
For getting simple sentences from complex sentence there are dataset and task like wiki_split that is available in hugging face datasets. This web_split is a very similar dataset. There some research paper which states that by combining these two datasets we if we train the model it will yield better results on both tests data.
This dataset is made from web NLG data.
All the dataset related details are provided in the below repository
Github link: https://github.com/shashiongithub/Split-and-Rephrase
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2648/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2648/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2643
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2643/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2643/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2643/events
|
https://github.com/huggingface/datasets/issues/2643
| 944,220,273
|
MDU6SXNzdWU5NDQyMjAyNzM=
| 2,643
|
Enum used in map functions will raise a RecursionError with dill.
|
{
"login": "jorgeecardona",
"id": 100702,
"node_id": "MDQ6VXNlcjEwMDcwMg==",
"avatar_url": "https://avatars.githubusercontent.com/u/100702?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jorgeecardona",
"html_url": "https://github.com/jorgeecardona",
"followers_url": "https://api.github.com/users/jorgeecardona/followers",
"following_url": "https://api.github.com/users/jorgeecardona/following{/other_user}",
"gists_url": "https://api.github.com/users/jorgeecardona/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jorgeecardona/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jorgeecardona/subscriptions",
"organizations_url": "https://api.github.com/users/jorgeecardona/orgs",
"repos_url": "https://api.github.com/users/jorgeecardona/repos",
"events_url": "https://api.github.com/users/jorgeecardona/events{/privacy}",
"received_events_url": "https://api.github.com/users/jorgeecardona/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"I'm running into this as well. (Thank you so much for reporting @jorgeecardona — was staring at this massive stack trace and unsure what exactly was wrong!)",
"Hi ! Thanks for reporting :)\r\n\r\nUntil this is fixed on `dill`'s side, we could implement a custom saving in our Pickler indefined in utils.py_utils.py\r\nThere is already a suggestion in this message about how to do it:\r\nhttps://github.com/uqfoundation/dill/issues/250#issuecomment-852566284\r\n\r\nLet me know if such a workaround could help, and feel free to open a PR if you want to contribute !",
"I have the same bug.\r\nthe code is as follows:\r\n\r\nthe error is: \r\n\r\n\r\nLook for the solution for this bug.",
"Hi ! I think your RecursionError comes from a different issue @BitcoinNLPer , could you open a separate issue please ?\r\n\r\nAlso which dataset are you using ? I tried loading `CodedotAI/code_clippy` but I get a different error\r\n```python\r\nTraceback (most recent call last):\r\n File \"<stdin>\", line 1, in <module>\r\n File \"/Users/quentinlhoest/Desktop/hf/datasets/src/datasets/load.py\", line 1615, in load_dataset\r\n **config_kwargs,\r\n File \"/Users/quentinlhoest/Desktop/hf/datasets/src/datasets/load.py\", line 1446, in load_dataset_builder\r\n builder_cls = import_main_class(dataset_module.module_path)\r\n File \"/Users/quentinlhoest/Desktop/hf/datasets/src/datasets/load.py\", line 101, in import_main_class\r\n module = importlib.import_module(module_path)\r\n File \"/Users/quentinlhoest/.virtualenvs/hf-datasets/lib/python3.7/importlib/__init__.py\", line 127, in import_module\r\n return _bootstrap._gcd_import(name[level:], package, level)\r\n File \"<frozen importlib._bootstrap>\", line 1006, in _gcd_import\r\n File \"<frozen importlib._bootstrap>\", line 983, in _find_and_load\r\n File \"<frozen importlib._bootstrap>\", line 967, in _find_and_load_unlocked\r\n File \"<frozen importlib._bootstrap>\", line 677, in _load_unlocked\r\n File \"<frozen importlib._bootstrap_external>\", line 728, in exec_module\r\n File \"<frozen importlib._bootstrap>\", line 219, in _call_with_frames_removed\r\n File \"/Users/quentinlhoest/.cache/huggingface/modules/datasets_modules/datasets/CodedotAI___code_clippy/d332f69d036e8c80f47bc9a96d676c3fa30cb50af7bb81e2d4d12e80b83efc4d/code_clippy.py\", line 66, in <module>\r\n url_elements = results.find_all(\"a\")\r\nAttributeError: 'NoneType' object has no attribute 'find_all'\r\n```"
] | 2021-07-14T09:16:08
| 2021-11-02T09:51:11
| null |
NONE
| null | null | null |
## Describe the bug
Enums used in functions pass to `map` will fail at pickling with a maximum recursion exception as described here: https://github.com/uqfoundation/dill/issues/250#issuecomment-852566284
In my particular case, I use an enum to define an argument with fixed options using the `TraininigArguments` dataclass as base class and the `HfArgumentParser`. In the same file I use a `ds.map` that tries to pickle the content of the module including the definition of the enum that runs into the dill bug described above.
## Steps to reproduce the bug
```python
from datasets import load_dataset
from enum import Enum
class A(Enum):
a = 'a'
def main():
a = A.a
def f(x):
return {} if a == a.a else x
ds = load_dataset('cnn_dailymail', '3.0.0')['test']
ds = ds.map(f, num_proc=15)
if __name__ == "__main__":
main()
```
## Expected results
The known problem with dill could be prevented as explained in the link above (workaround.) Since `HFArgumentParser` nicely uses the enum class for choices it makes sense to also deal with this bug under the hood.
## Actual results
```python
File "/home/xxxx/miniconda3/lib/python3.8/site-packages/dill/_dill.py", line 1373, in save_type
pickler.save_reduce(_create_type, (type(obj), obj.__name__,
File "/home/xxxx/miniconda3/lib/python3.8/pickle.py", line 690, in save_reduce
save(args)
File "/home/xxxx/miniconda3/lib/python3.8/pickle.py", line 558, in save
f(self, obj) # Call unbound method with explicit self
File "/home/xxxx/miniconda3/lib/python3.8/pickle.py", line 899, in save_tuple
save(element)
File "/home/xxxx/miniconda3/lib/python3.8/pickle.py", line 534, in save
self.framer.commit_frame()
File "/home/xxxx/miniconda3/lib/python3.8/pickle.py", line 220, in commit_frame
if f.tell() >= self._FRAME_SIZE_TARGET or force:
RecursionError: maximum recursion depth exceeded while calling a Python object
```
## Environment info
- `datasets` version: 1.8.0
- Platform: Linux-5.9.0-4-amd64-x86_64-with-glibc2.10
- Python version: 3.8.5
- PyArrow version: 3.0.0
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2643/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2643/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2642
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2642/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2642/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2642/events
|
https://github.com/huggingface/datasets/issues/2642
| 944,175,697
|
MDU6SXNzdWU5NDQxNzU2OTc=
| 2,642
|
Support multi-worker with streaming dataset (IterableDataset).
|
{
"login": "cccntu",
"id": 31893406,
"node_id": "MDQ6VXNlcjMxODkzNDA2",
"avatar_url": "https://avatars.githubusercontent.com/u/31893406?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/cccntu",
"html_url": "https://github.com/cccntu",
"followers_url": "https://api.github.com/users/cccntu/followers",
"following_url": "https://api.github.com/users/cccntu/following{/other_user}",
"gists_url": "https://api.github.com/users/cccntu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/cccntu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/cccntu/subscriptions",
"organizations_url": "https://api.github.com/users/cccntu/orgs",
"repos_url": "https://api.github.com/users/cccntu/repos",
"events_url": "https://api.github.com/users/cccntu/events{/privacy}",
"received_events_url": "https://api.github.com/users/cccntu/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! This is a great idea :)\r\nI think we could have something similar to what we have in `datasets.Dataset.map`, i.e. a `num_proc` parameter that tells how many processes to spawn to parallelize the data processing. \r\n\r\nRegarding AUTOTUNE, this could be a nice feature as well, we could see how to add it in a second step",
"Any update on this feature request?",
"Not yet, I'm happy to provide some guidance if someone wants to give it a shot though.\r\n\r\nThe code that applies the `map` function is in `iterable_dataset.py`, in `MappedExamplesIterable.__iter__`"
] | 2021-07-14T08:22:58
| 2024-05-03T10:11:04
| null |
CONTRIBUTOR
| null | null | null |
**Is your feature request related to a problem? Please describe.**
The current `.map` does not support multi-process, CPU can become bottleneck if the pre-processing is complex (e.g. t5 span masking).
**Describe the solution you'd like**
Ideally `.map` should support multi-worker like tfds, with `AUTOTUNE`.
**Describe alternatives you've considered**
A simpler solution is to shard the dataset and process it in parallel with pytorch dataloader. The shard does not need to be of equal size.
* https://pytorch.org/docs/stable/data.html#torch.utils.data.IterableDataset
**Additional context**
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2642/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2642/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2573
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2573/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2573/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2573/events
|
https://github.com/huggingface/datasets/issues/2573
| 934,584,745
|
MDU6SXNzdWU5MzQ1ODQ3NDU=
| 2,573
|
Finding right block-size with JSON loading difficult for user
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"This was actually a second error arising from a too small block-size in the json reader.\r\n\r\nFinding the right block size is difficult for the layman user"
] | 2021-07-01T08:48:35
| 2021-07-01T19:10:53
| null |
MEMBER
| null | null | null |
As reported by @thomwolf, while loading a JSON Lines file with "json" loading script, he gets
> json.decoder.JSONDecodeError: Extra data: line 2 column 1 (char 383)
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2573/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2573/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2547
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2547/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2547/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2547/events
|
https://github.com/huggingface/datasets/issues/2547
| 929,192,329
|
MDU6SXNzdWU5MjkxOTIzMjk=
| 2,547
|
Dataset load_from_disk is too slow
|
{
"login": "avacaondata",
"id": 35173563,
"node_id": "MDQ6VXNlcjM1MTczNTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/35173563?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/avacaondata",
"html_url": "https://github.com/avacaondata",
"followers_url": "https://api.github.com/users/avacaondata/followers",
"following_url": "https://api.github.com/users/avacaondata/following{/other_user}",
"gists_url": "https://api.github.com/users/avacaondata/gists{/gist_id}",
"starred_url": "https://api.github.com/users/avacaondata/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/avacaondata/subscriptions",
"organizations_url": "https://api.github.com/users/avacaondata/orgs",
"repos_url": "https://api.github.com/users/avacaondata/repos",
"events_url": "https://api.github.com/users/avacaondata/events{/privacy}",
"received_events_url": "https://api.github.com/users/avacaondata/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! It looks like an issue with the virtual disk you are using.\r\n\r\nWe load datasets using memory mapping. In general it makes it possible to load very big files instantaneously since it doesn't have to read the file (it just assigns virtual memory to the file on disk).\r\nHowever there happens to be issues with virtual disks (for example on spot instances), for which memory mapping does a pass over the entire file, and this takes a while. We are discussing about this issue here: #2252 \r\n\r\nMemory mapping is something handled by the OS so we can't do much about it, though we're still trying to figure out what's causing this behavior exactly to see what we can do.",
"Okay, that's exactly my case, with spot instances... Therefore this isn't something we can change in any way to be able to load the dataset faster? I mean, what do you do internally at huggingface for being able to use spot instances with datasets efficiently?",
"There are no solutions yet unfortunately.\r\nWe're still trying to figure out a way to make the loading instantaneous on such disks, I'll keep you posted"
] | 2021-06-24T12:45:44
| 2021-06-25T14:56:38
| null |
NONE
| null | null | null |
@lhoestq
## Describe the bug
It's not normal that I have to wait 7-8 hours for a dataset to be loaded from disk, as there are no preprocessing steps, it's only loading it with load_from_disk. I have 96 cpus, however only 1 is used for this, which is inefficient. Moreover, its usage is at 1%... This is happening in the context of a language model training, therefore I'm wasting 100$ each time I have to load the dataset from disk again (because the spot instance was stopped by aws and I need to relaunch it for example).
## Steps to reproduce the bug
Just get the oscar in spanish (around 150GGB) and try to first save in disk and then load the processed dataset. It's not dependent on the task you're doing, it just depends on the size of the text dataset.
## Expected results
I expect the dataset to be loaded in a normal time, by using the whole machine for loading it, I mean if you store the dataset in multiple files (.arrow) and then load it from multiple files, you can use multiprocessing for that and therefore don't waste so much time.
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.8.0
- Platform: Ubuntu 18
- Python version: 3.8
I've seen you're planning to include a streaming mode for load_dataset, but that only saves the downloading and processing time, that's not being a problem for me, you cannot save the pure loading from disk time, therefore that's not a solution for my use case or for anyone who wants to use your library for training a language model.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2547/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2547/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2538
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2538/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2538/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2538/events
|
https://github.com/huggingface/datasets/issues/2538
| 927,940,691
|
MDU6SXNzdWU5Mjc5NDA2OTE=
| 2,538
|
Loading partial dataset when debugging
|
{
"login": "reachtarunhere",
"id": 9061913,
"node_id": "MDQ6VXNlcjkwNjE5MTM=",
"avatar_url": "https://avatars.githubusercontent.com/u/9061913?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/reachtarunhere",
"html_url": "https://github.com/reachtarunhere",
"followers_url": "https://api.github.com/users/reachtarunhere/followers",
"following_url": "https://api.github.com/users/reachtarunhere/following{/other_user}",
"gists_url": "https://api.github.com/users/reachtarunhere/gists{/gist_id}",
"starred_url": "https://api.github.com/users/reachtarunhere/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/reachtarunhere/subscriptions",
"organizations_url": "https://api.github.com/users/reachtarunhere/orgs",
"repos_url": "https://api.github.com/users/reachtarunhere/repos",
"events_url": "https://api.github.com/users/reachtarunhere/events{/privacy}",
"received_events_url": "https://api.github.com/users/reachtarunhere/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi ! `load_dataset` downloads the full dataset once and caches it, so that subsequent calls to `load_dataset` just reloads the dataset from your disk.\r\nThen when you specify a `split` in `load_dataset`, it will just load the requested split from the disk. If your specified split is a sliced split (e.g. `\"train[:10]\"`), then it will load the 10 first rows of the train split that you have on disk.\r\n\r\nTherefore, as long as you don't delete your cache, all your calls to `load_dataset` will be very fast. Except the first call that downloads the dataset of course ^^",
"That’s a use case for the new streaming feature, no?",
"Hi @reachtarunhere.\r\n\r\nBesides the above insights provided by @lhoestq and @thomwolf, there is also a Dataset feature in progress (I plan to finish it this week): #2249, which will allow you, when calling `load_dataset`, to pass the option to download/preprocess/cache only some specific split(s), which will definitely speed up your workflow.\r\n\r\nIf this feature is interesting for you, I can ping you once it will be merged into the master branch.",
"Thanks all for responding.\r\n\r\nHey @albertvillanova \r\n\r\nThanks. Yes, I would be interested.\r\n\r\n@lhoestq I think even if a small split is specified it loads up the full dataset from the disk (please correct me if this is not the case). Because it does seem to be slow to me even on subsequent calls. There is no repeated downloading so it seems that the cache is working.\r\n\r\nI am not aware of the streaming feature @thomwolf mentioned. So I might need to read up on it.",
"@reshinthadithyan I use the .select function to have a fraction of indices.",
"If I want to create a dataset, containing only the 10 elements of a given dataset (slice it), how do I do that?",
"```python \r\nsmall_ds = ds.select(range(10))\r\n```",
"\r\n\r\n> ```python\r\n> small_ds = ds.select(range(10))\r\n> ```\r\n\r\nThanks, but this doesn't help me to save time during initial loading, right?",
"Indeed by default load_dataset would download and prepare everything as Arrow files. And passing `split=train[:10]` memory maps only the beginning of the full dataset that has been prepared on disk.\r\n\r\nIf you don't want to download everything, you can use streaming : \r\n```python \r\nids = load_dataset(..., streaming=True)\r\nfirst_samples = list(ids[\"train\"].take(10))\r\n```\r\n\r\nTo get a Dataset you can use \r\n```python \r\nds = Dataset.from_generator(ids.take(10).__iter__)\r\n```\r\n\r\nedit: fixed small bug",
"Thanks @lhoestq, but I don't think it is 100% accurate, as it doesn't keep the dataset structure exactly the same.\r\nTo load the full dataset, I do:\r\n```\r\ndata = load_dataset(\"json\", data_files=\"a.json\")\r\ntrain_data = data[\"train\"].shuffle()\r\n```\r\n\r\nBut when I am changing it as per your instructions: \r\n```\r\nids = load_dataset(\"json\", data_files=\"a.json\", streaming=True)\r\ndata = Dataset.from_generator(ids[\"train\"].take(1).__iter__)\r\ntrain_data = data[\"train\"].shuffle()\r\n```\r\nIt throws KeyError.\r\nI need a simple way, like you suggested, to have a subset of a Dataset, which exactly the same attributes.\r\n",
"Whoops I fixed my code sorry\r\n```diff\r\n- ds = Dataset.from_generator(ids[\"train\"].take(10).__iter__)\r\n+ ds = Dataset.from_generator(ids.take(10).__iter__)\r\n```\r\n\r\nin your case that means running\r\n```python\r\ntrain_data = data.shuffle()\r\n```\r\n\r\nwithout `[\"train\"]`"
] | 2021-06-23T07:19:52
| 2023-04-19T11:05:38
| null |
NONE
| null | null | null |
I am using PyTorch Lightning along with datasets (thanks for so many datasets already prepared and the great splits).
Every time I execute load_dataset for the imdb dataset it takes some time even if I specify a split involving very few samples. I guess this due to hashing as per the other issues.
Is there a way to only load part of the dataset on load_dataset? This would really speed up my workflow.
Something like a debug mode would really help. Thanks!
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2538/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2538/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2526
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2526/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2526/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2526/events
|
https://github.com/huggingface/datasets/issues/2526
| 925,929,228
|
MDU6SXNzdWU5MjU5MjkyMjg=
| 2,526
|
Add COCO datasets
|
{
"login": "NielsRogge",
"id": 48327001,
"node_id": "MDQ6VXNlcjQ4MzI3MDAx",
"avatar_url": "https://avatars.githubusercontent.com/u/48327001?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/NielsRogge",
"html_url": "https://github.com/NielsRogge",
"followers_url": "https://api.github.com/users/NielsRogge/followers",
"following_url": "https://api.github.com/users/NielsRogge/following{/other_user}",
"gists_url": "https://api.github.com/users/NielsRogge/gists{/gist_id}",
"starred_url": "https://api.github.com/users/NielsRogge/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/NielsRogge/subscriptions",
"organizations_url": "https://api.github.com/users/NielsRogge/orgs",
"repos_url": "https://api.github.com/users/NielsRogge/repos",
"events_url": "https://api.github.com/users/NielsRogge/events{/privacy}",
"received_events_url": "https://api.github.com/users/NielsRogge/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
},
{
"id": 3608941089,
"node_id": "LA_kwDODunzps7XHBIh",
"url": "https://api.github.com/repos/huggingface/datasets/labels/vision",
"name": "vision",
"color": "bfdadc",
"default": false,
"description": "Vision datasets"
}
] |
open
| false
|
{
"login": "merveenoyan",
"id": 53175384,
"node_id": "MDQ6VXNlcjUzMTc1Mzg0",
"avatar_url": "https://avatars.githubusercontent.com/u/53175384?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/merveenoyan",
"html_url": "https://github.com/merveenoyan",
"followers_url": "https://api.github.com/users/merveenoyan/followers",
"following_url": "https://api.github.com/users/merveenoyan/following{/other_user}",
"gists_url": "https://api.github.com/users/merveenoyan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/merveenoyan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/merveenoyan/subscriptions",
"organizations_url": "https://api.github.com/users/merveenoyan/orgs",
"repos_url": "https://api.github.com/users/merveenoyan/repos",
"events_url": "https://api.github.com/users/merveenoyan/events{/privacy}",
"received_events_url": "https://api.github.com/users/merveenoyan/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "merveenoyan",
"id": 53175384,
"node_id": "MDQ6VXNlcjUzMTc1Mzg0",
"avatar_url": "https://avatars.githubusercontent.com/u/53175384?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/merveenoyan",
"html_url": "https://github.com/merveenoyan",
"followers_url": "https://api.github.com/users/merveenoyan/followers",
"following_url": "https://api.github.com/users/merveenoyan/following{/other_user}",
"gists_url": "https://api.github.com/users/merveenoyan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/merveenoyan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/merveenoyan/subscriptions",
"organizations_url": "https://api.github.com/users/merveenoyan/orgs",
"repos_url": "https://api.github.com/users/merveenoyan/repos",
"events_url": "https://api.github.com/users/merveenoyan/events{/privacy}",
"received_events_url": "https://api.github.com/users/merveenoyan/received_events",
"type": "User",
"site_admin": false
}
] | null |
[
"I'm currently adding it, the entire dataset is quite big around 30 GB so I add splits separately. You can take a look here https://huggingface.co/datasets/merve/coco",
"I talked to @lhoestq and it's best if I download this dataset through TensorFlow datasets instead, so I'll be implementing that one really soon.\r\n@NielsRogge ",
"I started adding COCO, will be done tomorrow EOD\r\nmy work so far https://github.com/merveenoyan/datasets (my fork)",
"Hi Merve @merveenoyan , thank you so much for your great contribution! May I ask about the current progress of your implementation? Cuz I see the pull request is still in progess here. Or can I just run the COCO scripts in your fork repo?",
"Hello @yixuanren I had another prioritized project about to be merged, but I'll start continuing today will finish up soon. ",
"> Hello @yixuanren I had another prioritized project about to be merged, but I'll start continuing today will finish up soon.\r\n\r\nIt's really nice of you!! I see you've commited another version just now",
"@yixuanren we're working on it, will be available soon, thanks a lot for your patience",
"Hi @NielsRogge and @merveenoyan, did you find a way to load a dataset with COCO annotations to HF's hub?\r\nI have a panoptic segmentation dataset in COCO format and would like to share it with the community.\r\nThanks in advance :)",
"The COCO format is not supported out of the box in the HF's hub - you'd need to reformat it to an [ImageFolder](https://huggingface.co/docs/datasets/image_dataset#imagefolder) with metadata format, or write a [loading script](https://huggingface.co/docs/datasets/image_dataset#loading-script)",
"> The COCO format is not supported out of the box in the HF's hub - you'd need to reformat it to an [ImageFolder](https://huggingface.co/docs/datasets/image_dataset#imagefolder) with metadata format, or write a [loading script](https://huggingface.co/docs/datasets/image_dataset#loading-script)\r\n\r\nHi @lhoestq , thank you for your quick reply.\r\nI've correctly created a metadata.jsonl file for a dataset with instance segmentation annotations [here](https://huggingface.co/datasets/lombardata/data_2017)\r\nbut do not understand how I can integrate panoptic annotations with the metadata format of ImageFolder datasets. The \"problem\" with panoptic annotations is that we have a folder with images, a json file with annotations and another folder with png annotations.\r\n\r\nI checked between all the datasets already published on HuggingFace and, the only one who has uploaded a correct panoptic dataset is @NielsRogge [here](https://huggingface.co/datasets/nielsr/coco-panoptic-val2017) and [here](https://huggingface.co/datasets/nielsr/ade20k-panoptic-demo). Indeed he accomplished to have three fields : \r\n1.image (image)\r\n2.label (image)\r\n3.segments_info (list)\r\nbut I not find the corresponding code that allows to upload a panoptic dataset from this 3 sources.\r\nCan you please share an example code?\r\nThanks !",
"Both were uploaded using `ds.push_to_hub()` :)\r\n\r\nYou can get a Dataset from a python dictionary using `ds = Dataset.from_dict(...)` and casts the paths to images to the `Image()` type using `ds = ds.cast_column(\"image\", Image())`.\r\n\r\n```python\r\nfrom datasets import Dataset, Image\r\n\r\nds = Dataset.from_dict(...)\r\nds = ds.cast_column(\"image\", Image())\r\nds = ds.cast_column(\"label\", Image())\r\nds.push_to_hub(...)\r\n```",
"> Both were uploaded using `ds.push_to_hub()` :)\r\n> \r\n> You can get a Dataset from a python dictionary using `ds = Dataset.from_dict(...)` and casts the paths to images to the `Image()` type using `ds = ds.cast_column(\"image\", Image())`.\r\n> \r\n> ```python\r\n> from datasets import Dataset, Image\r\n> \r\n> ds = Dataset.from_dict(...)\r\n> ds = ds.cast_column(\"image\", Image())\r\n> ds = ds.cast_column(\"label\", Image())\r\n> ds.push_to_hub(...)\r\n> ```\r\n\r\nThank you very much @lhoestq , I succesfully created a hf dataset [here](https://huggingface.co/datasets/lombardata/panoptic_2023_06_21) with the two fields :\r\n1.image (image)\r\n2.label (image)\r\nfollowing your suggestions. Now still remain the problem of uploading **segments_info** information to the dataset.\r\nThere is a function that easily imports the _panoptic_coco_annotation.json_ file to a segment_info field?\r\nI think we must define a **list_of_segment**, i.e. a list of lists of this type : \r\n```python\r\n[ { \"area\": 214858, \"bbox\": [ 0, 0, 511, 760 ], \"category_id\": 0, \"id\": 7895160, \"iscrowd\": 0 }, { \"area\": 73067, \"bbox\": [ 98, 719, 413, 253 ], \"category_id\": 3, \"id\": 3289680, \"iscrowd\": 0 }, { \"area\": 832, \"bbox\": [ 53, 0, 101, 16 ], \"category_id\": 5, \"id\": 5273720, \"iscrowd\": 0 }, { \"area\": 70668, \"bbox\": [ 318, 60, 191, 392 ], \"category_id\": 8, \"id\": 15132390, \"iscrowd\": 0 }, { \"area\": 32696, \"bbox\": [ 0, 100, 78, 872 ], \"category_id\": 18, \"id\": 472063, \"iscrowd\": 0 }, { \"area\": 76045, \"bbox\": [ 42, 48, 264, 924 ], \"category_id\": 37, \"id\": 16713830, \"iscrowd\": 0 }, { \"area\": 27103, \"bbox\": [ 288, 482, 216, 306 ], \"category_id\": 47, \"id\": 16753408, \"iscrowd\": 0 } ]\r\n```\r\nand then apply again the **cast_column** function [here](https://github.com/huggingface/datasets/blob/2.13.0/src/datasets/arrow_dataset.py#L2060) but with a list as a second argument, like : \r\n```python\r\nfrom datasets import Dataset, Image\r\nds = ds.cast_column(\"image\", Image())\r\nds = ds.cast_column(\"label\", Image())\r\nds = ds.cast_column(\"segments_info\", list)\r\n```\r\nbut I do not see how to transfer the information of the _panoptic_coco_annotation.json_ to a list of lists of this type : \r\n```python\r\n[ { \"area\": 214858, \"bbox\": [ 0, 0, 511, 760 ], \"category_id\": 0, \"id\": 7895160, \"iscrowd\": 0 }, { \"area\": 73067, \"bbox\": [ 98, 719, 413, 253 ], \"category_id\": 3, \"id\": 3289680, \"iscrowd\": 0 }, { \"area\": 832, \"bbox\": [ 53, 0, 101, 16 ], \"category_id\": 5, \"id\": 5273720, \"iscrowd\": 0 }, { \"area\": 70668, \"bbox\": [ 318, 60, 191, 392 ], \"category_id\": 8, \"id\": 15132390, \"iscrowd\": 0 }, { \"area\": 32696, \"bbox\": [ 0, 100, 78, 872 ], \"category_id\": 18, \"id\": 472063, \"iscrowd\": 0 }, { \"area\": 76045, \"bbox\": [ 42, 48, 264, 924 ], \"category_id\": 37, \"id\": 16713830, \"iscrowd\": 0 }, { \"area\": 27103, \"bbox\": [ 288, 482, 216, 306 ], \"category_id\": 47, \"id\": 16753408, \"iscrowd\": 0 } ]\r\n```\r\nlike @NielsRogge has done [here](https://huggingface.co/datasets/nielsr/coco-panoptic-val2017) and [here](https://huggingface.co/datasets/nielsr/ade20k-panoptic-demo).\r\nThank you again for your help and have a good day !",
"You can pass this data in .from_dict() - no need to cast anything for this column\r\n\r\n```python\r\nds = Dataset.from_dict({\r\n \"image\": [...],\r\n \"label\": [...],\r\n \"segments_info\": [...],\r\n)}\r\n```\r\n\r\nwhere `segments_info` is the list of the segment_infos of all the examples in the dataset, and therefore is a list of lists of dicts.",
"> You can pass this data in .from_dict() - no need to cast anything for this column\r\n> \r\n> ```python\r\n> ds = Dataset.from_dict({\r\n> \"image\": [...],\r\n> \"label\": [...],\r\n> \"segments_info\": [...],\r\n> )}\r\n> ```\r\n> \r\n> where `segments_info` is the list of the segment_infos of all the examples in the dataset, and therefore is a list of lists of dicts.\r\n\r\nThank you for the quick reply @lhoestq , but then how to generate the `segments_info` list of lists of dicts starting from a _panoptic_coco_annotation.json_ file ?\r\n\r\n\r\n",
"You read the JSON file and transform the data yourself. I don't think there's an automatic converter somewhere",
"> You read the JSON file and transform the data yourself. I don't think there's an automatic converter somewhere\r\n\r\nPerfect, I've done it and succesfully uploaded a new dataset [here](https://huggingface.co/datasets/lombardata/panoptic_2023_06_22), but I've (I hope) a last problem.\r\nThe dataset has currently 302 images and, when I upload it to the hub, only the first page of images is correctly uploaded.\r\nWhen I try to see the second/third/fourth page of items of my dataset, I can see that the fields **segments_info** and **image_name** are correctly uploaded, while the images are not (the \"null\" string is shown everywhere).\r\n\r\nI've checked the path of images that are not uploaded and they exists, is there a problem with the size of the dataset ?\r\nHow can I upload the whole dataset to the hub ?\r\nThank you again @lhoestq and have a good day !",
"Awesome ! Your dataset looks all good 🤗 \r\n\r\nThe `null` in the viewer is a bug on our side, let me investigate"
] | 2021-06-21T07:48:32
| 2023-06-22T14:12:18
| null |
CONTRIBUTOR
| null | null | null |
## Adding a Dataset
- **Name:** COCO
- **Description:** COCO is a large-scale object detection, segmentation, and captioning dataset.
- **Paper + website:** https://cocodataset.org/#home
- **Data:** https://cocodataset.org/#download
- **Motivation:** It would be great to have COCO available in HuggingFace datasets, as we are moving beyond just text. COCO includes multi-modalities (images + text), as well as a huge amount of images annotated with objects, segmentation masks, keypoints etc., on which models like DETR (which I recently added to HuggingFace Transformers) are trained. Currently, one needs to download everything from the website and place it in a local folder, but it would be much easier if we can directly access it through the datasets API.
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2526/reactions",
"total_count": 3,
"+1": 3,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2526/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2516
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2516/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2516/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2516/events
|
https://github.com/huggingface/datasets/issues/2516
| 924,597,470
|
MDU6SXNzdWU5MjQ1OTc0NzA=
| 2,516
|
datasets.map pickle issue resulting in invalid mapping function
|
{
"login": "david-waterworth",
"id": 5028974,
"node_id": "MDQ6VXNlcjUwMjg5NzQ=",
"avatar_url": "https://avatars.githubusercontent.com/u/5028974?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/david-waterworth",
"html_url": "https://github.com/david-waterworth",
"followers_url": "https://api.github.com/users/david-waterworth/followers",
"following_url": "https://api.github.com/users/david-waterworth/following{/other_user}",
"gists_url": "https://api.github.com/users/david-waterworth/gists{/gist_id}",
"starred_url": "https://api.github.com/users/david-waterworth/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/david-waterworth/subscriptions",
"organizations_url": "https://api.github.com/users/david-waterworth/orgs",
"repos_url": "https://api.github.com/users/david-waterworth/repos",
"events_url": "https://api.github.com/users/david-waterworth/events{/privacy}",
"received_events_url": "https://api.github.com/users/david-waterworth/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! `map` calls `__getstate__` using `dill` to hash your map function. This is used by the caching mechanism to recover previously computed results. That's why you don't see any `__setstate__` call.\r\n\r\nWhy do you change an attribute of your tokenizer when `__getstate__` is called ?",
"@lhoestq because if I try to pickle my custom tokenizer (it contains a pure python pretokenization step in an otherwise rust backed tokenizer) I get\r\n\r\n> Exception: Error while attempting to pickle Tokenizer: Custom PreTokenizer cannot be serialized\r\n\r\nSo I remove the Custom PreTokenizer in `__getstate__` and then restore it in `__setstate__` (since it doesn't contain any state). This is what my `__getstate__` / `__setstate__` looks like:\r\n\r\n def __getstate__(self):\r\n \"\"\"\r\n Removes pre_tokenizer since it cannot be pickled\r\n \"\"\"\r\n logger.debug(\"Copy state dict\")\r\n out = self.__dict__.copy()\r\n logger.debug(\"Detaching pre_tokenizer\")\r\n out['_tokenizer'].pre_tokenizer = tokenizers.pre_tokenizers.Sequence([]) \r\n return out\r\n\r\n def __setstate__(self, d):\r\n \"\"\"\r\n Reinstates pre_tokenizer\r\n \"\"\"\r\n logger.debug(\"Reattaching pre_tokenizer\")\r\n self.__dict__ = d\r\n self.backend_tokenizer.pre_tokenizer = self._pre_tokenizer()\r\n\r\nIf this is the case can you think of another way of avoiding my issue?",
"Actually, maybe I need to deep copy `self.__dict__`? That way `self` isn't modified. That was my intention and I thought it was working - I'll double-check after the weekend.",
"Doing a deep copy results in the warning:\r\n\r\n> 06/20/2021 16:02:15 - WARNING - datasets.fingerprint - Parameter 'function'=<function tokenize_function at 0x7f1e95f05d40> of the transform datasets.arrow_dataset.Dataset._map_single couldn't be hashed properly, a random hash was used instead. Make sure your transforms and parameters are serializable with pickle or dill for the dataset fingerprinting and caching to work. If you reuse this transform, the caching mechanism will consider it to be different from the previous calls and recompute everything. This warning is only showed once. Subsequent hashing failures won't be showed.\r\n\r\n\r\n```\r\ndef __getstate__(self):\r\n \"\"\"\r\n Removes pre_tokenizer since it cannot be pickled\r\n \"\"\"\r\n logger.debug(\"Copy state dict\")\r\n out = copy.deepcopy(self.__dict__)\r\n logger.debug(\"Detaching pre_tokenizer\")\r\n out['_tokenizer'].pre_tokenizer = tokenizers.pre_tokenizers.Sequence([]) \r\n return out\r\n```",
"Looks like there is still an object that is not pickable in your `tokenize_function` function.\r\n\r\nYou can test if an object can be pickled and hashed by using \r\n```python\r\nfrom datasets.fingerprint import Hasher\r\n\r\nHasher.hash(my_object)\r\n```\r\n\r\nUnder the hood it pickles the object to compute its hash, so it calls `__getstate__` when applicable.",
"I figured it out, the problem is deep copy itself uses pickle (unless you implement `__deepcopy__`). So when I changed `__getstate__` it started throwing an error.\r\n\r\nI'm sure there's a better way of doing this, but in order to return the `__dict__` without the non-pikelable pre-tokeniser and without modifying self I removed the pre-tokenizers, did a deep copy and then re-generated it.\r\n\r\nIt does work - although I noticed Hasher doesn't call `__hash__` if the object being hashed implements it which I feel it should? If it did I could return a hash of the tokenizers.json file instead.\r\n\r\n```\r\n def __getstate__(self):\r\n \"\"\"\r\n Removes pre_tokenizer since it cannot be pickled\r\n \"\"\"\r\n logger.debug(\"Copy state dict\")\r\n self.backend_tokenizer.pre_tokenizer = tokenizers.pre_tokenizers.Sequence([])\r\n out = copy.deepcopy(self.__dict__) #self.__dict__.copy()\r\n self.backend_tokenizer.pre_tokenizer = self._pre_tokenizer()\r\n\r\n return out\r\n```\r\n",
"I'm glad you figured something out :)\r\n\r\nRegarding hashing: we're not using hashing for the same purpose as the python `__hash__` purpose (which is in general for dictionary lookups). For example it is allowed for python hashing to not return the same hash across sessions, while our hashing must return the same hashes across sessions for the caching to work properly."
] | 2021-06-18T06:47:26
| 2021-06-23T13:47:49
| null |
NONE
| null | null | null |
I trained my own tokenizer, and I needed to use a python custom class. Because of this I have to detach the custom step before saving and reattach after restore. I did this using the standard pickle `__get_state__` / `__set_state__` mechanism. I think it's correct but it fails when I use it inside a function which is mapped to a dataset, i.e. in the manner of run_mlm.py and other huggingface scripts.
The following reproduces the issue - most likely I'm missing something
A simulated tokeniser which can be pickled
```
class CustomTokenizer:
def __init__(self):
self.state = "init"
def __getstate__(self):
print("__getstate__ called")
out = self.__dict__.copy()
self.state = "pickled"
return out
def __setstate__(self, d):
print("__setstate__ called")
self.__dict__ = d
self.state = "restored"
tokenizer = CustomTokenizer()
```
Test that it actually works - prints "__getstate__ called" and "__setstate__ called"
```
import pickle
serialized = pickle.dumps(tokenizer)
restored = pickle.loads(serialized)
assert restored.state == "restored"
```
Simulate a function that tokenises examples, when dataset.map is called, this function
```
def tokenize_function(examples):
assert tokenizer.state == "restored" # this shouldn't fail but it does
output = tokenizer(examples) # this will fail as tokenizer isn't really a tokenizer
return output
```
Use map to simulate tokenization
```
import glob
from datasets import load_dataset
assert tokenizer.state == "restored"
train_files = glob.glob('train*.csv')
validation_files = glob.glob('validation*.csv')
datasets = load_dataset("csv", data_files=dict(train=train_files, validation=validation_files))
tokenized_datasets = datasets.map(
tokenize_function,
batched=True,
)
```
What's happening is I can see that __getstate__ is called but not __setstate__, so the state of `tokenize_function` is invalid at the point that it's actually executed. This doesn't matter as far as I can see for the standard tokenizers as they don't use __getstate__ / __setstate__. I'm not sure if there's another hook I'm supposed to implement as well?
---------------------------------------------------------------------------
AssertionError Traceback (most recent call last)
<ipython-input-22-a2aef4f74aaa> in <module>
8 tokenized_datasets = datasets.map(
9 tokenize_function,
---> 10 batched=True,
11 )
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/dataset_dict.py in map(self, function, with_indices, input_columns, batched, batch_size, remove_columns, keep_in_memory, load_from_cache_file, cache_file_names, writer_batch_size, features, disable_nullable, fn_kwargs, num_proc, desc)
487 desc=desc,
488 )
--> 489 for k, dataset in self.items()
490 }
491 )
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/dataset_dict.py in <dictcomp>(.0)
487 desc=desc,
488 )
--> 489 for k, dataset in self.items()
490 }
491 )
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in map(self, function, with_indices, input_columns, batched, batch_size, drop_last_batch, remove_columns, keep_in_memory, load_from_cache_file, cache_file_name, writer_batch_size, features, disable_nullable, fn_kwargs, num_proc, suffix_template, new_fingerprint, desc)
1633 fn_kwargs=fn_kwargs,
1634 new_fingerprint=new_fingerprint,
-> 1635 desc=desc,
1636 )
1637 else:
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in wrapper(*args, **kwargs)
184 }
185 # apply actual function
--> 186 out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs)
187 datasets: List["Dataset"] = list(out.values()) if isinstance(out, dict) else [out]
188 # re-apply format to the output
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/fingerprint.py in wrapper(*args, **kwargs)
395 # Call actual function
396
--> 397 out = func(self, *args, **kwargs)
398
399 # Update fingerprint of in-place transforms + update in-place history of transforms
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in _map_single(self, function, with_indices, input_columns, batched, batch_size, drop_last_batch, remove_columns, keep_in_memory, load_from_cache_file, cache_file_name, writer_batch_size, features, disable_nullable, fn_kwargs, new_fingerprint, rank, offset, desc)
1961 indices,
1962 check_same_num_examples=len(input_dataset.list_indexes()) > 0,
-> 1963 offset=offset,
1964 )
1965 except NumExamplesMismatch:
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in apply_function_on_filtered_inputs(inputs, indices, check_same_num_examples, offset)
1853 effective_indices = [i + offset for i in indices] if isinstance(indices, list) else indices + offset
1854 processed_inputs = (
-> 1855 function(*fn_args, effective_indices, **fn_kwargs) if with_indices else function(*fn_args, **fn_kwargs)
1856 )
1857 if update_data is None:
<ipython-input-21-8ee4a8ba5b1b> in tokenize_function(examples)
1 def tokenize_function(examples):
----> 2 assert tokenizer.state == "restored"
3 tokenizer(examples)
4 return examples
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2516/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2516/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2514
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2514/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2514/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2514/events
|
https://github.com/huggingface/datasets/issues/2514
| 924,417,172
|
MDU6SXNzdWU5MjQ0MTcxNzI=
| 2,514
|
Can datasets remove duplicated rows?
|
{
"login": "liuxinglan",
"id": 16516583,
"node_id": "MDQ6VXNlcjE2NTE2NTgz",
"avatar_url": "https://avatars.githubusercontent.com/u/16516583?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/liuxinglan",
"html_url": "https://github.com/liuxinglan",
"followers_url": "https://api.github.com/users/liuxinglan/followers",
"following_url": "https://api.github.com/users/liuxinglan/following{/other_user}",
"gists_url": "https://api.github.com/users/liuxinglan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/liuxinglan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/liuxinglan/subscriptions",
"organizations_url": "https://api.github.com/users/liuxinglan/orgs",
"repos_url": "https://api.github.com/users/liuxinglan/repos",
"events_url": "https://api.github.com/users/liuxinglan/events{/privacy}",
"received_events_url": "https://api.github.com/users/liuxinglan/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! For now this is probably the best option.\r\nWe might add a feature like this in the feature as well.\r\n\r\nDo you know any deduplication method that works on arbitrary big datasets without filling up RAM ?\r\nOtherwise we can have do the deduplication in memory like pandas but I feel like this is going to be limiting for some cases",
"Yes, I'd like to work on this feature once I'm done with #2500, but first I have to do some research, and see if the implementation wouldn't be too complex.\r\n\r\nIn the meantime, maybe [this lib](https://github.com/TomScheffers/pyarrow_ops) can help. However, note that this lib operates directly on pyarrow tables and relies only on `hash` to find duplicates (e.g. `-1` and `-2` have the same hash in Python 3, so this lib will treat them as duplicates), which doesn't make much sense.",
"> Hi ! For now this is probably the best option.\r\n> We might add a feature like this in the feature as well.\r\n> \r\n> Do you know any deduplication method that works on arbitrary big datasets without filling up RAM ?\r\n> Otherwise we can have do the deduplication in memory like pandas but I feel like this is going to be limiting for some cases\r\n\r\nGreat if this is can be done. Thanks!!\r\n\r\nNot sure if you are asking me. In any case I don't know of any unfortunately :( in practice if data is really large we normally do it with spark (only for info. I understand this is not useful in developing this library..)",
"Hello,\r\n\r\nI'm also interested in this feature.\r\nHas there been progress on this issue?\r\n\r\nCould we use a similar trick as above, but with a better hashing algorithm like SHA?\r\n\r\nWe could also use a [bloom filter](https://en.wikipedia.org/wiki/Bloom_filter), should we care a lot about collision in this case?",
"For reference, we can get a solution fairly easily if we assume that we can hold in memory all unique values. \r\n\r\n```python\r\nfrom datasets import Dataset\r\nfrom itertools import cycle\r\nfrom functools import partial\r\n\r\nmemory = set()\r\ndef is_unique(elem:Any , column: str, memory: set) -> bool:\r\n if elem[column] in memory:\r\n return False\r\n else:\r\n memory.add(elem[column])\r\n return True\r\n\r\n# Example dataset\r\nds = Dataset.from_dict({\"col1\" : [sent for i, sent in zip(range(10), cycle([\"apple\", \"orange\", \"pear\"]))],\r\n \"col2\": [i % 5 for i in range(10)]})\r\n\r\n# Drop duplicates in `ds` on \"col1\"\r\nds2 = ds.filter(partial(is_unique, column=\"col1\", memory=memory))\r\n```\r\n\r\nOf course, we can improve the API so that we can introduce `Dataset.drop_duplicates`.\r\nFor the parallel version, we can use a shared memory set.",
"An approach that works assuming you can hold the all the unique document hashes in memory:\r\n```python\r\nfrom datasets import load_dataset\r\n\r\ndef get_hash(example):\r\n \"\"\"Get hash of content field.\"\"\"\r\n return {\"hash\": hash(example[\"content\"])} # can use any hashing function here\r\n \r\ndef check_uniques(example, uniques):\r\n \"\"\"Check if current hash is still in set of unique hashes and remove if true.\"\"\"\r\n if example[\"hash\"] in uniques:\r\n uniques.remove(example[\"hash\"])\r\n return True\r\n else:\r\n return False\r\n\r\nds = load_dataset(\"some_dataset\")\r\nds = ds.map(get_hash)\r\nuniques = set(ds.unique(\"hash\"))\r\nds_filter = ds.filter(check_uniques, fn_kwargs={\"uniques\": uniques})\r\n```\r\nIf the `uniques` could be stored in arrow then no additional memory would used at all but I don't know if this is possible.\r\n",
"@lvwerra hey, could you tell me how reliable is this deduplication method. i am currently using the same deduplication strategy to deduplicate a large text corpus to pretrain LLMs ~ 11B to 20B. just needed to ensure if this strategy would be fine on large datasets for LLMs pretraining. ",
"Hi @StephennFernandes I'm also trying to pretrain an llm, and need to do deduplication for my dataset, \r\nwhich method you applied please?",
"Hey @Manel-Hik \n\nThe following is a simpler yet really effective deduplication code that i has used in the past. \n\ngiven that I have limited training corpus for the languages I wanted to train i relied on this code. https://huggingface.co/datasets/Finnish-NLP/mc4_fi_cleaned/blob/main/deduplicate.py\n\n\nfor more robust and stronger deduplication, refer to this huggingface repo, that's newly released: https://github.com/huggingface/datatrove\n\n\n\n",
"Thanks a lot Sure I will check it @StephennFernandes ",
"Hi, is there any updates? Thanks!"
] | 2021-06-17T23:35:38
| 2024-03-28T05:37:41
| null |
NONE
| null | null | null |
**Is your feature request related to a problem? Please describe.**
i find myself more and more relying on datasets just to do all the preprocessing. One thing however, for removing duplicated rows, I couldn't find out how and am always converting datasets to pandas to do that..
**Describe the solution you'd like**
have a functionality of " remove duplicated rows"
**Describe alternatives you've considered**
convert dataset to pandas, remove duplicate, and convert back...
**Additional context**
no
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2514/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2514/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2503
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2503/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2503/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2503/events
|
https://github.com/huggingface/datasets/issues/2503
| 920,636,186
|
MDU6SXNzdWU5MjA2MzYxODY=
| 2,503
|
SubjQA wrong boolean values in entries
|
{
"login": "arnaudstiegler",
"id": 26485052,
"node_id": "MDQ6VXNlcjI2NDg1MDUy",
"avatar_url": "https://avatars.githubusercontent.com/u/26485052?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/arnaudstiegler",
"html_url": "https://github.com/arnaudstiegler",
"followers_url": "https://api.github.com/users/arnaudstiegler/followers",
"following_url": "https://api.github.com/users/arnaudstiegler/following{/other_user}",
"gists_url": "https://api.github.com/users/arnaudstiegler/gists{/gist_id}",
"starred_url": "https://api.github.com/users/arnaudstiegler/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/arnaudstiegler/subscriptions",
"organizations_url": "https://api.github.com/users/arnaudstiegler/orgs",
"repos_url": "https://api.github.com/users/arnaudstiegler/repos",
"events_url": "https://api.github.com/users/arnaudstiegler/events{/privacy}",
"received_events_url": "https://api.github.com/users/arnaudstiegler/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
] | null |
[
"Hi @arnaudstiegler, thanks for reporting. I'm investigating it.",
"@arnaudstiegler I have just checked that these mismatches are already present in the original dataset: https://github.com/megagonlabs/SubjQA\r\n\r\nWe are going to contact the dataset owners to report this.",
"I have:\r\n- opened an issue in their repo: https://github.com/megagonlabs/SubjQA/issues/3\r\n- written an email to all the paper authors",
"Please [see my response](https://github.com/megagonlabs/SubjQA/issues/3#issuecomment-905160010). There will be a fix in a couple of days."
] | 2021-06-14T17:42:46
| 2021-08-25T03:52:06
| null |
NONE
| null | null | null |
## Describe the bug
SubjQA seems to have a boolean that's consistently wrong.
It defines:
- question_subj_level: The subjectiviy level of the question (on a 1 to 5 scale with 1 being the most subjective).
- is_ques_subjective: A boolean subjectivity label derived from question_subj_level (i.e., scores below 4 are considered as subjective)
However, `is_ques_subjective` seems to have wrong values in the entire dataset.
For instance, in the example in the dataset card, we have:
- "question_subj_level": 2
- "is_ques_subjective": false
However, according to the description, the question should be subjective since the `question_subj_level` is below 4
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2503/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2503/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2499
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2499/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2499/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2499/events
|
https://github.com/huggingface/datasets/issues/2499
| 920,413,021
|
MDU6SXNzdWU5MjA0MTMwMjE=
| 2,499
|
Python Programming Puzzles
|
{
"login": "VictorSanh",
"id": 16107619,
"node_id": "MDQ6VXNlcjE2MTA3NjE5",
"avatar_url": "https://avatars.githubusercontent.com/u/16107619?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/VictorSanh",
"html_url": "https://github.com/VictorSanh",
"followers_url": "https://api.github.com/users/VictorSanh/followers",
"following_url": "https://api.github.com/users/VictorSanh/following{/other_user}",
"gists_url": "https://api.github.com/users/VictorSanh/gists{/gist_id}",
"starred_url": "https://api.github.com/users/VictorSanh/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/VictorSanh/subscriptions",
"organizations_url": "https://api.github.com/users/VictorSanh/orgs",
"repos_url": "https://api.github.com/users/VictorSanh/repos",
"events_url": "https://api.github.com/users/VictorSanh/events{/privacy}",
"received_events_url": "https://api.github.com/users/VictorSanh/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[
"👀 @TalSchuster",
"Thanks @VictorSanh!\r\nThere's also a [notebook](https://aka.ms/python_puzzles) and [demo](https://aka.ms/python_puzzles_study) available now to try out some of the puzzles"
] | 2021-06-14T13:27:18
| 2021-06-15T18:14:14
| null |
MEMBER
| null | null | null |
## Adding a Dataset
- **Name:** Python Programming Puzzles
- **Description:** Programming challenge called programming puzzles, as an objective and comprehensive evaluation of program synthesis
- **Paper:** https://arxiv.org/pdf/2106.05784.pdf
- **Data:** https://github.com/microsoft/PythonProgrammingPuzzles ([Scrolling through the data](https://github.com/microsoft/PythonProgrammingPuzzles/blob/main/problems/README.md))
- **Motivation:** Spans a large range of difficulty, problems, and domains. A useful resource for evaluation as we don't have a clear understanding of the abilities and skills of extremely large LMs.
Note: it's a growing dataset (contributions are welcome), so we'll need careful versioning for this dataset.
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2499/reactions",
"total_count": 2,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 1,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2499/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2498
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2498/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2498/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2498/events
|
https://github.com/huggingface/datasets/issues/2498
| 920,411,285
|
MDU6SXNzdWU5MjA0MTEyODU=
| 2,498
|
Improve torch formatting performance
|
{
"login": "vblagoje",
"id": 458335,
"node_id": "MDQ6VXNlcjQ1ODMzNQ==",
"avatar_url": "https://avatars.githubusercontent.com/u/458335?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/vblagoje",
"html_url": "https://github.com/vblagoje",
"followers_url": "https://api.github.com/users/vblagoje/followers",
"following_url": "https://api.github.com/users/vblagoje/following{/other_user}",
"gists_url": "https://api.github.com/users/vblagoje/gists{/gist_id}",
"starred_url": "https://api.github.com/users/vblagoje/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/vblagoje/subscriptions",
"organizations_url": "https://api.github.com/users/vblagoje/orgs",
"repos_url": "https://api.github.com/users/vblagoje/repos",
"events_url": "https://api.github.com/users/vblagoje/events{/privacy}",
"received_events_url": "https://api.github.com/users/vblagoje/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"That’s interesting thanks, let’s see what we can do. Can you detail your last sentence? I’m not sure I understand it well.",
"Hi ! I just re-ran a quick benchmark and using `to_numpy()` seems to be faster now:\r\n\r\n```python\r\nimport pyarrow as pa # I used pyarrow 3.0.0\r\nimport numpy as np\r\n\r\nn, max_length = 1_000, 512\r\nlow, high, size = 0, 2 << 16, (n, max_length)\r\n\r\ntable = pa.Table.from_pydict({\r\n \"input_ids\": np.random.default_rng(42).integers(low=low, high=high, size=size).tolist()\r\n})\r\n\r\n\r\n%%timeit\r\n_ = table.to_pandas()[\"input_ids\"].to_numpy()\r\n# 1.44 ms ± 80.1 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\r\n\r\n%%timeit\r\n_ = table[\"input_ids\"].to_pandas().to_numpy()\r\n# 461 µs ± 14.2 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\r\n\r\n%%timeit\r\n_ = table[\"input_ids\"].to_numpy()\r\n# 317 µs ± 5.06 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\r\n```\r\n\r\nCurrently the conversion from arrow to numpy is done in the NumpyArrowExtractor here:\r\n\r\nhttps://github.com/huggingface/datasets/blob/d6d0ede9486ffad7944642ca9a326e058b676788/src/datasets/formatting/formatting.py#L143-L166\r\n\r\nLet's update the NumpyArrowExtractor to call `to_numpy` directly and see how our github benchmarks evolve ?__",
"Sounds like a plan @lhoestq If you create a PR I'll pick it up and try it out right away! ",
"@lhoestq I can also prepare the PR, just lmk. ",
"I’m not exactly sure how to read the graph but it seems that to_categorical take a lot of time here. Could you share more informations on the features/stats of your datasets so we could maybe design a synthetic datasets that looks more similar for debugging testing?",
"I created https://github.com/huggingface/datasets/pull/2505 if you want to play with it @vblagoje ",
"> I’m not exactly sure how to read the graph but it seems that to_categorical take a lot of time here. Could you share more informations on the features/stats of your datasets so we could maybe design a synthetic datasets that looks more similar for debugging testing?\r\n\r\n@thomwolf starting from the top, each rectangle represents the cumulative amount of it takes to execute the method call. Therefore, format_batch in torch_formatter.py takes ~20 sec, and the largest portion of that call is taken by to_pandas call and the smaller portion (grey rectangle) by the other method invocation(s) in format_batch (series_to_numpy etc). \r\n\r\nFeatures of the dataset are BERT pre-training model input columns i.e:\r\n```\r\nf = Features({ \r\n \"input_ids\": Sequence(feature=Value(dtype=\"int32\")), \r\n \"attention_mask\": Sequence(feature=Value(dtype=\"int8\")), \r\n \"token_type_ids\": Sequence(feature=Value(dtype=\"int8\")), \r\n \"labels\": Sequence(feature=Value(dtype=\"int32\")), \r\n \"next_sentence_label\": Value(dtype=\"int8\")\r\n})\r\n```\r\n\r\nI'll work with @lhoestq till we get to the bottom of this one. \r\n ",
"@lhoestq the proposed branch is faster, but overall training speedup is a few percentage points. I couldn't figure out how to include the GitHub branch into setup.py, so I couldn't start NVidia optimized Docker-based pre-training run. But on bare metal, there is a slight improvement. I'll do some more performance traces. ",
"Hi @vblagoje, to install Datasets from @lhoestq PR reference #2505, you can use:\r\n```shell\r\npip install git+ssh://git@github.com/huggingface/datasets.git@refs/pull/2505/head#egg=datasets\r\n```",
"Hey @albertvillanova yes thank you, I am aware, I can easily pull it from a terminal command line but then I can't automate docker image builds as dependencies are picked up from setup.py and for some reason setup.py doesn't accept this string format.",
"@vblagoje in that case, you can add this to your `setup.py`:\r\n```python\r\n install_requires=[\r\n \"datasets @ git+ssh://git@github.com/huggingface/datasets.git@refs/pull/2505/head\",\r\n```",
"@lhoestq @thomwolf @albertvillanova The new approach is definitely faster, dataloader now takes less than 3% cumulative time (pink rectangle two rectangles to the right of tensor.py backward invocation)\r\n\r\n\r\n\r\nWhen we drill down into dataloader next invocation we get:\r\n\r\n\r\n\r\nAnd finally format_batch:\r\n\r\n\r\n\r\n\r\nNot sure this could be further improved but this is definitely a decent step forward.\r\n\r\n",
"> ```python\r\n> datasets @ git+ssh://git@github.com/huggingface/datasets.git@refs/pull/2505/head\r\n> ```\r\n\r\n@albertvillanova how would I replace datasets dependency in https://github.com/huggingface/transformers/blob/master/setup.py as the above approach is not working. ",
"@vblagoje I tested my proposed approach before posting it here and it worked for me. \r\n\r\nIs it not working in your case because of the SSH protocol? In that case you could try the same approach but using HTTPS:\r\n```\r\n\"datasets @ git+https://github.com/huggingface/datasets.git@refs/pull/2505/head\",\r\n``` ",
"Also note the blanks before and after the `@`.",
"@albertvillanova of course it works. Apologies. I needed to change datasets in all deps references , like [here](https://github.com/huggingface/transformers/blob/master/setup.py#L235) for example. ",
"Is time spent casting an issue here? See https://github.com/huggingface/datasets/issues/4676 that Datasets can spend huge amounts of time repeatedly casting to Python objects."
] | 2021-06-14T13:25:24
| 2022-07-15T17:12:04
| null |
CONTRIBUTOR
| null | null | null |
**Is your feature request related to a problem? Please describe.**
It would be great, if possible, to further improve read performance of raw encoded datasets and their subsequent conversion to torch tensors.
A bit more background. I am working on LM pre-training using HF ecosystem. We use encoded HF Wikipedia and BookCorpus datasets. The training machines are similar to DGX-1 workstations. We use HF trainer torch.distributed training approach on a single machine with 8 GPUs.
The current performance is about 30% slower than NVidia optimized BERT [examples](https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/LanguageModeling) baseline. Quite a bit of customized code and training loop tricks were used to achieve the baseline performance. It would be great to achieve the same performance while using nothing more than off the shelf HF ecosystem. Perhaps, in the future, with @stas00 work on deepspeed integration, it could even be exceeded.
**Describe the solution you'd like**
Using profiling tools we've observed that appx. 25% of cumulative run time is spent on data loader next call.
As you can observe most of the data loader next call is spent in HF datasets torch_formatter.py format_batch call.
Digging a bit deeper into format_batch we can see the following profiler data:
Once again, a lot of time is spent in pyarrow table conversion to pandas which seems like an intermediary step. Offline @lhoestq told me that this approach was, for some unknown reason, faster than direct to numpy conversion.
**Describe alternatives you've considered**
I am not familiar with pyarrow and have not yet considered the alternatives to the current approach.
Most of the online advice around data loader performance improvements revolve around increasing number of workers, using pin memory for copying tensors from host device to gpus but we've already tried these avenues without much performance improvement. Weights & Biases dashboard for the pre-training task reports CPU utilization of ~ 10%, GPUs are completely saturated (GPU utilization is above 95% on all GPUs), while disk utilization is above 90%.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2498/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2498/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2494
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2494/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2494/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2494/events
|
https://github.com/huggingface/datasets/issues/2494
| 920,149,183
|
MDU6SXNzdWU5MjAxNDkxODM=
| 2,494
|
Improve docs on Enhancing performance
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892861,
"node_id": "MDU6TGFiZWwxOTM1ODkyODYx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/documentation",
"name": "documentation",
"color": "0075ca",
"default": true,
"description": "Improvements or additions to documentation"
}
] |
open
| false
| null |
[] | null |
[
"Hi @albertvillanova, I hope you are doing well.\r\n\r\nI am interested in this issue, is this still unresolved and open ?\r\n\r\nThe link you have provided in the above message directs to a webpage that does not exist.\r\n\r\nThanks and Regards"
] | 2021-06-14T08:11:48
| 2024-01-20T19:48:38
| null |
MEMBER
| null | null | null |
In the ["Enhancing performance"](https://huggingface.co/docs/datasets/loading_datasets.html#enhancing-performance) section of docs, add specific use cases:
- How to make datasets the fastest
- How to make datasets take the less RAM
- How to make datasets take the less hard drive mem
cc: @thomwolf
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2494/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2494/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2485
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2485/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2485/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2485/events
|
https://github.com/huggingface/datasets/issues/2485
| 919,099,218
|
MDU6SXNzdWU5MTkwOTkyMTg=
| 2,485
|
Implement layered building
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[] | 2021-06-11T18:54:25
| 2021-06-11T18:54:25
| null |
MEMBER
| null | null | null |
As discussed with @stas00 and @lhoestq (see also here https://github.com/huggingface/datasets/issues/2481#issuecomment-859712190):
> My suggestion for this would be to have this enabled by default.
>
> Plus I don't know if there should be a dedicated issue to that is another functionality. But I propose layered building rather than all at once. That is:
>
> 1. uncompress a handful of files via a generator enough to generate one arrow file
> 2. process arrow file 1
> 3. delete all the files that went in and aren't needed anymore.
>
> rinse and repeat.
>
> 1. This way much less disc space will be required - e.g. on JZ we won't be running into inode limitation, also it'd help with the collaborative hub training project
> 2. The user doesn't need to go and manually clean up all the huge files that were left after pre-processing
> 3. It would already include deleting temp files this issue is talking about
>
> I wonder if the new streaming API would be of help, except here the streaming would be into arrow files as the destination, rather than dataloaders.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2485/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2485/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2480
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2480/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2480/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2480/events
|
https://github.com/huggingface/datasets/issues/2480
| 918,678,578
|
MDU6SXNzdWU5MTg2Nzg1Nzg=
| 2,480
|
Set download/extracted paths configurable
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
] | null |
[
"For example to be able to send uncompressed and temp build files to another volume/partition, so that the user gets the minimal disk usage on their primary setup - and ends up with just the downloaded compressed data + arrow files, but outsourcing the huge files and building to another partition. e.g. on JZ there is a special partition for fast data, but it's also volatile, so only temp files should go there.\r\n\r\nThink of it as `TMPDIR` so we need the equivalent for `datasets`."
] | 2021-06-11T12:20:24
| 2021-06-15T14:23:49
| null |
MEMBER
| null | null | null |
As discussed with @stas00 and @lhoestq, setting these paths configurable may allow to overcome disk space limitation on different partitions/drives.
TODO:
- [x] Set configurable extracted datasets path: #2487
- [x] Set configurable downloaded datasets path: #2488
- [ ] Set configurable "incomplete" datasets path?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2480/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 1,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2480/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2478
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2478/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2478/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2478/events
|
https://github.com/huggingface/datasets/issues/2478
| 918,507,510
|
MDU6SXNzdWU5MTg1MDc1MTA=
| 2,478
|
Create release script
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
] | null |
[
"I've aligned the release script with Transformers in #6004, so I think this issue can be closed."
] | 2021-06-11T09:38:02
| 2023-07-20T13:22:23
| null |
MEMBER
| null | null | null |
Create a script so that releases can be done automatically (as done in `transformers`).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2478/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2478/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2462
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2462/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2462/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2462/events
|
https://github.com/huggingface/datasets/issues/2462
| 915,384,613
|
MDU6SXNzdWU5MTUzODQ2MTM=
| 2,462
|
Merge DatasetDict and Dataset
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 2067400324,
"node_id": "MDU6TGFiZWwyMDY3NDAwMzI0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/generic%20discussion",
"name": "generic discussion",
"color": "c5def5",
"default": false,
"description": "Generic discussion on the library"
}
] |
open
| false
| null |
[] |
{
"url": "https://api.github.com/repos/huggingface/datasets/milestones/8",
"html_url": "https://github.com/huggingface/datasets/milestone/8",
"labels_url": "https://api.github.com/repos/huggingface/datasets/milestones/8/labels",
"id": 6968069,
"node_id": "MI_kwDODunzps4AalMF",
"number": 8,
"title": "1.12",
"description": "Next minor release",
"creator": {
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
},
"open_issues": 4,
"closed_issues": 2,
"state": "open",
"created_at": "2021-07-21T15:34:56",
"updated_at": "2021-10-13T10:26:33",
"due_on": "2021-08-30T07:00:00",
"closed_at": null
}
|
[
"Any update on this? @lhoestq ",
"Unless there is high demande I don't think we will end up implementing this. This is a lot of work with very few advantages"
] | 2021-06-08T19:22:04
| 2023-08-16T09:34:34
| null |
MEMBER
| null | null | null |
As discussed in #2424 and #2437 (please see there for detailed conversation):
- It would be desirable to improve UX with respect the confusion between DatasetDict and Dataset.
- The difference between Dataset and DatasetDict is an additional abstraction complexity that confuses "typical" end users.
- A user expects a "Dataset" (whatever it contains multiple or a single split) and maybe it could be interesting to try to simplify the user-facing API as much as possible to hide this complexity from the end user.
Here is a proposal for discussion and refined (and potential abandon if it's not good enough):
- let's consider that a DatasetDict is also a Dataset with the various split concatenated one after the other
- let's disallow the use of integers in split names (probably not a very big breaking change)
- when you index with integers you access the examples progressively in split after the other is finished (in a deterministic order)
- when you index with strings/split name you have the same behavior as now (full backward compat)
- let's then also have all the methods of a Dataset on the DatasetDict
The end goal would be to merge both Dataset and DatasetDict object in a single object that would be (pretty much totally) backward compatible with both.
There are a few things that we could discuss if we want to merge Dataset and DatasetDict:
1. what happens if you index by a string ? Does it return the column or the split ? We could disallow conflicts between column names and split names to avoid ambiguities. It can be surprising to be able to get a column or a split using the same indexing feature
```
from datasets import load_dataset
dataset = load_dataset(...)
dataset["train"]
dataset["input_ids"]
```
2. what happens when you iterate over the object ? I guess it should iterate over the examples as a Dataset object, but a DatasetDict used to iterate over the splits as they are the dictionary keys. This is a breaking change that we can discuss.
Moreover regarding your points:
- integers are not allowed as split names already
- it's definitely doable to have all the methods. Maybe some of them like train_test_split that is currently only available for Dataset can be tweaked to work for a split dataset
cc: @thomwolf @lhoestq
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2462/reactions",
"total_count": 2,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2462/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2396
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2396/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2396/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2396/events
|
https://github.com/huggingface/datasets/issues/2396
| 899,016,308
|
MDU6SXNzdWU4OTkwMTYzMDg=
| 2,396
|
strange datasets from OSCAR corpus
|
{
"login": "jerryIsHere",
"id": 50871412,
"node_id": "MDQ6VXNlcjUwODcxNDEy",
"avatar_url": "https://avatars.githubusercontent.com/u/50871412?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jerryIsHere",
"html_url": "https://github.com/jerryIsHere",
"followers_url": "https://api.github.com/users/jerryIsHere/followers",
"following_url": "https://api.github.com/users/jerryIsHere/following{/other_user}",
"gists_url": "https://api.github.com/users/jerryIsHere/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jerryIsHere/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jerryIsHere/subscriptions",
"organizations_url": "https://api.github.com/users/jerryIsHere/orgs",
"repos_url": "https://api.github.com/users/jerryIsHere/repos",
"events_url": "https://api.github.com/users/jerryIsHere/events{/privacy}",
"received_events_url": "https://api.github.com/users/jerryIsHere/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! Thanks for reporting\r\ncc @pjox is this an issue from the data ?\r\n\r\nAnyway we should at least mention that OSCAR could contain such contents in the dataset card, you're totally right @jerryIsHere ",
"Hi @jerryIsHere , sorry for the late response! Sadly this is normal, the problem comes form fasttext's classifier which we used to create the original corpus. In general the classifier is not really capable of properly recognizing Yue Chineese so the file ends un being just noise from Common Crawl. Some of these problems with OSCAR were already discussed [here](https://arxiv.org/pdf/2103.12028.pdf) but we are working on explicitly documenting the problems by language on our website. In fact, could please you open an issue on [our repo](https://github.com/oscar-corpus/oscar-website/issues) as well so that we can track it?"
] | 2021-05-23T13:06:02
| 2021-06-17T13:54:37
| null |
CONTRIBUTOR
| null | null | null |
From the [official site ](https://oscar-corpus.com/), the Yue Chinese dataset should have 2.2KB data.
7 training instances is obviously not a right number.
As I can read Yue Chinese, I call tell the last instance is definitely not something that would appear on Common Crawl.
And even if you don't read Yue Chinese, you can tell the first six instance are problematic.
(It is embarrassing, as the 7 training instances look exactly like something from a pornographic novel or flitting messages in a chat of a dating app)
It might not be the problem of the huggingface/datasets implementation, because when I tried to download the dataset from the official site, I found out that the zip file is corrupted.
I will try to inform the host of OSCAR corpus later.
Awy a remake about this dataset in huggingface/datasets is needed, perhaps after the host of the dataset fixes the issue.
> Hi @jerryIsHere , sorry for the late response! Sadly this is normal, the problem comes form fasttext's classifier which we used to create the original corpus. In general the classifier is not really capable of properly recognizing Yue Chineese so the file ends un being just noise from Common Crawl. Some of these problems with OSCAR were already discussed [here](https://arxiv.org/pdf/2103.12028.pdf) but we are working on explicitly documenting the problems by language on our website. In fact, could please you open an issue on [our repo](https://github.com/oscar-corpus/oscar-website/issues) as well so that we can track it?
Thanks a lot, the new post is here:
https://github.com/oscar-corpus/oscar-website/issues/11
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2396/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2396/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2378
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2378/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2378/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2378/events
|
https://github.com/huggingface/datasets/issues/2378
| 895,131,774
|
MDU6SXNzdWU4OTUxMzE3NzQ=
| 2,378
|
Add missing dataset_infos.json files
|
{
"login": "lewtun",
"id": 26859204,
"node_id": "MDQ6VXNlcjI2ODU5MjA0",
"avatar_url": "https://avatars.githubusercontent.com/u/26859204?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lewtun",
"html_url": "https://github.com/lewtun",
"followers_url": "https://api.github.com/users/lewtun/followers",
"following_url": "https://api.github.com/users/lewtun/following{/other_user}",
"gists_url": "https://api.github.com/users/lewtun/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lewtun/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lewtun/subscriptions",
"organizations_url": "https://api.github.com/users/lewtun/orgs",
"repos_url": "https://api.github.com/users/lewtun/repos",
"events_url": "https://api.github.com/users/lewtun/events{/privacy}",
"received_events_url": "https://api.github.com/users/lewtun/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
|
{
"login": "lewtun",
"id": 26859204,
"node_id": "MDQ6VXNlcjI2ODU5MjA0",
"avatar_url": "https://avatars.githubusercontent.com/u/26859204?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lewtun",
"html_url": "https://github.com/lewtun",
"followers_url": "https://api.github.com/users/lewtun/followers",
"following_url": "https://api.github.com/users/lewtun/following{/other_user}",
"gists_url": "https://api.github.com/users/lewtun/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lewtun/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lewtun/subscriptions",
"organizations_url": "https://api.github.com/users/lewtun/orgs",
"repos_url": "https://api.github.com/users/lewtun/repos",
"events_url": "https://api.github.com/users/lewtun/events{/privacy}",
"received_events_url": "https://api.github.com/users/lewtun/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "lewtun",
"id": 26859204,
"node_id": "MDQ6VXNlcjI2ODU5MjA0",
"avatar_url": "https://avatars.githubusercontent.com/u/26859204?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lewtun",
"html_url": "https://github.com/lewtun",
"followers_url": "https://api.github.com/users/lewtun/followers",
"following_url": "https://api.github.com/users/lewtun/following{/other_user}",
"gists_url": "https://api.github.com/users/lewtun/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lewtun/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lewtun/subscriptions",
"organizations_url": "https://api.github.com/users/lewtun/orgs",
"repos_url": "https://api.github.com/users/lewtun/repos",
"events_url": "https://api.github.com/users/lewtun/events{/privacy}",
"received_events_url": "https://api.github.com/users/lewtun/received_events",
"type": "User",
"site_admin": false
}
] | null |
[] | 2021-05-19T08:11:12
| 2021-05-19T08:11:12
| null |
MEMBER
| null | null | null |
Some of the datasets in `datasets` are missing a `dataset_infos.json` file, e.g.
```
[PosixPath('datasets/chr_en/chr_en.py'), PosixPath('datasets/chr_en/README.md')]
[PosixPath('datasets/telugu_books/README.md'), PosixPath('datasets/telugu_books/telugu_books.py')]
[PosixPath('datasets/reclor/README.md'), PosixPath('datasets/reclor/reclor.py')]
[PosixPath('datasets/json/README.md')]
[PosixPath('datasets/csv/README.md')]
[PosixPath('datasets/wikihow/wikihow.py'), PosixPath('datasets/wikihow/README.md')]
[PosixPath('datasets/c4/c4.py'), PosixPath('datasets/c4/README.md')]
[PosixPath('datasets/text/README.md')]
[PosixPath('datasets/lm1b/README.md'), PosixPath('datasets/lm1b/lm1b.py')]
[PosixPath('datasets/pandas/README.md')]
```
For `json`, `text`, csv`, and `pandas` this is expected, but not for the others which should be fixed
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2378/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2378/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2377
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2377/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2377/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2377/events
|
https://github.com/huggingface/datasets/issues/2377
| 894,918,927
|
MDU6SXNzdWU4OTQ5MTg5Mjc=
| 2,377
|
ArrowDataset.save_to_disk produces files that cannot be read using pyarrow.feather
|
{
"login": "Ark-kun",
"id": 1829149,
"node_id": "MDQ6VXNlcjE4MjkxNDk=",
"avatar_url": "https://avatars.githubusercontent.com/u/1829149?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/Ark-kun",
"html_url": "https://github.com/Ark-kun",
"followers_url": "https://api.github.com/users/Ark-kun/followers",
"following_url": "https://api.github.com/users/Ark-kun/following{/other_user}",
"gists_url": "https://api.github.com/users/Ark-kun/gists{/gist_id}",
"starred_url": "https://api.github.com/users/Ark-kun/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/Ark-kun/subscriptions",
"organizations_url": "https://api.github.com/users/Ark-kun/orgs",
"repos_url": "https://api.github.com/users/Ark-kun/repos",
"events_url": "https://api.github.com/users/Ark-kun/events{/privacy}",
"received_events_url": "https://api.github.com/users/Ark-kun/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! This is because we are actually using the arrow streaming format. We plan to switch to the arrow IPC format.\r\nMore info at #1933 ",
"Not sure if this was resolved, but I am getting a similar error when trying to load a dataset.arrow file directly: `ArrowInvalid: Not an Arrow file`",
"Since we're using the streaming format, you need to use `open_stream`:\r\n\r\n```python\r\nimport pyarrow as pa\r\n\r\ndef in_memory_arrow_table_from_file(filename: str) -> pa.Table:\r\n in_memory_stream = pa.input_stream(filename)\r\n opened_stream = pa.ipc.open_stream(in_memory_stream)\r\n pa_table = opened_stream.read_all()\r\n return pa_table\r\n\r\ndef memory_mapped_arrow_table_from_file(filename: str) -> pa.Table:\r\n memory_mapped_stream = pa.memory_map(filename)\r\n opened_stream = pa.ipc.open_stream(memory_mapped_stream)\r\n pa_table = opened_stream.read_all()\r\n return pa_table\r\n```",
"> 由于我们使用流格式，因此您需要使用`open_stream`：\r\n> \r\n> ```python\r\n> import pyarrow as pa\r\n> \r\n> def in_memory_arrow_table_from_file(filename: str) -> pa.Table:\r\n> in_memory_stream = pa.input_stream(filename)\r\n> opened_stream = pa.ipc.open_stream(in_memory_stream)\r\n> pa_table = opened_stream.read_all()\r\n> return pa_table\r\n> \r\n> def memory_mapped_arrow_table_from_file(filename: str) -> pa.Table:\r\n> memory_mapped_stream = pa.memory_map(filename)\r\n> opened_stream = pa.ipc.open_stream(memory_mapped_stream)\r\n> pa_table = opened_stream.read_all()\r\n> return pa_table\r\n> ```\r\nThank you very much for providing the code that can read arrow file to pa_table and finally to dict, but how to implement the reverse process, how to restore a dict to arrow file?\r\n"
] | 2021-05-19T02:04:37
| 2024-01-18T08:06:15
| null |
NONE
| null | null | null |
## Describe the bug
A clear and concise description of what the bug is.
## Steps to reproduce the bug
```python
from datasets import load_dataset
from pyarrow import feather
dataset = load_dataset('imdb', split='train')
dataset.save_to_disk('dataset_dir')
table = feather.read_table('dataset_dir/dataset.arrow')
```
## Expected results
I expect that the saved dataset can be read by the official Apache Arrow methods.
## Actual results
```
File "/usr/local/lib/python3.7/site-packages/pyarrow/feather.py", line 236, in read_table
reader.open(source, use_memory_map=memory_map)
File "pyarrow/feather.pxi", line 67, in pyarrow.lib.FeatherReader.open
File "pyarrow/error.pxi", line 123, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 85, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: Not a Feather V1 or Arrow IPC file
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: datasets-1.6.2
- Platform: Linux
- Python version: 3.7
- PyArrow version: 0.17.1, also 2.0.0
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2377/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2377/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2360
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2360/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2360/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2360/events
|
https://github.com/huggingface/datasets/issues/2360
| 891,965,964
|
MDU6SXNzdWU4OTE5NjU5NjQ=
| 2,360
|
Automatically detect datasets with compatible task schemas
|
{
"login": "lewtun",
"id": 26859204,
"node_id": "MDQ6VXNlcjI2ODU5MjA0",
"avatar_url": "https://avatars.githubusercontent.com/u/26859204?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lewtun",
"html_url": "https://github.com/lewtun",
"followers_url": "https://api.github.com/users/lewtun/followers",
"following_url": "https://api.github.com/users/lewtun/following{/other_user}",
"gists_url": "https://api.github.com/users/lewtun/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lewtun/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lewtun/subscriptions",
"organizations_url": "https://api.github.com/users/lewtun/orgs",
"repos_url": "https://api.github.com/users/lewtun/repos",
"events_url": "https://api.github.com/users/lewtun/events{/privacy}",
"received_events_url": "https://api.github.com/users/lewtun/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
|
{
"login": "lewtun",
"id": 26859204,
"node_id": "MDQ6VXNlcjI2ODU5MjA0",
"avatar_url": "https://avatars.githubusercontent.com/u/26859204?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lewtun",
"html_url": "https://github.com/lewtun",
"followers_url": "https://api.github.com/users/lewtun/followers",
"following_url": "https://api.github.com/users/lewtun/following{/other_user}",
"gists_url": "https://api.github.com/users/lewtun/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lewtun/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lewtun/subscriptions",
"organizations_url": "https://api.github.com/users/lewtun/orgs",
"repos_url": "https://api.github.com/users/lewtun/repos",
"events_url": "https://api.github.com/users/lewtun/events{/privacy}",
"received_events_url": "https://api.github.com/users/lewtun/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"login": "lewtun",
"id": 26859204,
"node_id": "MDQ6VXNlcjI2ODU5MjA0",
"avatar_url": "https://avatars.githubusercontent.com/u/26859204?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lewtun",
"html_url": "https://github.com/lewtun",
"followers_url": "https://api.github.com/users/lewtun/followers",
"following_url": "https://api.github.com/users/lewtun/following{/other_user}",
"gists_url": "https://api.github.com/users/lewtun/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lewtun/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lewtun/subscriptions",
"organizations_url": "https://api.github.com/users/lewtun/orgs",
"repos_url": "https://api.github.com/users/lewtun/repos",
"events_url": "https://api.github.com/users/lewtun/events{/privacy}",
"received_events_url": "https://api.github.com/users/lewtun/received_events",
"type": "User",
"site_admin": false
}
] | null |
[] | 2021-05-14T14:23:40
| 2021-05-14T14:23:40
| null |
MEMBER
| null | null | null |
See description of #2255 for details.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2360/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2360/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2344
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2344/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2344/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2344/events
|
https://github.com/huggingface/datasets/issues/2344
| 885,331,505
|
MDU6SXNzdWU4ODUzMzE1MDU=
| 2,344
|
Is there a way to join multiple datasets in one?
|
{
"login": "avacaondata",
"id": 35173563,
"node_id": "MDQ6VXNlcjM1MTczNTYz",
"avatar_url": "https://avatars.githubusercontent.com/u/35173563?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/avacaondata",
"html_url": "https://github.com/avacaondata",
"followers_url": "https://api.github.com/users/avacaondata/followers",
"following_url": "https://api.github.com/users/avacaondata/following{/other_user}",
"gists_url": "https://api.github.com/users/avacaondata/gists{/gist_id}",
"starred_url": "https://api.github.com/users/avacaondata/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/avacaondata/subscriptions",
"organizations_url": "https://api.github.com/users/avacaondata/orgs",
"repos_url": "https://api.github.com/users/avacaondata/repos",
"events_url": "https://api.github.com/users/avacaondata/events{/privacy}",
"received_events_url": "https://api.github.com/users/avacaondata/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! We don't have `join`/`merge` on a certain column as in pandas.\r\nMaybe you can just use the [concatenate_datasets](https://huggingface.co/docs/datasets/package_reference/main_classes.html?highlight=concatenate#datasets.concatenate_datasets) function.\r\n",
"Hi! You can use `datasets_sql` for that now. As of recently, PyArrow also supports querying tables via Substrait, so I think we can start adding these methods to the API soon."
] | 2021-05-10T23:16:10
| 2022-10-05T17:27:05
| null |
NONE
| null | null | null |
**Is your feature request related to a problem? Please describe.**
I need to join 2 datasets, one that is in the hub and another I've created from my files. Is there an easy way to join these 2?
**Describe the solution you'd like**
Id like to join them with a merge or join method, just like pandas dataframes.
**Additional context**
If you want to extend an existing dataset with more data, for example for training a language model, you need that functionality. I've not found it in the documentation.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2344/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2344/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2343
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2343/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2343/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2343/events
|
https://github.com/huggingface/datasets/issues/2343
| 883,208,539
|
MDU6SXNzdWU4ODMyMDg1Mzk=
| 2,343
|
Columns are removed before or after map function applied?
|
{
"login": "taghizad3h",
"id": 8199406,
"node_id": "MDQ6VXNlcjgxOTk0MDY=",
"avatar_url": "https://avatars.githubusercontent.com/u/8199406?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/taghizad3h",
"html_url": "https://github.com/taghizad3h",
"followers_url": "https://api.github.com/users/taghizad3h/followers",
"following_url": "https://api.github.com/users/taghizad3h/following{/other_user}",
"gists_url": "https://api.github.com/users/taghizad3h/gists{/gist_id}",
"starred_url": "https://api.github.com/users/taghizad3h/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/taghizad3h/subscriptions",
"organizations_url": "https://api.github.com/users/taghizad3h/orgs",
"repos_url": "https://api.github.com/users/taghizad3h/repos",
"events_url": "https://api.github.com/users/taghizad3h/events{/privacy}",
"received_events_url": "https://api.github.com/users/taghizad3h/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Hi! Columns are removed **after** applying the function and **before** updating the examples with the function's output (as per the docs [here](https://huggingface.co/docs/datasets/package_reference/main_classes#datasets.Dataset.map.remove_columns)). I agree the docs on this should be more clear."
] | 2021-05-10T02:36:20
| 2022-10-24T11:31:55
| null |
NONE
| null | null | null |
## Describe the bug
According to the documentation when applying map function the [remove_columns ](https://huggingface.co/docs/datasets/processing.html#removing-columns) will be removed after they are passed to the function, but in the [source code](https://huggingface.co/docs/datasets/package_reference/main_classes.html#datasets.Dataset.map) it's documented that they are removed before applying function. I thinks the source code doc is more accurate, right?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2343/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2343/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2331
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2331/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2331/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2331/events
|
https://github.com/huggingface/datasets/issues/2331
| 879,031,427
|
MDU6SXNzdWU4NzkwMzE0Mjc=
| 2,331
|
Add Topical-Chat
|
{
"login": "ktangri",
"id": 22266659,
"node_id": "MDQ6VXNlcjIyMjY2NjU5",
"avatar_url": "https://avatars.githubusercontent.com/u/22266659?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ktangri",
"html_url": "https://github.com/ktangri",
"followers_url": "https://api.github.com/users/ktangri/followers",
"following_url": "https://api.github.com/users/ktangri/following{/other_user}",
"gists_url": "https://api.github.com/users/ktangri/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ktangri/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ktangri/subscriptions",
"organizations_url": "https://api.github.com/users/ktangri/orgs",
"repos_url": "https://api.github.com/users/ktangri/repos",
"events_url": "https://api.github.com/users/ktangri/events{/privacy}",
"received_events_url": "https://api.github.com/users/ktangri/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[] | 2021-05-07T13:43:59
| 2021-05-07T13:43:59
| null |
NONE
| null | null | null |
## Adding a Dataset
- **Name:** Topical-Chat
- **Description:** a knowledge-grounded human-human conversation dataset where the underlying knowledge spans 8 broad topics and conversation partners don’t have explicitly defined roles
- **Paper:** https://www.isca-speech.org/archive/Interspeech_2019/pdfs/3079.pdf
- **Data:** https://github.com/alexa/Topical-Chat
- **Motivation:** Good quality, knowledge-grounded dataset that spans a broad range of topics
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2331/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2331/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2294
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2294/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2294/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2294/events
|
https://github.com/huggingface/datasets/issues/2294
| 872,136,075
|
MDU6SXNzdWU4NzIxMzYwNzU=
| 2,294
|
Slow #0 when using map to tokenize.
|
{
"login": "VerdureChen",
"id": 31714566,
"node_id": "MDQ6VXNlcjMxNzE0NTY2",
"avatar_url": "https://avatars.githubusercontent.com/u/31714566?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/VerdureChen",
"html_url": "https://github.com/VerdureChen",
"followers_url": "https://api.github.com/users/VerdureChen/followers",
"following_url": "https://api.github.com/users/VerdureChen/following{/other_user}",
"gists_url": "https://api.github.com/users/VerdureChen/gists{/gist_id}",
"starred_url": "https://api.github.com/users/VerdureChen/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/VerdureChen/subscriptions",
"organizations_url": "https://api.github.com/users/VerdureChen/orgs",
"repos_url": "https://api.github.com/users/VerdureChen/repos",
"events_url": "https://api.github.com/users/VerdureChen/events{/privacy}",
"received_events_url": "https://api.github.com/users/VerdureChen/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi ! Have you tried other values for `preprocessing_num_workers` ? Is it always process 0 that is slower ?\r\nThere are no difference between process 0 and the others except that it processes the first shard of the dataset.",
"Hi, I have found the reason of it. Before using the map function to tokenize the data, I concatenate the wikipedia and bookcorpus first, like this:\r\n```if args.dataset_name1 is not None:\r\n dataset1 = load_dataset(args.dataset_name1, args.dataset_config_name1, split=\"train\")\r\n dataset1 = dataset1.remove_columns('title')\r\n if args.dataset_name2 is not None:\r\n dataset2 = load_dataset(args.dataset_name2, args.dataset_config_name2,split=\"train\")\r\n assert dataset1.features.type == dataset2.features.type, str(dataset1.features.type)+';'+str(dataset2.features.type)\r\n datasets12 = concatenate_datasets([dataset1, dataset2], split='train')\r\n```\r\nWhen I just use one datasets, e.g. wikipedia, the problem seems no longer exist:\r\n\r\n\r\nBookcorpus has more row numbers than Wikipedia, however, it takes much more time to process each batch of wiki than that of bookcorpus. When we first concatenate two datasets and then use _map_ to process the concatenated datasets, e.g. `num_proc=5`, process 0 has to process all of the wikipedia data, causing the problem that #0 takes a longer time to finish the job. \r\n\r\nThe problem is caused by the different characteristic of different datasets. One solution might be using _map_ first to process two datasets seperately, then concatenate the tokenized and processed datasets before input to the `Dataloader`.\r\n\r\n",
"That makes sense ! You can indeed use `map` on both datasets separately and then concatenate.\r\nAnother option is to concatenate, then shuffle, and then `map`."
] | 2021-04-30T08:00:33
| 2021-05-04T11:00:11
| null |
NONE
| null | null | null |
Hi, _datasets_ is really amazing! I am following [run_mlm_no_trainer.py](url) to pre-train BERT, and it uses `tokenized_datasets = raw_datasets.map(
tokenize_function,
batched=True,
num_proc=args.preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=not args.overwrite_cache,
)` to tokenize by multiprocessing. However, I have found that when `num_proc`>1，the process _#0_ is much slower than others.
It looks like this:
It takes more than 12 hours for #0, while others just about half an hour. Could anyone tell me it is normal or not, and is there any methods to speed up it?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2294/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2294/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2277
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2277/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2277/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2277/events
|
https://github.com/huggingface/datasets/pull/2277
| 870,071,994
|
MDExOlB1bGxSZXF1ZXN0NjI1MzI5NjIz
| 2,277
|
Create CacheManager
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2851292821,
"node_id": "MDU6TGFiZWwyODUxMjkyODIx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/refactoring",
"name": "refactoring",
"color": "B67A40",
"default": false,
"description": "Restructuring existing code without changing its external behavior"
}
] |
open
| false
| null |
[] |
{
"url": "https://api.github.com/repos/huggingface/datasets/milestones/8",
"html_url": "https://github.com/huggingface/datasets/milestone/8",
"labels_url": "https://api.github.com/repos/huggingface/datasets/milestones/8/labels",
"id": 6968069,
"node_id": "MI_kwDODunzps4AalMF",
"number": 8,
"title": "1.12",
"description": "Next minor release",
"creator": {
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
},
"open_issues": 4,
"closed_issues": 2,
"state": "open",
"created_at": "2021-07-21T15:34:56",
"updated_at": "2021-10-13T10:26:33",
"due_on": "2021-08-30T07:00:00",
"closed_at": null
}
|
[] | 2021-04-28T15:23:42
| 2022-07-06T15:19:48
| null |
MEMBER
| null | false
|
{
"url": "https://api.github.com/repos/huggingface/datasets/pulls/2277",
"html_url": "https://github.com/huggingface/datasets/pull/2277",
"diff_url": "https://github.com/huggingface/datasets/pull/2277.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/2277.patch",
"merged_at": null
}
|
Perform refactoring to decouple cache functionality (method `as_dataset`).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2277/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2277/timeline
| null | null | true
|
https://api.github.com/repos/huggingface/datasets/issues/2267
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2267/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2267/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2267/events
|
https://github.com/huggingface/datasets/issues/2267
| 868,291,129
|
MDU6SXNzdWU4NjgyOTExMjk=
| 2,267
|
DatasetDict save load Failing test in 1.6 not in 1.5
|
{
"login": "timothyjlaurent",
"id": 2000204,
"node_id": "MDQ6VXNlcjIwMDAyMDQ=",
"avatar_url": "https://avatars.githubusercontent.com/u/2000204?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/timothyjlaurent",
"html_url": "https://github.com/timothyjlaurent",
"followers_url": "https://api.github.com/users/timothyjlaurent/followers",
"following_url": "https://api.github.com/users/timothyjlaurent/following{/other_user}",
"gists_url": "https://api.github.com/users/timothyjlaurent/gists{/gist_id}",
"starred_url": "https://api.github.com/users/timothyjlaurent/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/timothyjlaurent/subscriptions",
"organizations_url": "https://api.github.com/users/timothyjlaurent/orgs",
"repos_url": "https://api.github.com/users/timothyjlaurent/repos",
"events_url": "https://api.github.com/users/timothyjlaurent/events{/privacy}",
"received_events_url": "https://api.github.com/users/timothyjlaurent/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Thanks for reporting ! We're looking into it",
"I'm not able to reproduce this, do you think you can provide a code that creates a DatasetDict that has this issue when saving and reloading ?",
"Hi, I just ran into a similar error. Here is the minimal code to reproduce:\r\n```python\r\nfrom datasets import load_dataset, DatasetDict\r\nds = load_dataset('super_glue', 'multirc')\r\n\r\nds.save_to_disk('tempds')\r\n\r\nds = DatasetDict.load_from_disk('tempds')\r\n\r\n```\r\n\r\n```bash\r\nReusing dataset super_glue (/home/idahl/.cache/huggingface/datasets/super_glue/multirc/1.0.2/2fb163bca9085c1deb906aff20f00c242227ff704a4e8c9cfdfe820be3abfc83)\r\nTraceback (most recent call last):\r\n File \"/home/idahl/eval-util-expl/multirc/tmp.py\", line 7, in <module>\r\n ds = DatasetDict.load_from_disk('tempds')\r\n File \"/home/idahl/miniconda3/envs/eval-util-expl/lib/python3.9/site-packages/datasets/dataset_dict.py\", line 710, in load_from_disk\r\n dataset_dict[k] = Dataset.load_from_disk(dataset_dict_split_path, fs, keep_in_memory=keep_in_memory)\r\n File \"/home/idahl/miniconda3/envs/eval-util-expl/lib/python3.9/site-packages/datasets/arrow_dataset.py\", line 687, in load_from_disk\r\n return Dataset(\r\n File \"/home/idahl/miniconda3/envs/eval-util-expl/lib/python3.9/site-packages/datasets/arrow_dataset.py\", line 274, in __init__\r\n raise ValueError(\r\nValueError: External features info don't match the dataset:\r\nGot\r\n{'answer': Value(dtype='string', id=None), 'idx': {'answer': Value(dtype='int32', id=None), 'paragraph': Value(dtype='int32', id=None), 'question': Value(dtype='int32', id=None)}, 'label': ClassLabel(num_classes=2, names=['False', 'True'], names_file=None, id=None), 'paragraph': Value(dtype='string', id=None), 'question': Value(dtype='string', id=None)}\r\nwith type\r\nstruct<answer: string, idx: struct<answer: int32, paragraph: int32, question: int32>, label: int64, paragraph: string, question: string>\r\n\r\nbut expected something like\r\n{'answer': Value(dtype='string', id=None), 'idx': {'paragraph': Value(dtype='int32', id=None), 'question': Value(dtype='int32', id=None), 'answer': Value(dtype='int32', id=None)}, 'label': Value(dtype='int64', id=None), 'paragraph': Value(dtype='string', id=None), 'question': Value(dtype='string', id=None)}\r\nwith type\r\nstruct<answer: string, idx: struct<paragraph: int32, question: int32, answer: int32>, label: int64, paragraph: string, question: string>\r\n\r\n```\r\n\r\nThe non-matching part seems to be\r\n`'label': ClassLabel(num_classes=2, names=['False', 'True'], names_file=None, id=None),`\r\nvs \r\n`'label': Value(dtype='int64', id=None),`\r\n\r\nAnd the order in the `<struct...` being different, which might cause the [features.type != inferred_features.type](https://github.com/huggingface/datasets/blob/master/src/datasets/arrow_dataset.py#L274) condition to become true and raise this ValueError.\r\n\r\n\r\nI am using datasets version 1.6.2.\r\n\r\nEdit: can confirm, this works without error in version 1.5.0",
"My current workaround is to remove the idx feature:\r\n\r\n```\r\n\r\nfrom datasets import load_dataset, DatasetDict, Value\r\nds = load_dataset('super_glue', 'multirc')\r\nds = ds.remove_columns('idx')\r\n\r\nds.save_to_disk('tempds')\r\n\r\nds = DatasetDict.load_from_disk('tempds')\r\n\r\n```\r\n\r\nworks.",
"It looks like this issue comes from the order of the fields in the 'idx' struct that is different for some reason.\r\nI'm looking into it. Note that as a workaround you can also flatten the nested features with `ds = ds.flatten()`",
"I just pushed a fix on `master`. We'll do a new release soon !\r\n\r\nThanks for reporting"
] | 2021-04-27T00:03:25
| 2021-05-28T15:27:34
| null |
NONE
| null | null | null |
## Describe the bug
We have a test that saves a DatasetDict to disk and then loads it from disk. In 1.6 there is an incompatibility in the schema.
Downgrading to `>1.6` -- fixes the problem.
## Steps to reproduce the bug
```python
### Load a dataset dict from jsonl
path = '/test/foo'
ds_dict.save_to_disk(path)
ds_from_disk = DatasetDict.load_from_disk(path). ## <-- this is where I see the error on 1.6
```
## Expected results
Upgrading to 1.6 shouldn't break that test. We should be able to serialize to and from disk.
## Actual results
```
# Infer features if None
inferred_features = Features.from_arrow_schema(arrow_table.schema)
if self.info.features is None:
self.info.features = inferred_features
# Infer fingerprint if None
if self._fingerprint is None:
self._fingerprint = generate_fingerprint(self)
# Sanity checks
assert self.features is not None, "Features can't be None in a Dataset object"
assert self._fingerprint is not None, "Fingerprint can't be None in a Dataset object"
if self.info.features.type != inferred_features.type:
> raise ValueError(
"External features info don't match the dataset:\nGot\n{}\nwith type\n{}\n\nbut expected something like\n{}\nwith type\n{}".format(
self.info.features, self.info.features.type, inferred_features, inferred_features.type
)
)
E ValueError: External features info don't match the dataset:
E Got
E {'_input_hash': Value(dtype='int64', id=None), '_task_hash': Value(dtype='int64', id=None), '_view_id': Value(dtype='string', id=None), 'answer': Value(dtype='string', id=None), 'encoding__ids': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'encoding__offsets': Sequence(feature=Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), length=-1, id=None), 'encoding__overflowing': Sequence(feature=Value(dtype='null', id=None), length=-1, id=None), 'encoding__tokens': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'encoding__words': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'ner_ids': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'ner_labels': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'relations': [{'child': Value(dtype='int64', id=None), 'child_span': {'end': Value(dtype='int64', id=None), 'label': Value(dtype='string', id=None), 'start': Value(dtype='int64', id=None), 'token_end': Value(dtype='int64', id=None), 'token_start': Value(dtype='int64', id=None)}, 'color': Value(dtype='string', id=None), 'head': Value(dtype='int64', id=None), 'head_span': {'end': Value(dtype='int64', id=None), 'label': Value(dtype='string', id=None), 'start': Value(dtype='int64', id=None), 'token_end': Value(dtype='int64', id=None), 'token_start': Value(dtype='int64', id=None)}, 'label': Value(dtype='string', id=None)}], 'spans': [{'end': Value(dtype='int64', id=None), 'label': Value(dtype='string', id=None), 'start': Value(dtype='int64', id=None), 'text': Value(dtype='string', id=None), 'token_end': Value(dtype='int64', id=None), 'token_start': Value(dtype='int64', id=None), 'type': Value(dtype='string', id=None)}], 'text': Value(dtype='string', id=None), 'tokens': [{'disabled': Value(dtype='bool', id=None), 'end': Value(dtype='int64', id=None), 'id': Value(dtype='int64', id=None), 'start': Value(dtype='int64', id=None), 'text': Value(dtype='string', id=None), 'ws': Value(dtype='bool', id=None)}]}
E with type
E struct<_input_hash: int64, _task_hash: int64, _view_id: string, answer: string, encoding__ids: list<item: int64>, encoding__offsets: list<item: list<item: int64>>, encoding__overflowing: list<item: null>, encoding__tokens: list<item: string>, encoding__words: list<item: int64>, ner_ids: list<item: int64>, ner_labels: list<item: string>, relations: list<item: struct<child: int64, child_span: struct<end: int64, label: string, start: int64, token_end: int64, token_start: int64>, color: string, head: int64, head_span: struct<end: int64, label: string, start: int64, token_end: int64, token_start: int64>, label: string>>, spans: list<item: struct<end: int64, label: string, start: int64, text: string, token_end: int64, token_start: int64, type: string>>, text: string, tokens: list<item: struct<disabled: bool, end: int64, id: int64, start: int64, text: string, ws: bool>>>
E
E but expected something like
E {'_input_hash': Value(dtype='int64', id=None), '_task_hash': Value(dtype='int64', id=None), '_view_id': Value(dtype='string', id=None), 'answer': Value(dtype='string', id=None), 'encoding__ids': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'encoding__offsets': Sequence(feature=Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), length=-1, id=None), 'encoding__overflowing': Sequence(feature=Value(dtype='null', id=None), length=-1, id=None), 'encoding__tokens': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'encoding__words': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'ner_ids': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'ner_labels': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'relations': [{'head': Value(dtype='int64', id=None), 'child': Value(dtype='int64', id=None), 'head_span': {'start': Value(dtype='int64', id=None), 'end': Value(dtype='int64', id=None), 'token_start': Value(dtype='int64', id=None), 'token_end': Value(dtype='int64', id=None), 'label': Value(dtype='string', id=None)}, 'child_span': {'start': Value(dtype='int64', id=None), 'end': Value(dtype='int64', id=None), 'token_start': Value(dtype='int64', id=None), 'token_end': Value(dtype='int64', id=None), 'label': Value(dtype='string', id=None)}, 'color': Value(dtype='string', id=None), 'label': Value(dtype='string', id=None)}], 'spans': [{'text': Value(dtype='string', id=None), 'start': Value(dtype='int64', id=None), 'token_start': Value(dtype='int64', id=None), 'token_end': Value(dtype='int64', id=None), 'end': Value(dtype='int64', id=None), 'type': Value(dtype='string', id=None), 'label': Value(dtype='string', id=None)}], 'text': Value(dtype='string', id=None), 'tokens': [{'text': Value(dtype='string', id=None), 'start': Value(dtype='int64', id=None), 'end': Value(dtype='int64', id=None), 'id': Value(dtype='int64', id=None), 'ws': Value(dtype='bool', id=None), 'disabled': Value(dtype='bool', id=None)}]}
E with type
E struct<_input_hash: int64, _task_hash: int64, _view_id: string, answer: string, encoding__ids: list<item: int64>, encoding__offsets: list<item: list<item: int64>>, encoding__overflowing: list<item: null>, encoding__tokens: list<item: string>, encoding__words: list<item: int64>, ner_ids: list<item: int64>, ner_labels: list<item: string>, relations: list<item: struct<head: int64, child: int64, head_span: struct<start: int64, end: int64, token_start: int64, token_end: int64, label: string>, child_span: struct<start: int64, end: int64, token_start: int64, token_end: int64, label: string>, color: string, label: string>>, spans: list<item: struct<text: string, start: int64, token_start: int64, token_end: int64, end: int64, type: string, label: string>>, text: string, tokens: list<item: struct<text: string, start: int64, end: int64, id: int64, ws: bool, disabled: bool>>>
../../../../../.virtualenvs/tf_ner_rel_lib/lib/python3.8/site-packages/datasets/arrow_dataset.py:274: ValueError
```
## Versions
- Datasets: 1.6.1
- Python: 3.8.5 (default, Jan 26 2021, 10:01:04)
[Clang 12.0.0 (clang-1200.0.32.2)]
- Platform: macOS-10.15.7-x86_64-i386-64bit
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2267/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2267/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2251
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2251/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2251/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2251/events
|
https://github.com/huggingface/datasets/issues/2251
| 865,848,705
|
MDU6SXNzdWU4NjU4NDg3MDU=
| 2,251
|
while running run_qa.py, ran into a value error
|
{
"login": "nlee0212",
"id": 44570724,
"node_id": "MDQ6VXNlcjQ0NTcwNzI0",
"avatar_url": "https://avatars.githubusercontent.com/u/44570724?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/nlee0212",
"html_url": "https://github.com/nlee0212",
"followers_url": "https://api.github.com/users/nlee0212/followers",
"following_url": "https://api.github.com/users/nlee0212/following{/other_user}",
"gists_url": "https://api.github.com/users/nlee0212/gists{/gist_id}",
"starred_url": "https://api.github.com/users/nlee0212/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/nlee0212/subscriptions",
"organizations_url": "https://api.github.com/users/nlee0212/orgs",
"repos_url": "https://api.github.com/users/nlee0212/repos",
"events_url": "https://api.github.com/users/nlee0212/events{/privacy}",
"received_events_url": "https://api.github.com/users/nlee0212/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2021-04-23T07:51:03
| 2021-04-23T07:51:03
| null |
NONE
| null | null | null |
command:
python3 run_qa.py --model_name_or_path hyunwoongko/kobart --dataset_name squad_kor_v2 --do_train --do_eval --per_device_train_batch_size 8 --learning_rate 3e-5 --num_train_epochs 3 --max_seq_length 512 --doc_stride 128 --output_dir /tmp/debug_squad/
error:
ValueError: External features info don't match the dataset:
Got
{'id': Value(dtype='string', id=None), 'title': Value(dtype='string', id=None), 'context': Value(dtype='string', id=None), 'question': Value(dtype='string', id=None), 'answer': {'text': Value(dtype='string', id=None), 'answer_start': Value(dtype='int32', id=None), 'html_answer_start': Value(dtype='int32', id=None)}, 'url': Value(dtype='string', id=None), 'raw_html': Value(dtype='string', id=None)}
with type
struct<answer: struct<text: string, answer_start: int32, html_answer_start: int32>, context: string, id: string, question: string, raw_html: string, title: string, url: string>
but expected something like
{'answer': {'answer_start': Value(dtype='int32', id=None), 'html_answer_start': Value(dtype='int32', id=None), 'text': Value(dtype='string', id=None)}, 'context': Value(dtype='string', id=None), 'id': Value(dtype='string', id=None), 'question': Value(dtype='string', id=None), 'raw_html': Value(dtype='string', id=None), 'title': Value(dtype='string', id=None), 'url': Value(dtype='string', id=None)}
with type
struct<answer: struct<answer_start: int32, html_answer_start: int32, text: string>, context: string, id: string, question: string, raw_html: string, title: string, url: string>
I didn't encounter this error 4 hours ago. any solutions for this kind of issue?
looks like gained dataset format refers to 'Data Fields', while expected refers to 'Data Instances'.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2251/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2251/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2249
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2249/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2249/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2249/events
|
https://github.com/huggingface/datasets/pull/2249
| 865,257,826
|
MDExOlB1bGxSZXF1ZXN0NjIxMzU1MzE3
| 2,249
|
Allow downloading/processing/caching only specific splits
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] |
{
"url": "https://api.github.com/repos/huggingface/datasets/milestones/8",
"html_url": "https://github.com/huggingface/datasets/milestone/8",
"labels_url": "https://api.github.com/repos/huggingface/datasets/milestones/8/labels",
"id": 6968069,
"node_id": "MI_kwDODunzps4AalMF",
"number": 8,
"title": "1.12",
"description": "Next minor release",
"creator": {
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
},
"open_issues": 4,
"closed_issues": 2,
"state": "open",
"created_at": "2021-07-21T15:34:56",
"updated_at": "2021-10-13T10:26:33",
"due_on": "2021-08-30T07:00:00",
"closed_at": null
}
|
[
"> If you pass a dictionary like this:\r\n> \r\n> ```\r\n> {\"main_metadata\": url_to_main_data,\r\n> \"secondary_metadata\": url_to_sec_data,\r\n> \"train\": url_train_data,\r\n> \"test\": url_test_data}\r\n> ```\r\n> \r\n> then only the train or test keys will be kept, which I feel not intuitive.\r\n> \r\n> For example if the users asks to load the \"train\" split, then the main and secondary metadata won't be downloaded.\r\n> You can fix that by keeping all the keys except the splits to ignore\r\n\r\nHi @lhoestq, I have been thinking about this and I think it is worth that we discuss about it.\r\n\r\nWhen I created this PR, my first idea was to create a \"hack\" inside the download manager that will be able to filter some split(s) without touching any dataset script. Of course, the download manager does not know about splits logic, and thus this trick would only work for some very specific datasets: only the ones containing that pass a dict to the download manager containing only the keys \"train\", \"validation\", \"test\" (or the one passed by the user for advanced users knowing they can do it), e.g. the `natural_questions` dataset (which was one of the targets).\r\n\r\nThe big inconvenient of this approach is that it is not applicable to many datasets (or worse, it should be constantly tweaked to cope with exceptional cases). One exceptional case is the one you pointed out. But I see others:\r\n- the split keys can be different: train, test, dev, val, validation, eval,...\r\n- in `hope_edi` dataset, the split keys are: TRAIN_DOWNLOAD_URL, VALIDATION_DOWNLOAD_URL\r\n- in `few_rel` dataset, the split keys are: train_wiki, val_nyt, val_pubmed,..., pid2name\r\n- in `curiosity_dialogs`, the split keys are: train, val, test, test_zero; this means that for every split we pass, we will also get test_zero\r\n- in `deal_or_no_dialog`, each of the splits URL is passed separately to the download manager, so all splits would be always downloaded\r\n- etc.\r\n\r\nThen after discussing, another idea emerged: pass a `split` parameter to `_split_generators`, which know about the splits logic, so that it can handle which splits are passed to the download manager. This approach is more accurate and can be tweaked so that it works with all the datasets we want. The only inconvenient is that then for every target dataset, we must modify its corresponding `_split_generators` script method.\r\n\r\nMy point is that I don't think it is a good idea to implement both approaches. They could even interfere with each other! \r\n\r\nIf you agree, I would implement ONLY the second one, which is simpler, more consistent and stable and will avoid future problems.",
"Hi @albertvillanova !\r\nYup I agree with you, implementing the 2nd approach seems to be the right solution"
] | 2021-04-22T17:51:44
| 2022-07-06T15:19:48
| null |
MEMBER
| null | false
|
{
"url": "https://api.github.com/repos/huggingface/datasets/pulls/2249",
"html_url": "https://github.com/huggingface/datasets/pull/2249",
"diff_url": "https://github.com/huggingface/datasets/pull/2249.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/2249.patch",
"merged_at": null
}
|
Allow downloading/processing/caching only specific splits without downloading/processing/caching the other splits.
This PR implements two steps to handle only specific splits:
- it allows processing/caching only specific splits into Arrow files
- for some simple cases, it allows downloading only specific splits (which is more intricate as it depends on the user-defined method `_split_generators`)
This PR makes several assumptions:
- `DownloadConfig` contains the configuration settings for downloading
- the parameter `split` passed to `load_dataset` is just a parameter for loading (from cache), not for downloading
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2249/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2249/timeline
| null | null | true
|
https://api.github.com/repos/huggingface/datasets/issues/2244
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2244/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2244/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2244/events
|
https://github.com/huggingface/datasets/pull/2244
| 863,029,946
|
MDExOlB1bGxSZXF1ZXN0NjE5NTAyODc0
| 2,244
|
Set specific cache directories per test function call
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] |
{
"url": "https://api.github.com/repos/huggingface/datasets/milestones/8",
"html_url": "https://github.com/huggingface/datasets/milestone/8",
"labels_url": "https://api.github.com/repos/huggingface/datasets/milestones/8/labels",
"id": 6968069,
"node_id": "MI_kwDODunzps4AalMF",
"number": 8,
"title": "1.12",
"description": "Next minor release",
"creator": {
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
},
"open_issues": 4,
"closed_issues": 2,
"state": "open",
"created_at": "2021-07-21T15:34:56",
"updated_at": "2021-10-13T10:26:33",
"due_on": "2021-08-30T07:00:00",
"closed_at": null
}
|
[
"@lhoestq, I think this reaches some memory limit on Linux instances... (?)",
"It looks like the `comet` metric test fails because it tries to load a model in memory.\r\nIn the tests I think we have `patch_comet` that mocks the model download + inference. Not sure why it didn't work though.\r\nI can take a look tomorrow (this afternoon is the pytorch ecosystem day)",
"@lhoestq thanks for the hint: I'm going to have a look at that mock... ;)",
"@lhoestq finally I did not find out why the mock is not used... If you can give me some other hint tomorrow..."
] | 2021-04-20T17:06:22
| 2022-07-06T15:19:48
| null |
MEMBER
| null | false
|
{
"url": "https://api.github.com/repos/huggingface/datasets/pulls/2244",
"html_url": "https://github.com/huggingface/datasets/pull/2244",
"diff_url": "https://github.com/huggingface/datasets/pull/2244.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/2244.patch",
"merged_at": null
}
|
Implement specific cache directories (datasets, metrics and modules) per test function call.
Currently, the cache directories are set within the temporary test directory, but they are shared across all test function calls.
This PR implements specific cache directories for each test function call, so that tests are atomic and there are no side effects.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2244/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2244/timeline
| null | null | true
|
https://api.github.com/repos/huggingface/datasets/issues/2237
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2237/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2237/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2237/events
|
https://github.com/huggingface/datasets/issues/2237
| 861,427,439
|
MDU6SXNzdWU4NjE0Mjc0Mzk=
| 2,237
|
Update Dataset.dataset_size after transformed with map
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"@albertvillanova I would like to take this up. It would be great if you could point me as to how the dataset size is calculated in HF. Thanks!"
] | 2021-04-19T15:19:38
| 2021-04-20T14:22:05
| null |
MEMBER
| null | null | null |
After loading a dataset, if we transform it by using `.map` its `dataset_size` attirbute is not updated.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2237/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2237/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2236
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2236/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2236/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2236/events
|
https://github.com/huggingface/datasets/issues/2236
| 861,388,145
|
MDU6SXNzdWU4NjEzODgxNDU=
| 2,236
|
Request to add StrategyQA dataset
|
{
"login": "sarahwie",
"id": 8027676,
"node_id": "MDQ6VXNlcjgwMjc2NzY=",
"avatar_url": "https://avatars.githubusercontent.com/u/8027676?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sarahwie",
"html_url": "https://github.com/sarahwie",
"followers_url": "https://api.github.com/users/sarahwie/followers",
"following_url": "https://api.github.com/users/sarahwie/following{/other_user}",
"gists_url": "https://api.github.com/users/sarahwie/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sarahwie/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sarahwie/subscriptions",
"organizations_url": "https://api.github.com/users/sarahwie/orgs",
"repos_url": "https://api.github.com/users/sarahwie/repos",
"events_url": "https://api.github.com/users/sarahwie/events{/privacy}",
"received_events_url": "https://api.github.com/users/sarahwie/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[] | 2021-04-19T14:46:26
| 2021-04-19T14:46:26
| null |
NONE
| null | null | null |
## Request to add StrategyQA dataset
- **Name:** StrategyQA
- **Description:** open-domain QA [(project page)](https://allenai.org/data/strategyqa)
- **Paper:** [url](https://arxiv.org/pdf/2101.02235.pdf)
- **Data:** [here](https://allenai.org/data/strategyqa)
- **Motivation:** uniquely-formulated dataset that also includes a question-decomposition breakdown and associated Wikipedia annotations for each step. Good for multi-hop reasoning modeling.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2236/reactions",
"total_count": 2,
"+1": 2,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2236/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2228
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2228/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2228/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2228/events
|
https://github.com/huggingface/datasets/pull/2228
| 859,795,563
|
MDExOlB1bGxSZXF1ZXN0NjE2ODE2MTQz
| 2,228
|
[WIP] Add ArrayXD support for fixed size list.
|
{
"login": "jblemoine",
"id": 22685854,
"node_id": "MDQ6VXNlcjIyNjg1ODU0",
"avatar_url": "https://avatars.githubusercontent.com/u/22685854?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jblemoine",
"html_url": "https://github.com/jblemoine",
"followers_url": "https://api.github.com/users/jblemoine/followers",
"following_url": "https://api.github.com/users/jblemoine/following{/other_user}",
"gists_url": "https://api.github.com/users/jblemoine/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jblemoine/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jblemoine/subscriptions",
"organizations_url": "https://api.github.com/users/jblemoine/orgs",
"repos_url": "https://api.github.com/users/jblemoine/repos",
"events_url": "https://api.github.com/users/jblemoine/events{/privacy}",
"received_events_url": "https://api.github.com/users/jblemoine/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Awesome thanks ! To fix the CI you just need to merge master into your branch.\r\nThe error is unrelated to your PR"
] | 2021-04-16T13:04:08
| 2022-07-06T15:19:48
| null |
NONE
| null | false
|
{
"url": "https://api.github.com/repos/huggingface/datasets/pulls/2228",
"html_url": "https://github.com/huggingface/datasets/pull/2228",
"diff_url": "https://github.com/huggingface/datasets/pull/2228.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/2228.patch",
"merged_at": null
}
|
Add support for fixed size list for ArrayXD when shape is known . See https://github.com/huggingface/datasets/issues/2146
Since offset are not stored anymore, the file size is now roughly equal to the actual data size.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2228/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2228/timeline
| null | null | true
|
https://api.github.com/repos/huggingface/datasets/issues/2224
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2224/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2224/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2224/events
|
https://github.com/huggingface/datasets/issues/2224
| 857,983,361
|
MDU6SXNzdWU4NTc5ODMzNjE=
| 2,224
|
Raise error if Windows max path length is not disabled
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2021-04-14T14:57:20
| 2021-04-14T14:59:13
| null |
MEMBER
| null | null | null |
On startup, raise an error if Windows max path length is not disabled; ask the user to disable it.
Linked to discussion in #2220.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2224/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2224/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2218
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2218/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2218/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2218/events
|
https://github.com/huggingface/datasets/issues/2218
| 857,238,435
|
MDU6SXNzdWU4NTcyMzg0MzU=
| 2,218
|
Duplicates in the LAMA dataset
|
{
"login": "amarasovic",
"id": 7276193,
"node_id": "MDQ6VXNlcjcyNzYxOTM=",
"avatar_url": "https://avatars.githubusercontent.com/u/7276193?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/amarasovic",
"html_url": "https://github.com/amarasovic",
"followers_url": "https://api.github.com/users/amarasovic/followers",
"following_url": "https://api.github.com/users/amarasovic/following{/other_user}",
"gists_url": "https://api.github.com/users/amarasovic/gists{/gist_id}",
"starred_url": "https://api.github.com/users/amarasovic/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/amarasovic/subscriptions",
"organizations_url": "https://api.github.com/users/amarasovic/orgs",
"repos_url": "https://api.github.com/users/amarasovic/repos",
"events_url": "https://api.github.com/users/amarasovic/events{/privacy}",
"received_events_url": "https://api.github.com/users/amarasovic/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi,\r\n\r\ncurrently the datasets API doesn't have a dedicated function to remove duplicate rows, but since the LAMA dataset is not too big (it fits in RAM), we can leverage pandas to help us remove duplicates:\r\n```python\r\n>>> from datasets import load_dataset, Dataset\r\n>>> dataset = load_dataset('lama', split='train')\r\n>>> dataset = Dataset.from_pandas(dataset.to_pandas().drop_duplicates(subset=...)) # specify a subset of the columns to consider in a list or use all of the columns if None\r\n```\r\n\r\nNote that the same can be achieved with the `Dataset.filter` method but this would requrie some extra work (filter function, speed?).",
"Oh, seems like my question wasn't specified well. I'm _not_ asking how to remove duplicates, but whether duplicates should be removed if I want to do the evaluation on the LAMA dataset as it was proposed in the original paper/repository? In other words, will I get the same result if evaluate on the de-duplicated dataset loaded from HF's `datasets` as the results I'd get if I use the original data format and data processing script in https://github.com/facebookresearch/LAMA? ",
"So it looks like the person who added LAMA to the library chose to have one item per piece of evidence rather than one per relation - and in this case, there are duplicate pieces of evidence for the target relation\r\n\r\nIf I understand correctly, to reproduce reported results, you would have to aggregate predictions for the several pieces of evidence provided for each relation (each unique `uuid`), but the original authors will know better \r\n\r\ncc @fabiopetroni "
] | 2021-04-13T18:59:49
| 2021-04-14T21:42:27
| null |
NONE
| null | null | null |
I observed duplicates in the LAMA probing dataset, see a minimal code below.
```
>>> import datasets
>>> dataset = datasets.load_dataset('lama')
No config specified, defaulting to: lama/trex
Reusing dataset lama (/home/anam/.cache/huggingface/datasets/lama/trex/1.1.0/97deffae13eca0a18e77dfb3960bb31741e973586f5c1fe1ec0d6b5eece7bddc)
>>> train_dataset = dataset['train']
>>> train_dataset[0]
{'description': 'language or languages a person has learned from early childhood', 'label': 'native language', 'masked_sentence': 'Louis Jules Trochu ([lwi ʒyl tʁɔʃy]; 12 March 1815 – 7 October 1896) was a [MASK] military leader and politician.', 'obj_label': 'French', 'obj_surface': 'French', 'obj_uri': 'Q150', 'predicate_id': 'P103', 'sub_label': 'Louis Jules Trochu', 'sub_surface': 'Louis Jules Trochu', 'sub_uri': 'Q441235', 'template': 'The native language of [X] is [Y] .', 'template_negated': '[X] is not owned by [Y] .', 'type': 'N-1', 'uuid': '40b2ed1c-0961-482e-844e-32596b6117c8'}
>>> train_dataset[1]
{'description': 'language or languages a person has learned from early childhood', 'label': 'native language', 'masked_sentence': 'Louis Jules Trochu ([lwi ʒyl tʁɔʃy]; 12 March 1815 – 7 October 1896) was a [MASK] military leader and politician.', 'obj_label': 'French', 'obj_surface': 'French', 'obj_uri': 'Q150', 'predicate_id': 'P103', 'sub_label': 'Louis Jules Trochu', 'sub_surface': 'Louis Jules Trochu', 'sub_uri': 'Q441235', 'template': 'The native language of [X] is [Y] .', 'template_negated': '[X] is not owned by [Y] .', 'type': 'N-1', 'uuid': '40b2ed1c-0961-482e-844e-32596b6117c8'}
```
I checked the original data available at https://dl.fbaipublicfiles.com/LAMA/data.zip. This particular duplicated comes from:
```
{"uuid": "40b2ed1c-0961-482e-844e-32596b6117c8", "obj_uri": "Q150", "obj_label": "French", "sub_uri": "Q441235", "sub_label": "Louis Jules Trochu", "predicate_id": "P103", "evidences": [{"sub_surface": "Louis Jules Trochu", "obj_surface": "French", "masked_sentence": "Louis Jules Trochu ([lwi \u0292yl t\u0281\u0254\u0283y]; 12 March 1815 \u2013 7 October 1896) was a [MASK] military leader and politician."}, {"sub_surface": "Louis Jules Trochu", "obj_surface": "French", "masked_sentence": "Louis Jules Trochu ([lwi \u0292yl t\u0281\u0254\u0283y]; 12 March 1815 \u2013 7 October 1896) was a [MASK] military leader and politician."}]}
```
What is the best way to deal with these duplicates if I want to use `datasets` to probe with LAMA?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2218/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2218/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2187
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2187/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2187/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2187/events
|
https://github.com/huggingface/datasets/issues/2187
| 852,939,736
|
MDU6SXNzdWU4NTI5Mzk3MzY=
| 2,187
|
Question (potential issue?) related to datasets caching
|
{
"login": "ioana-blue",
"id": 17202292,
"node_id": "MDQ6VXNlcjE3MjAyMjky",
"avatar_url": "https://avatars.githubusercontent.com/u/17202292?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ioana-blue",
"html_url": "https://github.com/ioana-blue",
"followers_url": "https://api.github.com/users/ioana-blue/followers",
"following_url": "https://api.github.com/users/ioana-blue/following{/other_user}",
"gists_url": "https://api.github.com/users/ioana-blue/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ioana-blue/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ioana-blue/subscriptions",
"organizations_url": "https://api.github.com/users/ioana-blue/orgs",
"repos_url": "https://api.github.com/users/ioana-blue/repos",
"events_url": "https://api.github.com/users/ioana-blue/events{/privacy}",
"received_events_url": "https://api.github.com/users/ioana-blue/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892912,
"node_id": "MDU6TGFiZWwxOTM1ODkyOTEy",
"url": "https://api.github.com/repos/huggingface/datasets/labels/question",
"name": "question",
"color": "d876e3",
"default": true,
"description": "Further information is requested"
}
] |
open
| false
| null |
[] | null |
[
"An educated guess: does this refer to the fact that depending on the custom column names in the dataset files (csv in this case), there is a dataset loader being created? and this dataset loader - using the \"custom data configuration\" is used among all jobs running using this particular csv files? (thinking out loud here...)\r\n\r\nIf this is the case, it may be ok for my use case (have to think about it more), still a bit surprising given that datasets caching is disabled (or so I hope) by the lines I pasted above. ",
"Hi ! Currently disabling the caching means that all the dataset transform like `map`, `filter` etc. ignore the cache: it doesn't write nor read processed cache files.\r\nHowever `load_dataset` reuses datasets that have already been prepared: it does reload prepared dataset files.\r\n\r\nIndeed from the documentation:\r\n> datasets.set_caching_enabled(boolean: bool)\r\n\r\n> When applying transforms on a dataset, the data are stored in cache files. The caching mechanism allows to reload an existing cache file if it’s already been computed.\r\n> Reloading a dataset is possible since the cache files are named using the dataset fingerprint, which is updated after each transform.\r\n> If disabled, the library will no longer reload cached datasets files when applying transforms to the datasets. More precisely, if the caching is disabled:\r\n> - cache files are always recreated\r\n> - cache files are written to a temporary directory that is deleted when session closes\r\n> - cache files are named using a random hash instead of the dataset fingerprint - use datasets.Dataset.save_to_disk() to save a transformed dataset or it will be deleted when session closes\r\n> - caching doesn’t affect datasets.load_dataset(). If you want to regenerate a dataset from scratch you should use the download_mode parameter in datasets.load_dataset().",
"Thank you for the clarification. \r\n\r\nThis is a bit confusing. On one hand, it says that cache files are always recreated and written to a temporary directory that is removed; on the other hand the last bullet point makes me think that since the default according to the docs for `download_mode (Optional datasets.GenerateMode) – select the download/generate mode - Default to REUSE_DATASET_IF_EXISTS` => it almost sounds that it could reload prepared dataset files. Where are these files stored? I guess not in the temporary directory that is removed... \r\n\r\nI find this type of api design error-prone. When I see as a programmer `datasets.set_caching_enabled(False)` I expect no reuse of anything in the cache. ",
"It would be nice if the documentation elaborated on all the possible values for `download_mode` and/or a link to `datasets.GenerateMode`. \r\nThis info here:\r\n```\r\n \"\"\"`Enum` for how to treat pre-existing downloads and data.\r\n The default mode is `REUSE_DATASET_IF_EXISTS`, which will reuse both\r\n raw downloads and the prepared dataset if they exist.\r\n The generations modes:\r\n | | Downloads | Dataset |\r\n | -----------------------------------|-----------|---------|\r\n | `REUSE_DATASET_IF_EXISTS` (default)| Reuse | Reuse |\r\n | `REUSE_CACHE_IF_EXISTS` | Reuse | Fresh |\r\n | `FORCE_REDOWNLOAD` | Fresh | Fresh |\r\n```",
"I have another question. Assuming that I understood correctly and there is reuse of datasets files when caching is disabled (!), I'm guessing there is a directory that is created based on some information on the dataset file. I'm interested in the situation where I'm loading a (custom) dataset from local disk. What information is used to create the directory/filenames where the files are stored?\r\n\r\nI'm concerned about the following scenario: if I have a file, let's say `train.csv` at path `the_path`, run once, the dataset is prepared, some models are run, etc. Now let's say there is an issue and I recreate `train.csv` at the same path `the_path`. Is there enough information in the temporary name/hash to *not* reload the *old* prepared dataset (e.g., timestamp of the file)? Or is it going to reload the *old* prepared file? ",
"Thanks for the feedback, we'll work in improving this aspect of the documentation.\r\n\r\n> Where are these files stored? I guess not in the temporary directory that is removed...\r\n\r\nWe're using the Arrow file format to load datasets. Therefore each time you load a dataset, it is prepared as an arrow file on your disk. By default the file is located in the ~/.cache/huggingface/datasets/<dataset_name>/<config_id>/<version> directory.\r\n\r\n> What information is used to create the directory/filenames where the files are stored?\r\n\r\nThe config_id contains a hash that takes into account:\r\n- the dataset loader used and its source code (e.g. the \"csv\" loader)\r\n- the arguments passed to the loader (e.g. the csv delimiter)\r\n- metadata of the local data files if any (e.g. their timestamps)\r\n\r\n> I'm concerned about the following scenario: if I have a file, let's say train.csv at path the_path, run once, the dataset is prepared, some models are run, etc. Now let's say there is an issue and I recreate train.csv at the same path the_path. Is there enough information in the temporary name/hash to not reload the old prepared dataset (e.g., timestamp of the file)? Or is it going to reload the old prepared file?\r\n\r\nYes the timestamp of the local csv file is taken into account. If you edit your csv file, the config_id will change and loading the dataset will create a new arrow file.",
"Thank you for all your clarifications, really helpful! \r\n\r\nIf you have the bandwidth, please do revisit the api wrt cache disabling. Anywhere in the computer stack (hardware included) where you disable the cache, one assumes there is no caching that happens. ",
"That makes total sense indeed !\r\nI think we can do the change",
"I have another question about caching, this time in the case where FORCE_REDOWNLOAD is used to load the dataset, the datasets cache is one directory as defined by HF_HOME and there are multiple concurrent jobs running in a cluster using the same local dataset (i.e., same local files in the cluster). Does anything in the naming convention and/or file access/locking that you're using prevent race conditions between the concurrent jobs on the caching of the local dataset they all use?\r\n\r\nI noticed some errors (can provide more details if helpful) in load_dataset/prepare_split that lead to my question above. \r\n\r\nLet me know if my question is clear, I can elaborate more if needed @lhoestq Thank you!",
"I got another error that convinces me there is a race condition (one of the test files had zero samples at prediction time). I think it comes down to the fact that the `config_id` above (used in the naming for the cache) has no information on who's touching the data. If I have 2 concurrent jobs, both loading the same dataset and forcing redownload, they may step on each other foot/caching of the dataset. ",
"We're using a locking mechanism to prevent two processes from writing at the same time. The locking is based on the `filelock` module.\r\nAlso directories that are being written use a suffix \".incomplete\" so that reading is not possible on a dataset being written.\r\n\r\nDo you think you could provide a simple code to reproduce the race condition you experienced ?",
"I can provide details about the code I'm running (it's really-really close to some official samples from the huggingface transformers examples, I can point to the exact sample file, I kept a record of that). I can also describe in which conditions this race occurs (I'm convinced it has to do with forcing the redownloading of the dataset, I've been running hundreds of experiments before and didn't have a problem before I forced the redownload). I also can provide samples of the different stack errors I get and some details about the level of concurrency of jobs I was running. I can also try to imagine how the race manifests (I'm fairly sure that it's a combo of one job cleaning up and another job being in the middle of the run).\r\n\r\nHowever, I have to cleanup all this to make sure I'm no spilling any info I shouldn't be spilling. I'll try to do it by the end of the week, if you think all this is helpful. \r\n\r\nFor now, I have a workaround. Don't use forcing redownloading. And to be ultra careful (although I don't think this is a problem), I run a series of jobs that will prepare the datasets and I know there is no concurrency wrt the dataset. Once that's done (and I believe even having multiple jobs loading the datasets at the same time doesn't create problems, as long as REUSE_DATASET_IF_EXISTS is the policy for loading the dataset, so the filelock mechanism you're using is working in that scenario), the prepared datasets will be reused, no race possible in any way. \r\n\r\nThanks for all the details you provided, it helped me understand the underlying implementation and coming up with workarounds when I ran into issues. ",
"Hi! I have the same challenge with caching, where the **.cache** folder is required even though it isn't possible for me.\r\n\r\nI'd like to run transformers in Snowflake, using Snowpark for Python, this would mean I could provide configurable transformers in real-time for business users without having data leave an environment (for security reasons). With no need for data transfer,n the compute is faster. It is a large use case - is it possible to entirely disable caching in certain scenarios?\r\n@lhoestq ?\r\n",
"You can try to change the location of the cache folder using the `HF_CACHE_HOME` environment variable, and set a location where you have read/write access.",
"Thanks @lhoestq \r\n\r\nI wanted to do that, however, snowflake does not allow it to write at all. I'm asking around to see if they can help me out with that issue 😅"
] | 2021-04-08T00:16:28
| 2023-01-03T18:30:38
| null |
NONE
| null | null | null |
I thought I had disabled datasets caching in my code, as follows:
```
from datasets import set_caching_enabled
...
def main():
# disable caching in datasets
set_caching_enabled(False)
```
However, in my log files I see messages like the following:
```
04/07/2021 18:34:42 - WARNING - datasets.builder - Using custom data configuration default-888a87931cbc5877
04/07/2021 18:34:42 - WARNING - datasets.builder - Reusing dataset csv (xxxx/cache-transformers/datasets/csv/default-888a87931cbc5877/0.0.0/965b6429be0fc05f975b608ce64e1fa941cc8fb4f30629b523d2390f3c0e1a93
```
Can you please let me know what this reusing dataset csv means? I wouldn't expect any reusing with the datasets caching disabled. Thank you!
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2187/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2187/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2170
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2170/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2170/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2170/events
|
https://github.com/huggingface/datasets/issues/2170
| 850,913,228
|
MDU6SXNzdWU4NTA5MTMyMjg=
| 2,170
|
Wikipedia historic dumps are deleted but hf/datasets hardcodes dump date
|
{
"login": "leezu",
"id": 946903,
"node_id": "MDQ6VXNlcjk0NjkwMw==",
"avatar_url": "https://avatars.githubusercontent.com/u/946903?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/leezu",
"html_url": "https://github.com/leezu",
"followers_url": "https://api.github.com/users/leezu/followers",
"following_url": "https://api.github.com/users/leezu/following{/other_user}",
"gists_url": "https://api.github.com/users/leezu/gists{/gist_id}",
"starred_url": "https://api.github.com/users/leezu/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/leezu/subscriptions",
"organizations_url": "https://api.github.com/users/leezu/orgs",
"repos_url": "https://api.github.com/users/leezu/repos",
"events_url": "https://api.github.com/users/leezu/events{/privacy}",
"received_events_url": "https://api.github.com/users/leezu/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"It seems that this can be fixed from user's end by including a `date` argument, like this:\r\n\r\n`dataset = datasets.load_dataset('wikipedia', '20200501.en', date='20210420')`\r\n\r\nYou can get available dates from [here](https://dumps.wikimedia.org/enwiki/).\r\n\r\nThis is not a proper fix however as all the files will still have '20200501' in their file names."
] | 2021-04-06T03:13:18
| 2021-06-16T01:10:50
| null |
NONE
| null | null | null |
Wikimedia does not keep all historical dumps. For example, as of today https://dumps.wikimedia.org/kowiki/ only provides
```
20201220/ 02-Feb-2021 01:36 -
20210101/ 21-Feb-2021 01:26 -
20210120/ 02-Mar-2021 01:25 -
20210201/ 21-Mar-2021 01:26 -
20210220/ 02-Apr-2021 01:26 -
20210301/ 03-Mar-2021 08:10 -
20210320/ 21-Mar-2021 18:13 -
20210401/ 03-Apr-2021 10:08 -
latest/ 03-Apr-2021 10:08 -
```
However, the wikipedia dataset provided in the library, only supports the following configs, none of which are applicable anymore when disregarding the cached datasets:
```
ValueError: BuilderConfig 20210401.ko not found. Available: ['20200501.aa', '20200501.ab', '20200501.ace', '20200501.ady', '20200501.af', '20200501.ak', '20200501.als', '20200501.am', '20200501.an', '20200501.ang', '20200501.ar', '20200501.arc', '20200501.arz', '20200501.as', '20200501.ast', '20200501.atj', '20200501.av', '20200501.ay', '20200501.az', '20200501.azb', '20200501.ba', '20200501.bar', '20200501.bat-smg', '20200501.bcl', '20200501.be', '20200501.be-x-old', '20200501.bg', '20200501.bh', '20200501.bi', '20200501.bjn', '20200501.bm', '20200501.bn', '20200501.bo', '20200501.bpy', '20200501.br', '20200501.bs', '20200501.bug', '20200501.bxr', '20200501.ca', '20200501.cbk-zam', '20200501.cdo', '20200501.ce', '20200501.ceb', '20200501.ch', '20200501.cho', '20200501.chr', '20200501.chy', '20200501.ckb', '20200501.co', '20200501.cr', '20200501.crh', '20200501.cs', '20200501.csb', '20200501.cu', '20200501.cv', '20200501.cy', '20200501.da', '20200501.de', '20200501.din', '20200501.diq', '20200501.dsb', '20200501.dty', '20200501.dv', '20200501.dz', '20200501.ee', '20200501.el', '20200501.eml', '20200501.en', '20200501.eo', '20200501.es', '20200501.et', '20200501.eu', '20200501.ext', '20200501.fa', '20200501.ff', '20200501.fi', '20200501.fiu-vro', '20200501.fj', '20200501.fo', '20200501.fr', '20200501.frp', '20200501.frr', '20200501.fur', '20200501.fy', '20200501.ga', '20200501.gag', '20200501.gan', '20200501.gd', '20200501.gl', '20200501.glk', '20200501.gn', '20200501.gom', '20200501.gor', '20200501.got', '20200501.gu', '20200501.gv', '20200501.ha', '20200501.hak', '20200501.haw', '20200501.he', '20200501.hi', '20200501.hif', '20200501.ho', '20200501.hr', '20200501.hsb', '20200501.ht', '20200501.hu', '20200501.hy', '20200501.ia', '20200501.id', '20200501.ie', '20200501.ig', '20200501.ii', '20200501.ik', '20200501.ilo', '20200501.inh', '20200501.io', '20200501.is', '20200501.it', '20200501.iu', '20200501.ja', '20200501.jam', '20200501.jbo', '20200501.jv', '20200501.ka', '20200501.kaa', '20200501.kab', '20200501.kbd', '20200501.kbp', '20200501.kg', '20200501.ki', '20200501.kj', '20200501.kk', '20200501.kl', '20200501.km', '20200501.kn', '20200501.ko', '20200501.koi', '20200501.krc', '20200501.ks', '20200501.ksh', '20200501.ku', '20200501.kv', '20200501.kw', '20200501.ky', '20200501.la', '20200501.lad', '20200501.lb', '20200501.lbe', '20200501.lez', '20200501.lfn', '20200501.lg', '20200501.li', '20200501.lij', '20200501.lmo', '20200501.ln', '20200501.lo', '20200501.lrc', '20200501.lt', '20200501.ltg', '20200501.lv', '20200501.mai', '20200501.map-bms', '20200501.mdf', '20200501.mg', '20200501.mh', '20200501.mhr', '20200501.mi', '20200501.min', '20200501.mk', '20200501.ml', '20200501.mn', '20200501.mr', '20200501.mrj', '20200501.ms', '20200501.mt', '20200501.mus', '20200501.mwl', '20200501.my', '20200501.myv', '20200501.mzn', '20200501.na', '20200501.nah', '20200501.nap', '20200501.nds', '20200501.nds-nl', '20200501.ne', '20200501.new', '20200501.ng', '20200501.nl', '20200501.nn', '20200501.no', '20200501.nov', '20200501.nrm', '20200501.nso', '20200501.nv', '20200501.ny', '20200501.oc', '20200501.olo', '20200501.om', '20200501.or', '20200501.os', '20200501.pa', '20200501.pag', '20200501.pam', '20200501.pap', '20200501.pcd', '20200501.pdc', '20200501.pfl', '20200501.pi', '20200501.pih', '20200501.pl', '20200501.pms', '20200501.pnb', '20200501.pnt', '20200501.ps', '20200501.pt', '20200501.qu', '20200501.rm', '20200501.rmy', '20200501.rn', '20200501.ro', '20200501.roa-rup', '20200501.roa-tara', '20200501.ru', '20200501.rue', '20200501.rw', '20200501.sa', '20200501.sah', '20200501.sat', '20200501.sc', '20200501.scn', '20200501.sco', '20200501.sd', '20200501.se', '20200501.sg', '20200501.sh', '20200501.si', '20200501.simple', '20200501.sk', '20200501.sl', '20200501.sm', '20200501.sn', '20200501.so', '20200501.sq', '20200501.sr', '20200501.srn', '20200501.ss', '20200501.st', '20200501.stq', '20200501.su', '20200501.sv', '20200501.sw', '20200501.szl', '20200501.ta', '20200501.tcy', '20200501.te', '20200501.tet', '20200501.tg', '20200501.th', '20200501.ti', '20200501.tk', '20200501.tl', '20200501.tn', '20200501.to', '20200501.tpi', '20200501.tr', '20200501.ts', '20200501.tt', '20200501.tum', '20200501.tw', '20200501.ty', '20200501.tyv', '20200501.udm', '20200501.ug', '20200501.uk', '20200501.ur', '20200501.uz', '20200501.ve', '20200501.vec', '20200501.vep', '20200501.vi', '20200501.vls', '20200501.vo', '20200501.wa', '20200501.war', '20200501.wo', '20200501.wuu', '20200501.xal', '20200501.xh', '20200501.xmf', '20200501.yi', '20200501.yo', '20200501.za', '20200501.zea', '20200501.zh', '20200501.zh-classical', '20200501.zh-min-nan', '20200501.zh-yue', '20200501.zu']
```
The cached datasets:
```
% aws s3 --no-sign-request --endpoint-url https://storage.googleapis.com ls s3://huggingface-nlp/cache/datasets/wikipedia/
PRE 20200501.de/
PRE 20200501.en/
PRE 20200501.fr/
PRE 20200501.frr/
PRE 20200501.it/
PRE 20200501.simple/
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2170/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2170/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2146
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2146/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2146/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2146/events
|
https://github.com/huggingface/datasets/issues/2146
| 844,673,244
|
MDU6SXNzdWU4NDQ2NzMyNDQ=
| 2,146
|
Dataset file size on disk is very large with 3D Array
|
{
"login": "jblemoine",
"id": 22685854,
"node_id": "MDQ6VXNlcjIyNjg1ODU0",
"avatar_url": "https://avatars.githubusercontent.com/u/22685854?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jblemoine",
"html_url": "https://github.com/jblemoine",
"followers_url": "https://api.github.com/users/jblemoine/followers",
"following_url": "https://api.github.com/users/jblemoine/following{/other_user}",
"gists_url": "https://api.github.com/users/jblemoine/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jblemoine/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jblemoine/subscriptions",
"organizations_url": "https://api.github.com/users/jblemoine/orgs",
"repos_url": "https://api.github.com/users/jblemoine/repos",
"events_url": "https://api.github.com/users/jblemoine/events{/privacy}",
"received_events_url": "https://api.github.com/users/jblemoine/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi ! In the arrow file we store all the integers as uint8.\r\nSo your arrow file should weigh around `height x width x n_channels x n_images` bytes.\r\n\r\nWhat feature type do your TFDS dataset have ?\r\n\r\nIf it uses a `tfds.features.Image` type, then what is stored is the encoded data (as png or jpg for example). Since these encodings are made for compression, the resulting tfrecord is smaller that the arrow file.\r\n\r\nWe are working on adding a similar feature in `datasets`: the ability to store the encoded data instead of the raw integers for images, but also for audio data. This way, arrow files will have similar sizes as tfrecords for images.",
"Thanks for the prompt response. You're right about the encoding, I have the `tfds.features.Image` feature type you mentioned.\r\nHowever, as described in the `dataset_info.json`, my dataset is made of 1479 (224x224x3) images. 1479 x 224 x 224 x 3 = 222630912 bytes which is far from the actual size 520803408 bytes. \r\n\r\nAnyway I look forward to the Image feature type in `datasets`. ",
"@lhoestq I changed the data structure so I have a 2D Array feature type instead of a 3D Array by grouping the two last dimensions ( a 224x672 2D Array instead of a 224x224x3 3D Array). The file size is now 223973964 bytes, nearly half the previous size! Which is around of what I would expect.\r\nI found similar behavior in existing `datasets` collection, when comparing black and white vs color image, for example MNIST vs CIFAR. ",
"Interesting !\r\nThis may be because of the offsets that are stored with the array data.\r\n\r\nCurrently the offsets are stored even if the `shape` of the arrays is fixed. This was needed because of some issues with pyarrow a few months ago. I think these issues have been addressed now, so we can probably try to remove them to make the file lighter.\r\n\r\nIdeally in your case the floats data should be 220 MB for both Array2D and Array3D",
"Yeah for sure, can you be a bit more specific about where the offset is stored in the code base ? And any reference to pyarrow issues if you have some. I would be very interested in contributing to `datasets` by trying to fix this issue. ",
"Pyarrow has two types of lists: variable length lists and fixed size lists.\r\nCurrently we store the ArrayXD data as variable length lists. They take more disk space because they must store both actual data and offsets.\r\nIn the `datasets` code this is done here:\r\n\r\nhttps://github.com/huggingface/nlp/blob/dbac87c8a083f806467f5afc4ec9b401a7e4c15c/src/datasets/features.py#L346-L352\r\n\r\nTo use a fixed length list, one should use the `list_size` argument of `pyarrow.list_()`.\r\nI believe this would work directly modulo some changes in the numpy conversion here:\r\n\r\nhttps://github.com/huggingface/nlp/blob/dbac87c8a083f806467f5afc4ec9b401a7e4c15c/src/datasets/features.py#L381-L395"
] | 2021-03-30T14:46:09
| 2021-04-16T13:07:02
| null |
NONE
| null | null | null |
Hi,
I have created my own dataset using the provided dataset loading script. It is an image dataset where images are stored as 3D Array with dtype=uint8.
The actual size on disk is surprisingly large. It takes 520 MB. Here is some info from `dataset_info.json`.
`{
"description": "",
"citation": "",
"homepage": "",
"license": "",
"features": {
"image": {
"shape": [224, 224, 3],
"dtype": "uint8",
"id": null,
"_type": "Array3D",
}
},
"post_processed": null,
"supervised_keys": null,
"builder_name": "shot_type_image_dataset",
"config_name": "default",
"version": {
"version_str": "0.0.0",
"description": null,
"major": 0,
"minor": 0,
"patch": 0,
},
"splits": {
"train": {
"name": "train",
"num_bytes": 520803408,
"num_examples": 1479,
"dataset_name": "shot_type_image_dataset",
}
},
"download_checksums": {
"": {
"num_bytes": 16940447118,
"checksum": "5854035705efe08b0ed8f3cf3da7b4d29cba9055c2d2d702c79785350d72ee03",
}
},
"download_size": 16940447118,
"post_processing_size": null,
"dataset_size": 520803408,
"size_in_bytes": 17461250526,
}`
I have created the same dataset with tensorflow_dataset and it takes only 125MB on disk.
I am wondering, is it normal behavior ? I understand `Datasets` uses Arrow for serialization wheres tf uses TF Records.
This might be a problem for large dataset.
Thanks for your help.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2146/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2146/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2144
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2144/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2144/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2144/events
|
https://github.com/huggingface/datasets/issues/2144
| 844,352,067
|
MDU6SXNzdWU4NDQzNTIwNjc=
| 2,144
|
Loading wikipedia 20200501.en throws pyarrow related error
|
{
"login": "TomPyonsuke",
"id": 26637405,
"node_id": "MDQ6VXNlcjI2NjM3NDA1",
"avatar_url": "https://avatars.githubusercontent.com/u/26637405?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/TomPyonsuke",
"html_url": "https://github.com/TomPyonsuke",
"followers_url": "https://api.github.com/users/TomPyonsuke/followers",
"following_url": "https://api.github.com/users/TomPyonsuke/following{/other_user}",
"gists_url": "https://api.github.com/users/TomPyonsuke/gists{/gist_id}",
"starred_url": "https://api.github.com/users/TomPyonsuke/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/TomPyonsuke/subscriptions",
"organizations_url": "https://api.github.com/users/TomPyonsuke/orgs",
"repos_url": "https://api.github.com/users/TomPyonsuke/repos",
"events_url": "https://api.github.com/users/TomPyonsuke/events{/privacy}",
"received_events_url": "https://api.github.com/users/TomPyonsuke/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"That's how I loaded the dataset\r\n```python\r\nfrom datasets import load_dataset\r\nds = load_dataset('wikipedia', '20200501.en', cache_dir='/usr/local/workspace/NAS_NLP/cache')\r\n```",
"Hi ! It looks like the arrow file in the folder\r\n`/usr/local/workspace/NAS_NLP/cache/wikipedia/20200501.en/1.0.0/50aa706aa417bb77d910ad61211cc672c0ef3e0f224225a5e0a18277ade8b931` is corrupted.\r\n\r\nCan you take a look and check that it's 18.3GB ?\r\n\r\nIf not, then maybe you need to redownload it:\r\n```python\r\nfrom datasets import load_dataset\r\nds = load_dataset('wikipedia', '20200501.en', cache_dir='/usr/local/workspace/NAS_NLP/cache', download_mode=\"force_redownload\")\r\n```",
"> Hi ! It looks like the arrow file in the folder\r\n> `/usr/local/workspace/NAS_NLP/cache/wikipedia/20200501.en/1.0.0/50aa706aa417bb77d910ad61211cc672c0ef3e0f224225a5e0a18277ade8b931` is corrupted.\r\n> \r\n> Can you take a look and check that it's 18.3GB ?\r\n> \r\n> If not, then maybe you need to redownload it:\r\n> \r\n> ```python\r\n> from datasets import load_dataset\r\n> ds = load_dataset('wikipedia', '20200501.en', cache_dir='/usr/local/workspace/NAS_NLP/cache', download_mode=\"force_redownload\")\r\n> ```\r\n\r\nHi Ihoestq, thanks for the reply! Actually i think my issue is i couldn't download the dataset beyond 10.7G. It feels like the whole dataset is split into different volumes and after the first one was downloaded it crashed before proceeding to the next one. I did try 'force_redownload' mode but still got the same issue.",
"I just tried on my side and got no issues.\r\nWhen downloading the dataset again, did it crash at 10.7GB as well ?",
"> I just tried on my side and got no issues.\r\n> When downloading the dataset again, did it crash at 10.7GB as well ?\r\n\r\nYes i have tried it multiple times on different machines. I am wondering if you could share the screenshot of your dependency versions and i will try to make them the same as yours?",
"I tried using `datasets` from `master` on macos with python 3.7.2\r\nI also have `requests==2.23.0` and `tqdm==4.45.0`."
] | 2021-03-30T10:38:31
| 2021-04-01T09:21:17
| null |
NONE
| null | null | null |
**Problem description**
I am getting the following error when trying to load wikipedia/20200501.en dataset.
**Error log**
Downloading and preparing dataset wikipedia/20200501.en (download: 16.99 GiB, generated: 17.07 GiB, post-processed: Unknown size, total: 34.06 GiB) to /usr/local/workspace/NAS_NLP/cache/wikipedia/20200501.en/1.0.0/50aa706aa417bb77d910ad61211cc672c0ef3e0f224225a5e0a18277ade8b931...
Downloading: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 14.6k/14.6k [00:00<00:00, 5.41MB/s]
Downloading: 59%|███████████████████████████████████████████████████████████████████████████████████████▊ | 10.7G/18.3G [11:30<08:08, 15.5MB/s]
Dataset wikipedia downloaded and prepared to /usr/local/workspace/NAS_NLP/cache/wikipedia/20200501.en/1.0.0/50aa706aa417bb77d910ad61211cc672c0ef3e0f224225a5e0a18277ade8b931. Subsequent calls will reuse this data.
Traceback (most recent call last):
File "load_wiki.py", line 2, in <module>
ds = load_dataset('wikipedia', '20200501.en', cache_dir='/usr/local/workspace/NAS_NLP/cache')
File "/usr/local/lib/python3.6/dist-packages/datasets/load.py", line 751, in load_dataset
ds = builder_instance.as_dataset(split=split, ignore_verifications=ignore_verifications, in_memory=keep_in_memory)
File "/usr/local/lib/python3.6/dist-packages/datasets/builder.py", line 746, in as_dataset
map_tuple=True,
File "/usr/local/lib/python3.6/dist-packages/datasets/utils/py_utils.py", line 204, in map_nested
_single_map_nested((function, obj, types, None, True)) for obj in tqdm(iterable, disable=disable_tqdm)
File "/usr/local/lib/python3.6/dist-packages/datasets/utils/py_utils.py", line 204, in <listcomp>
_single_map_nested((function, obj, types, None, True)) for obj in tqdm(iterable, disable=disable_tqdm)
File "/usr/local/lib/python3.6/dist-packages/datasets/utils/py_utils.py", line 142, in _single_map_nested
return function(data_struct)
File "/usr/local/lib/python3.6/dist-packages/datasets/builder.py", line 763, in _build_single_dataset
in_memory=in_memory,
File "/usr/local/lib/python3.6/dist-packages/datasets/builder.py", line 835, in _as_dataset
in_memory=in_memory,
File "/usr/local/lib/python3.6/dist-packages/datasets/arrow_reader.py", line 215, in read
return self.read_files(files=files, original_instructions=instructions, in_memory=in_memory)
File "/usr/local/lib/python3.6/dist-packages/datasets/arrow_reader.py", line 236, in read_files
pa_table = self._read_files(files, in_memory=in_memory)
File "/usr/local/lib/python3.6/dist-packages/datasets/arrow_reader.py", line 171, in _read_files
pa_table: pa.Table = self._get_dataset_from_filename(f_dict, in_memory=in_memory)
File "/usr/local/lib/python3.6/dist-packages/datasets/arrow_reader.py", line 302, in _get_dataset_from_filename
pa_table = ArrowReader.read_table(filename, in_memory=in_memory)
File "/usr/local/lib/python3.6/dist-packages/datasets/arrow_reader.py", line 324, in read_table
pa_table = f.read_all()
File "pyarrow/ipc.pxi", line 544, in pyarrow.lib.RecordBatchReader.read_all
File "pyarrow/error.pxi", line 99, in pyarrow.lib.check_status
OSError: Expected to be able to read 9176784 bytes for message body, got 4918712
**Detailed version info**
datasets==1.5.0
- dataclasses [required: Any, installed: 0.8]
- dill [required: Any, installed: 0.3.3]
- fsspec [required: Any, installed: 0.8.7]
- importlib-metadata [required: Any, installed: 1.7.0]
- zipp [required: >=0.5, installed: 3.1.0]
- huggingface-hub [required: <0.1.0, installed: 0.0.7]
- filelock [required: Any, installed: 3.0.12]
- importlib-metadata [required: Any, installed: 1.7.0]
- zipp [required: >=0.5, installed: 3.1.0]
- requests [required: Any, installed: 2.24.0]
- certifi [required: >=2017.4.17, installed: 2020.6.20]
- chardet [required: >=3.0.2,<4, installed: 3.0.4]
- idna [required: >=2.5,<3, installed: 2.6]
- urllib3 [required: >=1.21.1,<1.26,!=1.25.1,!=1.25.0, installed: 1.25.10]
- tqdm [required: Any, installed: 4.49.0]
- importlib-metadata [required: Any, installed: 1.7.0]
- zipp [required: >=0.5, installed: 3.1.0]
- multiprocess [required: Any, installed: 0.70.11.1]
- dill [required: >=0.3.3, installed: 0.3.3]
- numpy [required: >=1.17, installed: 1.17.0]
- pandas [required: Any, installed: 1.1.5]
- numpy [required: >=1.15.4, installed: 1.17.0]
- python-dateutil [required: >=2.7.3, installed: 2.8.0]
- six [required: >=1.5, installed: 1.15.0]
- pytz [required: >=2017.2, installed: 2020.1]
- pyarrow [required: >=0.17.1, installed: 3.0.0]
- numpy [required: >=1.16.6, installed: 1.17.0]
- requests [required: >=2.19.0, installed: 2.24.0]
- certifi [required: >=2017.4.17, installed: 2020.6.20]
- chardet [required: >=3.0.2,<4, installed: 3.0.4]
- idna [required: >=2.5,<3, installed: 2.6]
- urllib3 [required: >=1.21.1,<1.26,!=1.25.1,!=1.25.0, installed: 1.25.10]
- tqdm [required: >=4.27,<4.50.0, installed: 4.49.0]
- xxhash [required: Any, installed: 2.0.0]
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2144/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2144/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2132
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2132/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2132/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2132/events
|
https://github.com/huggingface/datasets/issues/2132
| 843,142,822
|
MDU6SXNzdWU4NDMxNDI4MjI=
| 2,132
|
TydiQA dataset is mixed and is not split per language
|
{
"login": "dorost1234",
"id": 79165106,
"node_id": "MDQ6VXNlcjc5MTY1MTA2",
"avatar_url": "https://avatars.githubusercontent.com/u/79165106?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dorost1234",
"html_url": "https://github.com/dorost1234",
"followers_url": "https://api.github.com/users/dorost1234/followers",
"following_url": "https://api.github.com/users/dorost1234/following{/other_user}",
"gists_url": "https://api.github.com/users/dorost1234/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dorost1234/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dorost1234/subscriptions",
"organizations_url": "https://api.github.com/users/dorost1234/orgs",
"repos_url": "https://api.github.com/users/dorost1234/repos",
"events_url": "https://api.github.com/users/dorost1234/events{/privacy}",
"received_events_url": "https://api.github.com/users/dorost1234/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"You can filter the languages this way:\r\n```python\r\ntydiqa_en = tydiqa_dataset.filter(lambda x: x[\"language\"] == \"english\")\r\n```\r\n\r\nOtherwise maybe we can have one configuration per language ?\r\nWhat do you think of this for example ?\r\n\r\n```python\r\nload_dataset(\"tydiqa\", \"primary_task.en\")\r\n```",
"Hi\nthank you very much for the great response, this will be really wonderful\nto have one configuration per language, as one need the dataset in majority\nof case per language for cross-lingual evaluations.\nThis becomes also then more close to TFDS format, which is separated per\nlanguage https://www.tensorflow.org/datasets/catalog/tydi_qa which will be\nreally awesome to have.\nthanks\n\nOn Mon, Mar 29, 2021 at 6:17 PM Quentin Lhoest ***@***.***>\nwrote:\n\n> You can filter the languages this way:\n>\n> tydiqa_en = tydiqa_dataset.filter(lambda x: x[\"language\"] == \"english\")\n>\n> Otherwise maybe we can have one configuration per language ?\n> What do you think of this for example ?\n>\n> load_dataset(\"tydiqa\", \"primary_task.en\")\n>\n> —\n> You are receiving this because you authored the thread.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/datasets/issues/2132#issuecomment-809516799>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AS37NMXPW2PWSQ2RHG73O7TTGCY4LANCNFSM4Z7ER7IA>\n> .\n>\n",
"@lhoestq I greatly appreciate any updates on this. thanks a lot"
] | 2021-03-29T08:56:21
| 2021-04-04T09:57:15
| null |
NONE
| null | null | null |
Hi @lhoestq
Currently TydiQA is mixed and user can only access the whole training set of all languages:
https://www.tensorflow.org/datasets/catalog/tydi_qa
for using this dataset, one need to train/evaluate in each separate language, and having them mixed, makes it hard to use this dataset. This is much convenient for user to have them split and I appreciate your help on this.
Meanwhile, till hopefully this is split per language, I greatly appreciate telling me how I can preprocess and get data per language. thanks a lot
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2132/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2132/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2124
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2124/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2124/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2124/events
|
https://github.com/huggingface/datasets/issues/2124
| 842,627,729
|
MDU6SXNzdWU4NDI2Mjc3Mjk=
| 2,124
|
Adding ScaNN library to do MIPS?
|
{
"login": "shamanez",
"id": 16892570,
"node_id": "MDQ6VXNlcjE2ODkyNTcw",
"avatar_url": "https://avatars.githubusercontent.com/u/16892570?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/shamanez",
"html_url": "https://github.com/shamanez",
"followers_url": "https://api.github.com/users/shamanez/followers",
"following_url": "https://api.github.com/users/shamanez/following{/other_user}",
"gists_url": "https://api.github.com/users/shamanez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/shamanez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/shamanez/subscriptions",
"organizations_url": "https://api.github.com/users/shamanez/orgs",
"repos_url": "https://api.github.com/users/shamanez/repos",
"events_url": "https://api.github.com/users/shamanez/events{/privacy}",
"received_events_url": "https://api.github.com/users/shamanez/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"I haven't played with it (yet) but it sounds really cool !\r\n"
] | 2021-03-28T00:07:00
| 2021-03-29T13:23:43
| null |
NONE
| null | null | null |
@lhoestq Hi I am thinking of adding this new google library to do the MIPS similar to **add_faiss_idex**. As the paper suggests, it is really fast when it comes to retrieving the nearest neighbors.
https://github.com/google-research/google-research/tree/master/scann
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2124/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2124/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2108
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2108/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2108/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2108/events
|
https://github.com/huggingface/datasets/issues/2108
| 840,181,055
|
MDU6SXNzdWU4NDAxODEwNTU=
| 2,108
|
Is there a way to use a GPU only when training an Index in the process of add_faisis_index?
|
{
"login": "shamanez",
"id": 16892570,
"node_id": "MDQ6VXNlcjE2ODkyNTcw",
"avatar_url": "https://avatars.githubusercontent.com/u/16892570?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/shamanez",
"html_url": "https://github.com/shamanez",
"followers_url": "https://api.github.com/users/shamanez/followers",
"following_url": "https://api.github.com/users/shamanez/following{/other_user}",
"gists_url": "https://api.github.com/users/shamanez/gists{/gist_id}",
"starred_url": "https://api.github.com/users/shamanez/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/shamanez/subscriptions",
"organizations_url": "https://api.github.com/users/shamanez/orgs",
"repos_url": "https://api.github.com/users/shamanez/repos",
"events_url": "https://api.github.com/users/shamanez/events{/privacy}",
"received_events_url": "https://api.github.com/users/shamanez/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892912,
"node_id": "MDU6TGFiZWwxOTM1ODkyOTEy",
"url": "https://api.github.com/repos/huggingface/datasets/labels/question",
"name": "question",
"color": "d876e3",
"default": true,
"description": "Further information is requested"
}
] |
open
| false
| null |
[] | null |
[] | 2021-03-24T21:32:16
| 2021-03-25T06:31:43
| null |
NONE
| null | null | null |
Motivation - Some FAISS indexes like IVF consist of the training step that clusters the dataset into a given number of indexes. It would be nice if we can use a GPU to do the training step and covert the index back to CPU as mention in [this faiss example](https://gist.github.com/mdouze/46d6bbbaabca0b9778fca37ed2bcccf6).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2108/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2108/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2106
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2106/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2106/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2106/events
|
https://github.com/huggingface/datasets/issues/2106
| 839,084,264
|
MDU6SXNzdWU4MzkwODQyNjQ=
| 2,106
|
WMT19 Dataset for Kazakh-English is not formatted correctly
|
{
"login": "trina731",
"id": 22580542,
"node_id": "MDQ6VXNlcjIyNTgwNTQy",
"avatar_url": "https://avatars.githubusercontent.com/u/22580542?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/trina731",
"html_url": "https://github.com/trina731",
"followers_url": "https://api.github.com/users/trina731/followers",
"following_url": "https://api.github.com/users/trina731/following{/other_user}",
"gists_url": "https://api.github.com/users/trina731/gists{/gist_id}",
"starred_url": "https://api.github.com/users/trina731/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/trina731/subscriptions",
"organizations_url": "https://api.github.com/users/trina731/orgs",
"repos_url": "https://api.github.com/users/trina731/repos",
"events_url": "https://api.github.com/users/trina731/events{/privacy}",
"received_events_url": "https://api.github.com/users/trina731/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067388877,
"node_id": "MDU6TGFiZWwyMDY3Mzg4ODc3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20bug",
"name": "dataset bug",
"color": "2edb81",
"default": false,
"description": "A bug in a dataset script provided in the library"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! Thanks for reporting\r\n\r\nBy looking at the raw `news-commentary-v14.en-kk.tsv` file, it looks like there are at least 17 lines with this issue.\r\nMoreover these issues are not always the same:\r\n- L97 is only `kk` text and must be appended at the end of the `kk` text of the **next** line\r\n- L2897 is only `kk` text and must be appended at the end of the `kk` text of the **previous** line\r\n- L1247 and L1248 are only `kk` texts and must be inserted at the **beginning** of the `kk` text of the next line\r\n- (and there are many others)\r\n\r\nIt would be nice to have a corrected version of this file ! The file is available in the `wmt/news-commentary` repository on the Datasets Hub here:\r\nhttps://huggingface.co/datasets/wmt/news-commentary/tree/main/v14/training\r\n\r\nThen maybe we can notify the WMT authors and host the corrected version somewhere"
] | 2021-03-23T20:14:47
| 2021-03-25T21:36:20
| null |
NONE
| null | null | null |
In addition to the bug of languages being switched from Issue @415, there are incorrect translations in the dataset because the English-Kazakh translations have a one off formatting error.
The News Commentary v14 parallel data set for kk-en from http://www.statmt.org/wmt19/translation-task.html has a bug here:
> Line 94. The Swiss National Bank, for its part, has been battling with the deflationary effects of the franc’s dramatic appreciation over the past few years. Швейцарияның Ұлттық банкі өз тарапынан, соңғы бірнеше жыл ішінде франк құнының қатты өсуінің дефляциялық әсерімен күресіп келеді.
>
> Line 95. Дефляциялық күштер 2008 жылы терең және ұзаққа созылған жаһандық дағдарысқа байланысты орын алған ірі экономикалық және қаржылық орын алмасулардың арқасында босатылды. Жеке қарыз қаражаты үлесінің қысқаруы орталық банктің рефляцияға жұмсалған күш-жігеріне тұрақты соққан қарсы желдей болды.
>
> Line 96. The deflationary forces were unleashed by the major economic and financial dislocations associated with the deep and protracted global crisis that erupted in 2008. Private deleveraging became a steady headwind to central bank efforts to reflate. 2009 жылы, алдыңғы қатарлы экономикалардың шамамен үштен бірі бағаның төмендеуін көрсетті, бұл соғыстан кейінгі жоғары деңгей болды.
As you can see, line 95 has only the Kazakh translation which should be part of line 96. This causes all of the following English-Kazakh translation pairs to be one off rendering ALL of those translations incorrect. This issue was not fixed when the dataset was imported to Huggingface. By running this code
```
import datasets
from datasets import load_dataset
dataset = load_dataset('wmt19', 'kk-en')
for key in dataset['train']['translation']:
if 'The deflationary forces were unleashed by the major economic and financial dislocations associated with the deep and protracted global crisis that erupted in 2008.' in key['kk']:
print(key['en'])
print(key['kk'])
break
```
we get:
> 2009 жылы, алдыңғы қатарлы экономикалардың шамамен үштен бірі бағаның төмендеуін көрсетті, бұл соғыстан кейінгі жоғары деңгей болды.
> The deflationary forces were unleashed by the major economic and financial dislocations associated with the deep and protracted global crisis that erupted in 2008. Private deleveraging became a steady headwind to central bank efforts to reflate.
which shows that the issue still persists in the Huggingface dataset. The Kazakh sentence matches up to the next English sentence in the dataset instead of the current one.
Please let me know if there's you have any ideas to fix this one-off error from the dataset or if this can be fixed by Huggingface.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2106/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2106/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2105
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2105/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2105/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2105/events
|
https://github.com/huggingface/datasets/issues/2105
| 839,059,226
|
MDU6SXNzdWU4MzkwNTkyMjY=
| 2,105
|
Request to remove S2ORC dataset
|
{
"login": "kyleclo",
"id": 13603748,
"node_id": "MDQ6VXNlcjEzNjAzNzQ4",
"avatar_url": "https://avatars.githubusercontent.com/u/13603748?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/kyleclo",
"html_url": "https://github.com/kyleclo",
"followers_url": "https://api.github.com/users/kyleclo/followers",
"following_url": "https://api.github.com/users/kyleclo/following{/other_user}",
"gists_url": "https://api.github.com/users/kyleclo/gists{/gist_id}",
"starred_url": "https://api.github.com/users/kyleclo/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/kyleclo/subscriptions",
"organizations_url": "https://api.github.com/users/kyleclo/orgs",
"repos_url": "https://api.github.com/users/kyleclo/repos",
"events_url": "https://api.github.com/users/kyleclo/events{/privacy}",
"received_events_url": "https://api.github.com/users/kyleclo/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hello @kyleclo! Currently, we are getting the data from your bucket, so if you remove it the HF script won't work anymore :) \r\n\r\nUntil you solve things on your end, @lhoestq suggested we just return a warning message when people try to load that dataset from HF. What would you like it to say?",
"Hi @kyleclo, as of today, you have not removed your bucket data yet, and therefore HuggingFace can download it from there.\r\n\r\nIs it OK? Are you planning to eventually delete it? Thank you.",
"Hi! Sorry I missed @yjernite 's previous message, thanks for responding! \r\n\r\nIs there an option where we can keep our data in our bucket, but the HF script no longer pulls data from it? "
] | 2021-03-23T19:43:06
| 2021-08-04T19:18:02
| null |
NONE
| null | null | null |
Hi! I was wondering if it's possible to remove [S2ORC](https://huggingface.co/datasets/s2orc) from hosting on Huggingface's platform? Unfortunately, there are some legal considerations about how we make this data available. Happy to add back to Huggingface's platform once we work out those hurdles! Thanks!
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2105/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 1
}
|
https://api.github.com/repos/huggingface/datasets/issues/2105/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/2076
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/2076/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/2076/comments
|
https://api.github.com/repos/huggingface/datasets/issues/2076/events
|
https://github.com/huggingface/datasets/issues/2076
| 834,445,296
|
MDU6SXNzdWU4MzQ0NDUyOTY=
| 2,076
|
Issue: Dataset download error
|
{
"login": "XuhuiZhou",
"id": 20436061,
"node_id": "MDQ6VXNlcjIwNDM2MDYx",
"avatar_url": "https://avatars.githubusercontent.com/u/20436061?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/XuhuiZhou",
"html_url": "https://github.com/XuhuiZhou",
"followers_url": "https://api.github.com/users/XuhuiZhou/followers",
"following_url": "https://api.github.com/users/XuhuiZhou/following{/other_user}",
"gists_url": "https://api.github.com/users/XuhuiZhou/gists{/gist_id}",
"starred_url": "https://api.github.com/users/XuhuiZhou/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/XuhuiZhou/subscriptions",
"organizations_url": "https://api.github.com/users/XuhuiZhou/orgs",
"repos_url": "https://api.github.com/users/XuhuiZhou/repos",
"events_url": "https://api.github.com/users/XuhuiZhou/events{/privacy}",
"received_events_url": "https://api.github.com/users/XuhuiZhou/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067388877,
"node_id": "MDU6TGFiZWwyMDY3Mzg4ODc3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20bug",
"name": "dataset bug",
"color": "2edb81",
"default": false,
"description": "A bug in a dataset script provided in the library"
}
] |
open
| false
| null |
[] | null |
[
"Hi @XuhuiZhou, thanks for reporting this issue. \r\n\r\nIndeed, the old links are no longer valid (404 Not Found error), and the script must be updated with the new links to Google Drive.",
"It would be nice to update the urls indeed !\r\n\r\nTo do this, you just need to replace the urls in `iwslt2017.py` and then update the dataset_infos.json file with\r\n```\r\ndatasets-cli test ./datasets/iwslt2017 --all_configs --save_infos --ignore_verifications\r\n```",
"Is this a command to update my local files or fix the file Github repo in general? (I am not so familiar with the datasets-cli command here)\r\n\r\nI also took a brief look at the **Sharing your dataset** section, looks like I could fix that locally and push it to the repo? I guess we are \"canonical\" category?",
"This command will update your local file. Then you can open a Pull Request to push your fix to the github repo :)\r\nAnd yes you are right, it is a \"canonical\" dataset, i.e. a dataset script defined in this github repo (as opposed to dataset repositories of users on the huggingface hub)",
"Hi, thanks for the answer. \r\n\r\nI gave a try to the problem today. But I encountered an upload error: \r\n\r\n```\r\ngit push -u origin fix_link_iwslt\r\nEnter passphrase for key '/home2/xuhuizh/.ssh/id_rsa': \r\nERROR: Permission to huggingface/datasets.git denied to XuhuiZhou.\r\nfatal: Could not read from remote repository.\r\n\r\nPlease make sure you have the correct access rights\r\nand the repository exists.\r\n```\r\n\r\nAny insight here? \r\n\r\nBy the way, when I run the datasets-cli command, it shows the following error, but does not seem to be the error coming from `iwslt.py`\r\n\r\n```\r\nTraceback (most recent call last):\r\n File \"/home2/xuhuizh/anaconda3/envs/UMT/bin/datasets-cli\", line 33, in <module>\r\n sys.exit(load_entry_point('datasets', 'console_scripts', 'datasets-cli')())\r\n File \"/home2/xuhuizh/projects/datasets/src/datasets/commands/datasets_cli.py\", line 35, in main\r\n service.run()\r\n File \"/home2/xuhuizh/projects/datasets/src/datasets/commands/test.py\", line 141, in run\r\n try_from_hf_gcs=False,\r\n File \"/home2/xuhuizh/projects/datasets/src/datasets/builder.py\", line 579, in download_and_prepare\r\n dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs\r\n File \"/home2/xuhuizh/projects/datasets/src/datasets/builder.py\", line 639, in _download_and_prepare\r\n self.info.download_checksums, dl_manager.get_recorded_sizes_checksums(), \"dataset source files\"\r\n File \"/home2/xuhuizh/projects/datasets/src/datasets/utils/info_utils.py\", line 32, in verify_checksums\r\n raise ExpectedMoreDownloadedFiles(str(set(expected_checksums) - set(recorded_checksums)))\r\ndatasets.utils.info_utils.ExpectedMoreDownloadedFiles: {'https://wit3.fbk.eu/archive/2017-01-trnmted//texts/DeEnItNlRo/DeEnItNlRo/DeEnItNlRo-DeEnItNlRo.tgz'}\r\n```",
"Hi ! To create a PR on this repo your must fork it and create a branch on your fork. See how to fork the repo [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md#start-by-preparing-your-environment).\r\nAnd to make the command work without the `ExpectedMoreDownloadedFiles` error, you just need to use the `--ignore_verifications` flag.",
"Hi @XuhuiZhou,\r\n\r\nAs @lhoestq has well explained, you need to fork HF's repository, create a feature branch in your fork, push your changes to it and then open a Pull Request to HF's upstream repository. This is so because at HuggingFace Datasets we follow a development model called \"Fork and Pull Model\". You can find more information here:\r\n- [Understanding the GitHub flow](https://guides.github.com/introduction/flow/)\r\n- [Forking Projects](https://guides.github.com/activities/forking/)\r\n\r\nAlternatively, if you find all these steps too complicated, you can use the GitHub official command line tool: [GitHub CLI](https://cli.github.com/). Once installed, in order to create a Pull Request, you only need to use this command:\r\n```shell\r\ngh pr create --web\r\n```\r\nThis utility will automatically create the fork, push your changes and open a Pull Request, under the hood."
] | 2021-03-18T06:36:06
| 2021-03-22T11:52:31
| null |
NONE
| null | null | null |
The download link in `iwslt2017.py` file does not seem to work anymore.
For example, `FileNotFoundError: Couldn't find file at https://wit3.fbk.eu/archive/2017-01-trnted/texts/zh/en/zh-en.tgz`
Would be nice if we could modify it script and use the new downloadable link?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/2076/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/2076/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1994
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1994/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1994/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1994/events
|
https://github.com/huggingface/datasets/issues/1994
| 822,871,238
|
MDU6SXNzdWU4MjI4NzEyMzg=
| 1,994
|
not being able to get wikipedia es language
|
{
"login": "dorost1234",
"id": 79165106,
"node_id": "MDQ6VXNlcjc5MTY1MTA2",
"avatar_url": "https://avatars.githubusercontent.com/u/79165106?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dorost1234",
"html_url": "https://github.com/dorost1234",
"followers_url": "https://api.github.com/users/dorost1234/followers",
"following_url": "https://api.github.com/users/dorost1234/following{/other_user}",
"gists_url": "https://api.github.com/users/dorost1234/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dorost1234/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dorost1234/subscriptions",
"organizations_url": "https://api.github.com/users/dorost1234/orgs",
"repos_url": "https://api.github.com/users/dorost1234/repos",
"events_url": "https://api.github.com/users/dorost1234/events{/privacy}",
"received_events_url": "https://api.github.com/users/dorost1234/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"@lhoestq I really appreciate if you could help me providiing processed datasets, I do not really have access to enough resources to run the apache-beam and need to run the codes on these datasets. Only en/de/fr currently works, but I need all the languages more or less. thanks ",
"Hi @dorost1234, I think I can help you a little. I’ve processed some Wikipedia datasets (Spanish inclusive) using the HF/datasets library during recent research.\r\n\r\n@lhoestq Could you help me to upload these preprocessed datasets to Huggingface's repositories? To be more precise, I've built datasets from the following languages using the 20201201 dumps: Spanish, Portuguese, Russian, French, Japanese, Chinese, and Turkish. Process these datasets have high costs that most of the community can't afford. I think these preprocessed datasets I have could be helpful for someone without access to high-resource machines to process Wikipedia's dumps like @dorost1234\r\n\r\n",
"Thank you so much @jonatasgrosman , I greatly appreciate your help with them. \r\nYes, I unfortunately does not have access to a good resource and need it for my\r\nresearch. I greatly appreciate @lhoestq your help with uploading the processed datasets in huggingface datasets. This would be really helpful for some users like me with not access to high-memory GPU resources.\r\n\r\nthank you both so much again.\r\n\r\nOn Sat, Mar 6, 2021 at 12:55 AM Jonatas Grosman <notifications@github.com>\r\nwrote:\r\n\r\n> Hi @dorost1234 <https://github.com/dorost1234>, I think I can help you a\r\n> little. I’ve processed some Wikipedia datasets (Spanish inclusive) using\r\n> the HF/datasets library during recent research.\r\n>\r\n> @lhoestq <https://github.com/lhoestq> Could you help me to upload these\r\n> preprocessed datasets to Huggingface's repositories? To be more precise,\r\n> I've built datasets from the following languages using the 20201201 dumps:\r\n> Spanish, Portuguese, Russian, French, Japanese, Chinese, and Turkish.\r\n> Process these datasets have high costs that most of the community can't\r\n> afford. I think these preprocessed datasets I have could be helpful for\r\n> someone without access to high-resource machines to process Wikipedia's\r\n> dumps like @dorost1234 <https://github.com/dorost1234>\r\n>\r\n> —\r\n> You are receiving this because you were mentioned.\r\n> Reply to this email directly, view it on GitHub\r\n> <https://github.com/huggingface/datasets/issues/1994#issuecomment-791798195>,\r\n> or unsubscribe\r\n> <https://github.com/notifications/unsubscribe-auth/AS37NMWMK5GFJFU3ACCJFUDTCFVNZANCNFSM4YUZIF4A>\r\n> .\r\n>\r\n",
"Hi @dorost1234, so sorry, but looking at my files here, I figure out that I've preprocessed files using the HF/datasets for all the languages previously listed by me (Portuguese, Russian, French, Japanese, Chinese, and Turkish) except the Spanish (on my tests I've used the [wikicorpus](https://www.cs.upc.edu/~nlp/wikicorpus/) instead).\r\n\r\nOnly with the Spanish Wikipedia's dump, I had the same `KeyError: '000nbsp'` problem already reported here https://github.com/huggingface/datasets/issues/577\r\n\r\nSo nowadays, even with access to a high resource machine, you couldn't be able to get Wikipedia's Spanish data using the HF/datasets :(\r\n\r\n\r\n\r\n\r\n",
"Thanks a lot for the information and help. This would be great to have\nthese datasets.\n@lhoestq <https://github.com/lhoestq> Do you know a way I could get\nsmaller amount of these data like 1 GBtype of each language to deal with\ncomputatioanl requirements? thanks\n\nOn Sat, Mar 6, 2021 at 5:36 PM Jonatas Grosman <notifications@github.com>\nwrote:\n\n> Hi @dorost1234 <https://github.com/dorost1234>, so sorry, but looking at\n> my files here, I figure out that I've preprocessed files using the\n> HF/datasets for all the languages previously listed by me (Portuguese,\n> Russian, French, Japanese, Chinese, and Turkish) except the Spanish (on my\n> tests I've used the wikicorpus <https://www.cs.upc.edu/~nlp/wikicorpus/>\n> instead).\n>\n> Only with the Spanish Wikipedia's dump, I had the same KeyError: '000nbsp'\n> problem already reported here #577\n> <https://github.com/huggingface/datasets/issues/577>\n>\n> So nowadays, even with access to a high resource machine, you couldn't be\n> able to get Wikipedia's Spanish data using the HF/datasets :(\n>\n> —\n> You are receiving this because you were mentioned.\n> Reply to this email directly, view it on GitHub\n> <https://github.com/huggingface/datasets/issues/1994#issuecomment-791985546>,\n> or unsubscribe\n> <https://github.com/notifications/unsubscribe-auth/AS37NMWMO7WOHWLOROPD6Q3TCJKXPANCNFSM4YUZIF4A>\n> .\n>\n",
"Hi ! As mentioned above the Spanish configuration have parsing issues from `mwparserfromhell`. I haven't tested with the latest `mwparserfromhell` >=0.6 though. Which version of `mwparserfromhell` are you using ?\r\n\r\n> @lhoestq Could you help me to upload these preprocessed datasets to Huggingface's repositories? To be more precise, I've built datasets from the following languages using the 20201201 dumps: Spanish, Portuguese, Russian, French, Japanese, Chinese, and Turkish. Process these datasets have high costs that most of the community can't afford. I think these preprocessed datasets I have could be helpful for someone without access to high-resource machines to process Wikipedia's dumps like @dorost1234\r\n\r\nThat would be awesome ! Feel free to ping me on slack so we can put the processed wikipedia files on google storage with the other ones we've already preprocessed.\r\n\r\n> Do you know a way I could get smaller amount of these data like 1 GBtype of each language to deal with computatioanl requirements? thanks\r\n\r\nI'd suggest to copy the [wikipedia.py](https://github.com/huggingface/datasets/blob/master/datasets/wikipedia/wikipedia.py) to a new script `custom_wikipedia.py` and modify it to only download and process only a subset of the raw data files.\r\nYou can for example replace [this line](https://github.com/huggingface/datasets/blob/64e59fc45ca2134218b3e42e83fddddbe840ff74/datasets/wikipedia/wikipedia.py#L446) by:\r\n```python\r\n if total_bytes >= (1 << 30): # stop if the total amount of data is >= 1GB\r\n break\r\n else:\r\n xml_urls.append(_base_url(lang) + fname)\r\n```\r\n\r\nThen you can load your custom wikipedia dataset with\r\n```python\r\nload_dataset(\"path/to/my/custom_wikipedia.py\", f\"{date}.{language}\")\r\n```",
"Hi @lhoestq!\r\n\r\n> Hi ! As mentioned above the Spanish configuration have parsing issues from mwparserfromhell. I haven't tested with the latest mwparserfromhell >=0.6 though. Which version of mwparserfromhell are you using ?\r\n\r\nI'm using the latest mwparserfromhell version (0.6)\r\n\r\n> That would be awesome ! Feel free to ping me on slack so we can put the processed wikipedia files on google storage with the other ones we've already preprocessed.\r\n\r\nI'll ping you there 👍 ",
"Thank you so much @jonatasgrosman and @lhoestq this would be a great help. I am really thankful to you both and to wonderful Huggingface dataset library allowing us to train models at scale."
] | 2021-03-05T08:31:48
| 2021-03-11T20:46:21
| null |
NONE
| null | null | null |
Hi
I am trying to run a code with wikipedia of config 20200501.es, getting:
Traceback (most recent call last):
File "run_mlm_t5.py", line 608, in <module>
main()
File "run_mlm_t5.py", line 359, in main
datasets = load_dataset(data_args.dataset_name, data_args.dataset_config_name)
File "/dara/libs/anaconda3/envs/success432/lib/python3.7/site-packages/datasets-1.2.1-py3.7.egg/datasets/load.py", line 612, in load_dataset
ignore_verifications=ignore_verifications,
File "/dara/libs/anaconda3/envs/success432/lib/python3.7/site-packages/datasets-1.2.1-py3.7.egg/datasets/builder.py", line 527, in download_and_prepare
dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
File "/dara/libs/anaconda3/envs/success432/lib/python3.7/site-packages/datasets-1.2.1-py3.7.egg/datasets/builder.py", line 1050, in _download_and_prepare
"\n\t`{}`".format(usage_example)
datasets.builder.MissingBeamOptions: Trying to generate a dataset using Apache Beam, yet no Beam Runner or PipelineOptions() has been provided in `load_dataset` or in the builder arguments. For big datasets it has to run on large-scale data processing tools like Dataflow, Spark, etc. More information about Apache Beam runners at https://beam.apache.org/documentation/runners/capability-matrix/
If you really want to run it locally because you feel like the Dataset is small enough, you can use the local beam runner called `DirectRunner` (you may run out of memory).
Example of usage:
`load_dataset('wikipedia', '20200501.es', beam_runner='DirectRunner')`
thanks @lhoestq for any suggestion/help
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1994/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1994/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1992
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1992/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1992/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1992/events
|
https://github.com/huggingface/datasets/issues/1992
| 822,672,238
|
MDU6SXNzdWU4MjI2NzIyMzg=
| 1,992
|
`datasets.map` multi processing much slower than single processing
|
{
"login": "hwijeen",
"id": 29157715,
"node_id": "MDQ6VXNlcjI5MTU3NzE1",
"avatar_url": "https://avatars.githubusercontent.com/u/29157715?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/hwijeen",
"html_url": "https://github.com/hwijeen",
"followers_url": "https://api.github.com/users/hwijeen/followers",
"following_url": "https://api.github.com/users/hwijeen/following{/other_user}",
"gists_url": "https://api.github.com/users/hwijeen/gists{/gist_id}",
"starred_url": "https://api.github.com/users/hwijeen/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/hwijeen/subscriptions",
"organizations_url": "https://api.github.com/users/hwijeen/orgs",
"repos_url": "https://api.github.com/users/hwijeen/repos",
"events_url": "https://api.github.com/users/hwijeen/events{/privacy}",
"received_events_url": "https://api.github.com/users/hwijeen/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Hi @hwijeen, you might want to look at issues #1796 and #1949. I think it could be something related to the I/O operations being performed.",
"I see that many people are experiencing the same issue. Is this problem considered an \"official\" bug that is worth a closer look? @lhoestq",
"Yes this looks like a bug. On my side I haven't managed to reproduce it but @theo-m has. We'll investigate this !",
"Thank you for the reply! I would be happy to follow the discussions related to the issue.\r\nIf you do not mind, could you also give a little more explanation on my p.s.2? I am having a hard time figuring out why the single processing `map` uses all of my cores.\r\n@lhoestq @theo-m ",
"Regarding your ps2: It depends what function you pass to `map`.\r\nFor example, fast tokenizers from `transformers` in Rust tokenize texts and parallelize the tokenization over all the cores.",
"I am still experiencing this issue with datasets 1.9.0..\r\nHas there been a further investigation? \r\n<img width=\"442\" alt=\"image\" src=\"https://user-images.githubusercontent.com/29157715/126143387-8b5ddca2-a896-4e18-abf7-4fbf62a48b41.png\">\r\n",
"Hi. Is there any update on this issue? I am desperately trying to decrease my times, and multiprocessing \"should\" be the solution, but it literally takes 5 times longer.",
"Which version of `datasets` are you using ?",
"Hi,\r\n\r\nI’m running into the same issue and trying to come up with a simple benchmark. \r\n\r\n# environment info\r\nI have a total of 80 CPUs.\r\n\r\n- `datasets` version: 2.4.0\r\n- Platform: Linux-4.18.0-305.65.1.el8_4.x86_64-x86_64-with-glibc2.28\r\n- Python version: 3.10.4\r\n- PyArrow version: 8.0.0\r\n- Pandas version: 1.4.3\r\n\r\n# How to reproduce\r\n\r\n```py\r\nIn [1]: from datasets import Dataset, set_caching_enabled \r\nIn [2]: import numpy as np \r\nIn [3]: set_caching_enabled(False) \r\nIn [4]: d = Dataset.from_dict({'foo': np.random.randn(1000,256)}) \r\nIn [9]: d.set_format('np')\r\nIn [14]: def sort(array): \r\n ...: np.sort(array) \r\n# multiprocessing disabled\r\nIn [19]: %%timeit \r\n ...: d.map(sort, input_columns='foo') \r\n78.8 ms ± 1.22 ms per loop (mean ± std. dev. of 7 runs, 10 loops each) \r\n# multiprocessing enabled \r\nIn [27]: %%timeit \r\n ...: d.map(sort, input_columns='foo',num_proc=10) \r\n858 ms ± 45.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each) \r\n```",
"Spawning multiple processes has an overhead. For small datasets the processing is likely to be faster than spawning the processes and passing the data to them.\r\n\r\nEspecially since your dataset is in memory: the data has to be copied to the subprocesses.\r\nOn the other hand, datasets loaded from disk are much faster to reload from a subprocess thanks to memory mapping.",
"Thanks for the clarifications! \r\n\r\nIndeed, when saving then loading the above dataset to disk, and increasing the number of rows to 10K or 100K, the performance gap narrows.\r\n\r\n```py\r\n# with 10000 rows\r\nIn [3]: %%timeit\r\n ...: d.map(sort, input_columns='foo')\r\n578 ms ± 5.89 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\r\nIn [4]: %%timeit \r\n ...: d.map(sort, input_columns='foo',num_proc=10) \r\n1.06 s ± 47.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\r\n\r\n# with 100000 rows\r\nIn [6]: %%timeit\r\n ...: d.map(sort, input_columns='foo')\r\n5.8 s ± 25.5 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\r\nIn [7]: %%timeit\r\n ...: d.map(sort, input_columns='foo',num_proc=10)\r\n7.23 s ± 154 ms per loop (mean ± std. dev. of 7 runs, 1 loop each\r\n```",
"any updates on this issue? \r\nI'm using `datasets=2.12.0`. Adding `num_proc` to the mapping function makes it at least 5x slower than using a single process.",
"What kind of function are you passing to `map` ? How many CPUs do you have and what did you set for `num_proc` ?"
] | 2021-03-05T02:10:02
| 2023-06-08T12:31:55
| null |
NONE
| null | null | null |
Hi, thank you for the great library.
I've been using datasets to pretrain language models, and it often involves datasets as large as ~70G.
My data preparation step is roughly two steps: `load_dataset` which splits corpora into a table of sentences, and `map` converts a sentence into a list of integers, using a tokenizer.
I noticed that `map` function with `num_proc=mp.cpu_count() //2` takes more than 20 hours to finish the job where as `num_proc=1` gets the job done in about 5 hours. The machine I used has 40 cores, with 126G of RAM. There were no other jobs when `map` function was running.
What could be the reason? I would be happy to provide information necessary to spot the reason.
p.s. I was experiencing the imbalance issue mentioned in [here](https://github.com/huggingface/datasets/issues/610#issuecomment-705177036) when I was using multi processing.
p.s.2 When I run `map` with `num_proc=1`, I see one tqdm bar but all the cores are working. When `num_proc=20`, only 20 cores work.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1992/reactions",
"total_count": 5,
"+1": 5,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1992/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1977
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1977/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1977/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1977/events
|
https://github.com/huggingface/datasets/issues/1977
| 820,312,022
|
MDU6SXNzdWU4MjAzMTIwMjI=
| 1,977
|
ModuleNotFoundError: No module named 'apache_beam' for wikipedia datasets
|
{
"login": "dorost1234",
"id": 79165106,
"node_id": "MDQ6VXNlcjc5MTY1MTA2",
"avatar_url": "https://avatars.githubusercontent.com/u/79165106?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dorost1234",
"html_url": "https://github.com/dorost1234",
"followers_url": "https://api.github.com/users/dorost1234/followers",
"following_url": "https://api.github.com/users/dorost1234/following{/other_user}",
"gists_url": "https://api.github.com/users/dorost1234/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dorost1234/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dorost1234/subscriptions",
"organizations_url": "https://api.github.com/users/dorost1234/orgs",
"repos_url": "https://api.github.com/users/dorost1234/repos",
"events_url": "https://api.github.com/users/dorost1234/events{/privacy}",
"received_events_url": "https://api.github.com/users/dorost1234/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"I sometimes also get this error with other languages of the same dataset:\r\n\r\n File \"/dara/libs/anaconda3/envs/code/lib/python3.7/site-packages/datasets-1.3.0-py3.7.egg/datasets/arrow_reader.py\", line 322, in read_table\r\n stream = stream_from(filename)\r\n File \"pyarrow/io.pxi\", line 782, in pyarrow.lib.memory_map\r\n File \"pyarrow/io.pxi\", line 743, in pyarrow.lib.MemoryMappedFile._open\r\n File \"pyarrow/error.pxi\", line 122, in pyarrow.lib.pyarrow_internal_check_status\r\n File \"pyarrow/error.pxi\", line 99, in pyarrow.lib.check_status\r\nOSError: Memory mapping file failed: Cannot allocate memory\r\n\r\n@lhoestq \r\n",
"Hi ! Thanks for reporting\r\nSome wikipedia configurations do require the user to have `apache_beam` in order to parse the wikimedia data.\r\n\r\nOn the other hand regarding your second issue\r\n```\r\nOSError: Memory mapping file failed: Cannot allocate memory\r\n```\r\nI've never experienced this, can you open a new issue for this specific error and provide more details please ?\r\nFor example what script did you use to get this, what language did you use, what's your environment details (os, python version, pyarrow version).."
] | 2021-03-02T19:21:28
| 2021-03-03T10:17:40
| null |
NONE
| null | null | null |
Hi
I am trying to run run_mlm.py code [1] of huggingface with following "wikipedia"/ "20200501.aa" dataset:
`python run_mlm.py --model_name_or_path bert-base-multilingual-cased --dataset_name wikipedia --dataset_config_name 20200501.aa --do_train --do_eval --output_dir /tmp/test-mlm --max_seq_length 256
`
I am getting this error, but as per documentation, huggingface dataset provide processed version of this dataset and users can load it without requiring setup extra settings for apache-beam. could you help me please to load this dataset?
Do you think I can run run_ml.py with this dataset? or anyway I could subsample and train the model? I greatly appreciate providing the processed version of all languages for this dataset, which allow the user to use them without setting up apache-beam,. thanks
I really appreciate your help.
@lhoestq
thanks.
[1] https://github.com/huggingface/transformers/blob/master/examples/language-modeling/run_mlm.py
error I get:
```
>>> import datasets
>>> datasets.load_dataset("wikipedia", "20200501.aa")
Downloading and preparing dataset wikipedia/20200501.aa (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /dara/temp/cache_home_2/datasets/wikipedia/20200501.aa/1.0.0/4021357e28509391eab2f8300d9b689e7e8f3a877ebb3d354b01577d497ebc63...
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/dara/temp/libs/anaconda3/envs/codes/lib/python3.7/site-packages/datasets-1.3.0-py3.7.egg/datasets/load.py", line 746, in load_dataset
use_auth_token=use_auth_token,
File "/dara/temp/libs/anaconda3/envs/codes/lib/python3.7/site-packages/datasets-1.3.0-py3.7.egg/datasets/builder.py", line 573, in download_and_prepare
dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
File "/dara/temp/libs/anaconda3/envs/codes/lib/python3.7/site-packages/datasets-1.3.0-py3.7.egg/datasets/builder.py", line 1099, in _download_and_prepare
import apache_beam as beam
ModuleNotFoundError: No module named 'apache_beam'
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1977/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1977/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1949
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1949/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1949/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1949/events
|
https://github.com/huggingface/datasets/issues/1949
| 816,986,936
|
MDU6SXNzdWU4MTY5ODY5MzY=
| 1,949
|
Enable Fast Filtering using Arrow Dataset
|
{
"login": "gchhablani",
"id": 29076344,
"node_id": "MDQ6VXNlcjI5MDc2MzQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/29076344?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gchhablani",
"html_url": "https://github.com/gchhablani",
"followers_url": "https://api.github.com/users/gchhablani/followers",
"following_url": "https://api.github.com/users/gchhablani/following{/other_user}",
"gists_url": "https://api.github.com/users/gchhablani/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gchhablani/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gchhablani/subscriptions",
"organizations_url": "https://api.github.com/users/gchhablani/orgs",
"repos_url": "https://api.github.com/users/gchhablani/repos",
"events_url": "https://api.github.com/users/gchhablani/events{/privacy}",
"received_events_url": "https://api.github.com/users/gchhablani/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi @gchhablani :)\r\nThanks for proposing your help !\r\n\r\nI'll be doing a refactor of some parts related to filtering in the scope of https://github.com/huggingface/datasets/issues/1877\r\nSo I would first wait for this refactor to be done before working on the filtering. In particular because I plan to make things simpler to manipulate.\r\n\r\nYour feedback on this refactor would also be appreciated since it also aims at making the core code more accessible (basically my goal is that no one's ever \"having troubles getting started\" ^^)\r\n\r\nThis will be available in a few days, I will be able to give you more details at that time if you don't mind waiting a bit !",
"Sure! I don't mind waiting. I'll check the refactor and try to understand what you're trying to do :)"
] | 2021-02-26T02:53:37
| 2021-02-26T19:18:29
| null |
CONTRIBUTOR
| null | null | null |
Hi @lhoestq,
As mentioned in Issue #1796, I would love to work on enabling fast filtering/mapping. Can you please share the expectations? It would be great if you could point me to the relevant methods/files involved. Or the docs or maybe an overview of `arrow_dataset.py`. I only ask this because I am having trouble getting started ;-;
Any help would be appreciated.
Thanks,
Gunjan
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1949/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1949/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1922
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1922/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1922/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1922/events
|
https://github.com/huggingface/datasets/issues/1922
| 813,140,806
|
MDU6SXNzdWU4MTMxNDA4MDY=
| 1,922
|
How to update the "wino_bias" dataset
|
{
"login": "JieyuZhao",
"id": 22306304,
"node_id": "MDQ6VXNlcjIyMzA2MzA0",
"avatar_url": "https://avatars.githubusercontent.com/u/22306304?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/JieyuZhao",
"html_url": "https://github.com/JieyuZhao",
"followers_url": "https://api.github.com/users/JieyuZhao/followers",
"following_url": "https://api.github.com/users/JieyuZhao/following{/other_user}",
"gists_url": "https://api.github.com/users/JieyuZhao/gists{/gist_id}",
"starred_url": "https://api.github.com/users/JieyuZhao/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/JieyuZhao/subscriptions",
"organizations_url": "https://api.github.com/users/JieyuZhao/orgs",
"repos_url": "https://api.github.com/users/JieyuZhao/repos",
"events_url": "https://api.github.com/users/JieyuZhao/events{/privacy}",
"received_events_url": "https://api.github.com/users/JieyuZhao/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi @JieyuZhao !\r\n\r\nYou can edit the dataset card of wino_bias to update the URL via a Pull Request. This would be really appreciated :)\r\n\r\nThe dataset card is the README.md file you can find at https://github.com/huggingface/datasets/tree/master/datasets/wino_bias\r\nAlso the homepage url is also mentioned in the wino_bias.py so feel free to update it there as well.\r\n\r\nYou can create a Pull Request directly from the github interface by editing the files you want and submit a PR, or from a local clone of the repository.\r\n\r\nThanks for noticing !"
] | 2021-02-22T05:39:39
| 2021-02-22T10:35:59
| null |
CONTRIBUTOR
| null | null | null |
Hi all,
Thanks for the efforts to collect all the datasets! But I think there is a problem with the wino_bias dataset. The current link is not correct. How can I update that?
Thanks!
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1922/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1922/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1911
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1911/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1911/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1911/events
|
https://github.com/huggingface/datasets/issues/1911
| 812,009,956
|
MDU6SXNzdWU4MTIwMDk5NTY=
| 1,911
|
Saving processed dataset running infinitely
|
{
"login": "ayubSubhaniya",
"id": 20911334,
"node_id": "MDQ6VXNlcjIwOTExMzM0",
"avatar_url": "https://avatars.githubusercontent.com/u/20911334?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ayubSubhaniya",
"html_url": "https://github.com/ayubSubhaniya",
"followers_url": "https://api.github.com/users/ayubSubhaniya/followers",
"following_url": "https://api.github.com/users/ayubSubhaniya/following{/other_user}",
"gists_url": "https://api.github.com/users/ayubSubhaniya/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ayubSubhaniya/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ayubSubhaniya/subscriptions",
"organizations_url": "https://api.github.com/users/ayubSubhaniya/orgs",
"repos_url": "https://api.github.com/users/ayubSubhaniya/repos",
"events_url": "https://api.github.com/users/ayubSubhaniya/events{/privacy}",
"received_events_url": "https://api.github.com/users/ayubSubhaniya/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"@thomwolf @lhoestq can you guys please take a look and recommend some solution.",
"am suspicious of this thing? what's the purpose of this? pickling and unplickling\r\n`self = pickle.loads(pickle.dumps(self))`\r\n\r\n```\r\n def save_to_disk(self, dataset_path: str, fs=None):\r\n \"\"\"\r\n Saves a dataset to a dataset directory, or in a filesystem using either :class:`datasets.filesystem.S3FileSystem` or any implementation of ``fsspec.spec.AbstractFileSystem``.\r\n\r\n Args:\r\n dataset_path (``str``): path (e.g. ``dataset/train``) or remote uri (e.g. ``s3://my-bucket/dataset/train``) of the dataset directory where the dataset will be saved to\r\n fs (Optional[:class:`datasets.filesystem.S3FileSystem`,``fsspec.spec.AbstractFileSystem``], `optional`, defaults ``None``): instance of :class:`datasets.filesystem.S3FileSystem` or ``fsspec.spec.AbstractFileSystem`` used to download the files from remote filesystem.\r\n \"\"\"\r\n assert (\r\n not self.list_indexes()\r\n ), \"please remove all the indexes using `dataset.drop_index` before saving a dataset\"\r\n self = pickle.loads(pickle.dumps(self))\r\n ```",
"It's been 24 hours and sadly it's still running. With not a single byte written",
"Tried finding the root cause but was unsuccessful.\r\nI am using lazy tokenization with `dataset.set_transform()`, it works like a charm with almost same performance as pre-compute.",
"Hi ! This very probably comes from the hack you used.\r\n\r\nThe pickling line was added an a sanity check because save_to_disk uses the same assumptions as pickling for a dataset object. The main assumption is that memory mapped pyarrow tables must be reloadable from the disk. In your case it's not possible since you altered the pyarrow table.\r\nI would suggest you to rebuild a valid Dataset object from your new pyarrow table. To do so you must first save your new table to a file, and then make a new Dataset object from that arrow file.\r\n\r\nYou can save the raw arrow table (without all the `datasets.Datasets` metadata) by calling `map` with `cache_file_name=\"path/to/outut.arrow\"` and `function=None`. Having `function=None` makes the `map` write your dataset on disk with no data transformation.\r\n\r\nOnce you have your new arrow file, load it with `datasets.Dataset.from_file` to have a brand new Dataset object :)\r\n\r\nIn the future we'll have a better support for the fast filtering method from pyarrow so you don't have to do this very unpractical workaround. Since it breaks somes assumptions regarding the core behavior of Dataset objects, this is very discouraged.",
"Thanks, @lhoestq for your response. Will try your solution and let you know."
] | 2021-02-19T13:09:19
| 2021-02-23T07:34:44
| null |
NONE
| null | null | null |
I have a text dataset of size 220M.
For pre-processing, I need to tokenize this and filter rows with the large sequence.
My tokenization took roughly 3hrs. I used map() with batch size 1024 and multi-process with 96 processes.
filter() function was way to slow, so I used a hack to use pyarrow filter table function, which is damm fast. Mentioned [here](https://github.com/huggingface/datasets/issues/1796)
```dataset._data = dataset._data.filter(...)```
It took 1 hr for the filter.
Then i use `save_to_disk()` on processed dataset and it is running forever.
I have been waiting since 8 hrs, it has not written a single byte.
Infact it has actually read from disk more than 100GB, screenshot below shows the stats using `iotop`.
Second process is the one.
<img width="1672" alt="Screenshot 2021-02-19 at 6 36 53 PM" src="https://user-images.githubusercontent.com/20911334/108508197-7325d780-72e1-11eb-8369-7c057d137d81.png">
I am not able to figure out, whether this is some issue with dataset library or that it is due to my hack for filter() function.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1911/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1911/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1906
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1906/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1906/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1906/events
|
https://github.com/huggingface/datasets/issues/1906
| 811,405,274
|
MDU6SXNzdWU4MTE0MDUyNzQ=
| 1,906
|
Feature Request: Support for Pandas `Categorical`
|
{
"login": "justin-yan",
"id": 7731709,
"node_id": "MDQ6VXNlcjc3MzE3MDk=",
"avatar_url": "https://avatars.githubusercontent.com/u/7731709?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/justin-yan",
"html_url": "https://github.com/justin-yan",
"followers_url": "https://api.github.com/users/justin-yan/followers",
"following_url": "https://api.github.com/users/justin-yan/following{/other_user}",
"gists_url": "https://api.github.com/users/justin-yan/gists{/gist_id}",
"starred_url": "https://api.github.com/users/justin-yan/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/justin-yan/subscriptions",
"organizations_url": "https://api.github.com/users/justin-yan/orgs",
"repos_url": "https://api.github.com/users/justin-yan/repos",
"events_url": "https://api.github.com/users/justin-yan/events{/privacy}",
"received_events_url": "https://api.github.com/users/justin-yan/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 2067400324,
"node_id": "MDU6TGFiZWwyMDY3NDAwMzI0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/generic%20discussion",
"name": "generic discussion",
"color": "c5def5",
"default": false,
"description": "Generic discussion on the library"
}
] |
open
| false
| null |
[] | null |
[
"We already have a ClassLabel type that does this kind of mapping between the label ids (integers) and actual label values (strings).\r\n\r\nI wonder if actually we should use the DictionaryType from Arrow and the Categorical type from pandas for the `datasets` ClassLabel feature type.\r\nCurrently ClassLabel corresponds to `pa.int64()` in pyarrow and `dtype('int64')` in pandas (so the label names are lost during conversions).\r\n\r\nWhat do you think ?",
"Now that I've heard you explain ClassLabel, that makes a lot of sense! While DictionaryType for Arrow (I think) can have arbitrarily typed keys, so it won't cover all potential cases, pandas' Category is *probably* the most common use for that pyarrow type, and ClassLabel should match that perfectly?\r\n\r\nOther thoughts:\r\n\r\n- changing the resulting patype on ClassLabel might be backward-incompatible? I'm not totally sure if users of the `datasets` library tend to directly access the `patype` attribute (I don't think we really do, but we haven't been using it for very long yet).\r\n- would ClassLabel's dtype change to `dict[int64, string]`? It seems like in practice a ClassLabel (when not explicitly specified) would be constructed from the DictionaryType branch of `generate_from_arrow_type`, so it's not totally clear to me that anyone ever actually accesses/uses that dtype?\r\n- I don't quite know how `.int2str` and `.str2int` are used in practice - would those be kept? Perhaps the implementation might actually be substantially smaller if we can just delegate to pyarrow's dict methods?\r\n\r\nAnother idea that just occurred to me: add a branch in here to generate a ClassLabel if the dict key is int64 and the values are string: https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L932 , and then don't touch anything else.\r\n\r\nIn practice, I don't think this would be backward-incompatible in a way anyone would care about since the current behavior just throws an exception, and this way, we could support *reading* a pandas Categorical into a `Dataset` as a ClassLabel. I *think* from there, while it would require some custom glue it wouldn't be too hard to convert the ClassLabel into a pandas Category if we want to go back - I think this would improve on the current behavior without risking changing the behavior of ClassLabel in a backward-incompat way.\r\n\r\nThoughts? I'm not sure if this is overly cautious. Whichever approach you think is better, I'd be happy to take it on!\r\n",
"I think we can first keep the int64 precision but with an arrow Dictionary for ClassLabel, and focus on the connection with arrow and pandas.\r\n\r\nIn this scope, I really like the idea of checking for the dictionary type:\r\n\r\n> Another idea that just occurred to me: add a branch in here to generate a ClassLabel if the dict key is int64 and the values are string: https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L932 , and then don't touch anything else.\r\n\r\nThis looks like a great start.\r\n\r\nThen as you said we'd have to add the conversion from classlabel to the correct arrow dictionary type. Arrow is already able to convert from arrow Dictionary to pandas Categorical so it should be enough.\r\n\r\nI can see two things that we must take case of to make this change backward compatible:\r\n- first we must still be able to load an arrow file with arrow int64 dtype and `datasets` ClassLabel type without crashing. This can be fixed by casting the arrow int64 array to an arrow Dictionary array on-the-fly when loading the table in the ArrowReader.\r\n- then we still have to return integers when accessing examples from a ClassLabel column. Currently it would return the strings values since it's based on the pandas behavior for converting from pandas to python/numpy. To do so we just have to adapt the python/numpy extractors in formatting.py (it takes care of converting an arrow table to a dictionary of python objects by doing arrow table -> pandas dataframe -> python dictionary)\r\n\r\nAny help on this matter is very much welcome :)"
] | 2021-02-18T19:46:05
| 2021-02-23T14:38:50
| null |
CONTRIBUTOR
| null | null | null |
```
from datasets import Dataset
import pandas as pd
import pyarrow
df = pd.DataFrame(pd.Series(["a", "b", "c", "a"], dtype="category"))
pyarrow.Table.from_pandas(df)
Dataset.from_pandas(df)
# Throws NotImplementedError
# TODO(thom) this will need access to the dictionary as well (for labels). I.e. to the py_table
```
I'm curious if https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L796 could be built out in a way similar to `Sequence`?
e.g. a `Map` class (or whatever name the maintainers might prefer) that can accept:
```
index_type = generate_from_arrow_type(pa_type.index_type)
value_type = generate_from_arrow_type(pa_type.value_type)
```
and then additional code points to modify:
- FeatureType: https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L694
- A branch to handle Map in get_nested_type: https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L719
- I don't quite understand what `encode_nested_example` does but perhaps a branch there? https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L755
- Similarly, I don't quite understand why `Sequence` is used this way in `generate_from_dict`, but perhaps a branch here? https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L775
I couldn't find other usages of `Sequence` outside of defining specific datasets, so I'm not sure if that's a comprehensive set of touchpoints.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1906/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1906/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1894
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1894/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1894/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1894/events
|
https://github.com/huggingface/datasets/issues/1894
| 809,609,654
|
MDU6SXNzdWU4MDk2MDk2NTQ=
| 1,894
|
benchmarking against MMapIndexedDataset
|
{
"login": "sshleifer",
"id": 6045025,
"node_id": "MDQ6VXNlcjYwNDUwMjU=",
"avatar_url": "https://avatars.githubusercontent.com/u/6045025?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sshleifer",
"html_url": "https://github.com/sshleifer",
"followers_url": "https://api.github.com/users/sshleifer/followers",
"following_url": "https://api.github.com/users/sshleifer/following{/other_user}",
"gists_url": "https://api.github.com/users/sshleifer/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sshleifer/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sshleifer/subscriptions",
"organizations_url": "https://api.github.com/users/sshleifer/orgs",
"repos_url": "https://api.github.com/users/sshleifer/repos",
"events_url": "https://api.github.com/users/sshleifer/events{/privacy}",
"received_events_url": "https://api.github.com/users/sshleifer/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi sam !\r\nIndeed we can expect the performances to be very close since both MMapIndexedDataset and the `datasets` implem use memory mapping. With memory mapping what determines the I/O performance is the speed of your hard drive/SSD.\r\n\r\nIn terms of performance we're pretty close to the optimal speed for reading text, even though I found recently that we could still slightly improve speed for big datasets (see [here](https://github.com/huggingface/datasets/issues/1803)).\r\n\r\nIn terms of number of examples and example sizes, the only limit is the available disk space you have.\r\n\r\nI haven't used `psrecord` yet but it seems to be a very interesting tool for benchmarking. Currently for benchmarks we only have github actions to avoid regressions in terms of speed. But it would be cool to have benchmarks with comparisons with other dataset tools ! This would be useful to many people",
"Also I would be interested to know what data types `MMapIndexedDataset` supports. Is there some documentation somewhere ?",
"no docs haha, it's written to support integer numpy arrays.\r\n\r\nYou can build one in fairseq with, roughly:\r\n```bash\r\n\r\nwget https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-103-raw-v1.zip\r\nunzip wikitext-103-raw-v1.zip\r\nexport dd=$HOME/fairseq-py/wikitext-103-raw\r\n\r\nexport mm_dir=$HOME/mmap_wikitext2\r\nmkdir -p gpt2_bpe\r\nwget -O gpt2_bpe/encoder.json https://dl.fbaipublicfiles.com/fairseq/gpt2_bpe/encoder.json\r\nwget -O gpt2_bpe/vocab.bpe https://dl.fbaipublicfiles.com/fairseq/gpt2_bpe/vocab.bpe\r\nwget -O gpt2_bpe/dict.txt https://dl.fbaipublicfiles.com/fairseq/gpt2_bpe/dict.txt\r\nfor SPLIT in train valid; do \\\r\n python -m examples.roberta.multiprocessing_bpe_encoder \\\r\n --encoder-json gpt2_bpe/encoder.json \\\r\n --vocab-bpe gpt2_bpe/vocab.bpe \\\r\n --inputs /scratch/stories_small/${SPLIT}.txt \\\r\n --outputs /scratch/stories_small/${SPLIT}.bpe \\\r\n --keep-empty \\\r\n --workers 60; \\\r\ndone\r\n\r\nmkdir -p $mm_dir\r\nfairseq-preprocess \\\r\n --only-source \\\r\n --srcdict gpt2_bpe/dict.txt \\\r\n --trainpref $dd/wiki.train.bpe \\\r\n --validpref $dd/wiki.valid.bpe \\\r\n --destdir $mm_dir \\\r\n --workers 60 \\\r\n --dataset-impl mmap\r\n```\r\n\r\nI'm noticing in my benchmarking that it's much smaller on disk than arrow (200mb vs 900mb), and that both incur significant cost by increasing the number of data loader workers. \r\nThis somewhat old [post](https://ray-project.github.io/2017/10/15/fast-python-serialization-with-ray-and-arrow.html) suggests there are some gains to be had from using `pyarrow.serialize(array).tobuffer()`. I haven't yet figured out how much of this stuff `pa.Table` does under the hood.\r\n\r\nThe `MMapIndexedDataset` bottlenecks we are working on improving (by using arrow) are:\r\n1) `MMapIndexedDataset`'s index, which stores offsets, basically gets read in its entirety by each dataloading process.\r\n2) we have separate, identical, `MMapIndexedDatasets` on each dataloading worker, so there's redundancy there; we wonder if there is a way that arrow can somehow dedupe these in shared memory.\r\n\r\nIt will take me a few hours to get `MMapIndexedDataset` benchmarks out of `fairseq`/onto a branch in this repo, but I'm happy to invest the time if you're interested in collaborating on some performance hacking."
] | 2021-02-16T20:04:58
| 2021-02-17T18:52:28
| null |
CONTRIBUTOR
| null | null | null |
I am trying to benchmark my datasets based implementation against fairseq's [`MMapIndexedDataset`](https://github.com/pytorch/fairseq/blob/master/fairseq/data/indexed_dataset.py#L365) and finding that, according to psrecord, my `datasets` implem uses about 3% more CPU memory and runs 1% slower for `wikitext103` (~1GB of tokens).
Questions:
1) Is this (basically identical) performance expected?
2) Is there a scenario where this library will outperform `MMapIndexedDataset`? (maybe more examples/larger examples?)
3) Should I be using different benchmarking tools than `psrecord`/how do you guys do benchmarks?
Thanks in advance! Sam
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1894/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1894/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1882
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1882/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1882/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1882/events
|
https://github.com/huggingface/datasets/pull/1882
| 808,716,576
|
MDExOlB1bGxSZXF1ZXN0NTczNzA4OTEw
| 1,882
|
Create Remote Manager
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"@lhoestq I have refactorized the logic. Instead of the previous hierarchy call (local temp file opening -> remote call -> use again temp local file logic but from within the remote caller scope), now it is flattened. Schematically:\r\n```python\r\nwith src.open() as src_file, dst.open() as dst_file:\r\n src_file.fetch(dst_file)\r\n```\r\n\r\nI have created `RemotePath` (analogue to Path) with method `.open()` that returns `FtpFile`/`HttpFile` (analogue to file-like).\r\n\r\nNow I am going to implement `RemotePath.exists()` method (analogue to the Path's method) to check if remote resource is accessible, using `Ftp/Http.head()`.",
"Quick update on this one:\r\nwe discussed offline with @albertvillanova on this PR and I think using `fsspec` can help a lot, since it already implements many parts of the abstraction we need to have nice download tools for both http and ftp (and others !)"
] | 2021-02-15T17:36:24
| 2022-07-06T15:19:47
| null |
MEMBER
| null | false
|
{
"url": "https://api.github.com/repos/huggingface/datasets/pulls/1882",
"html_url": "https://github.com/huggingface/datasets/pull/1882",
"diff_url": "https://github.com/huggingface/datasets/pull/1882.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/1882.patch",
"merged_at": null
}
|
Refactoring to separate the concern of remote (HTTP/FTP requests) management.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1882/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1882/timeline
| null | null | true
|
https://api.github.com/repos/huggingface/datasets/issues/1863
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1863/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1863/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1863/events
|
https://github.com/huggingface/datasets/issues/1863
| 806,171,311
|
MDU6SXNzdWU4MDYxNzEzMTE=
| 1,863
|
Add WikiCREM
|
{
"login": "NielsRogge",
"id": 48327001,
"node_id": "MDQ6VXNlcjQ4MzI3MDAx",
"avatar_url": "https://avatars.githubusercontent.com/u/48327001?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/NielsRogge",
"html_url": "https://github.com/NielsRogge",
"followers_url": "https://api.github.com/users/NielsRogge/followers",
"following_url": "https://api.github.com/users/NielsRogge/following{/other_user}",
"gists_url": "https://api.github.com/users/NielsRogge/gists{/gist_id}",
"starred_url": "https://api.github.com/users/NielsRogge/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/NielsRogge/subscriptions",
"organizations_url": "https://api.github.com/users/NielsRogge/orgs",
"repos_url": "https://api.github.com/users/NielsRogge/repos",
"events_url": "https://api.github.com/users/NielsRogge/events{/privacy}",
"received_events_url": "https://api.github.com/users/NielsRogge/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[
"Hi @NielsRogge I would like to work on this dataset.\r\n\r\nThanks!",
"Hi @udapy, are you working on this?"
] | 2021-02-11T08:16:00
| 2021-03-07T07:27:13
| null |
CONTRIBUTOR
| null | null | null |
## Adding a Dataset
- **Name:** WikiCREM
- **Description:** A large unsupervised corpus for coreference resolution.
- **Paper:** https://arxiv.org/abs/1905.06290
- **Github repo:**: https://github.com/vid-koci/bert-commonsense
- **Data:** https://ora.ox.ac.uk/objects/uuid:c83e94bb-7584-41a1-aef9-85b0e764d9e3
- **Motivation:** Coreference resolution, common sense reasoning
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1863/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1863/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1843
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1843/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1843/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1843/events
|
https://github.com/huggingface/datasets/issues/1843
| 803,565,393
|
MDU6SXNzdWU4MDM1NjUzOTM=
| 1,843
|
MustC Speech Translation
|
{
"login": "patrickvonplaten",
"id": 23423619,
"node_id": "MDQ6VXNlcjIzNDIzNjE5",
"avatar_url": "https://avatars.githubusercontent.com/u/23423619?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/patrickvonplaten",
"html_url": "https://github.com/patrickvonplaten",
"followers_url": "https://api.github.com/users/patrickvonplaten/followers",
"following_url": "https://api.github.com/users/patrickvonplaten/following{/other_user}",
"gists_url": "https://api.github.com/users/patrickvonplaten/gists{/gist_id}",
"starred_url": "https://api.github.com/users/patrickvonplaten/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/patrickvonplaten/subscriptions",
"organizations_url": "https://api.github.com/users/patrickvonplaten/orgs",
"repos_url": "https://api.github.com/users/patrickvonplaten/repos",
"events_url": "https://api.github.com/users/patrickvonplaten/events{/privacy}",
"received_events_url": "https://api.github.com/users/patrickvonplaten/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
},
{
"id": 2725241052,
"node_id": "MDU6TGFiZWwyNzI1MjQxMDUy",
"url": "https://api.github.com/repos/huggingface/datasets/labels/speech",
"name": "speech",
"color": "d93f0b",
"default": false,
"description": ""
}
] |
open
| false
| null |
[] | null |
[
"Hi @patrickvonplaten I would like to work on this dataset. \r\n\r\nThanks! ",
"That's awesome! Actually, I just noticed that this dataset might become a bit too big!\r\n\r\nMuST-C is the main dataset used for IWSLT19 and should probably be added as a standalone dataset. Would you be interested also in adding `datasets/MuST-C` instead?\r\n\r\nDescription: \r\n_MuST-C is a multilingual speech translation corpus whose size and quality facilitates the training of end-to-end systems for speech translation from English into several languages. For each target language, MuST-C comprises several hundred hours of audio recordings from English TED Talks, which are automatically aligned at the sentence level with their manual transcriptions and translations._\r\n\r\nPaper: https://www.aclweb.org/anthology/N19-1202.pdf\r\n\r\nDataset: https://ict.fbk.eu/must-c/ (One needs to fill out a short from to download the data, but it's very easy).\r\n\r\nIt would be awesome if you're interested in adding this datates. I'm very happy to guide you through the PR! I think the easiest way to start would probably be to read [this README on how to add a dataset](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md) and open a PR. Think you can copy & paste some code from:\r\n\r\n- Librispeech_asr: https://github.com/huggingface/datasets/blob/master/datasets/librispeech_asr/librispeech_asr.py\r\n- Flores Translation: https://github.com/huggingface/datasets/blob/master/datasets/flores/flores.py\r\n\r\nThink all the rest can be handled on the PR :-) ",
"Hi @patrickvonplaten \r\nI have tried downloading this dataset, but the connection seems to reset all the time. I have tried it via the browser, wget, and using gdown . But it gives me an error message. _\"The server is busy or down, pls try again\"_ (rephrasing the message here)\r\n\r\nI have completed adding 4 datasets in the previous data sprint (including the IWSLT dataset #1676 ) ...so just checking if you are able to download it at your end. Otherwise will write to the dataset authors to update the links. \r\n\r\n\r\n\r\n\r\n",
"Let me check tomorrow! Thanks for leaving this message!",
"cc @patil-suraj for notification ",
"@skyprince999, I think I'm getting the same error you're getting :-/\r\n\r\n```\r\nSorry, you can't view or download this file at this time.\r\n\r\nToo many users have viewed or downloaded this file recently. Please try accessing the file again later. If the file you are trying to access is particularly large or is shared with many people, it may take up to 24 hours to be able to view or download the file. If you still can't access a file after 24 hours, contact your domain administrator.\r\n```\r\n\r\nIt would be great if you could write the authors to see whether they can fix it.\r\nAlso cc @lhoestq - do you think we could mirror the dataset? ",
"Also there are huge those datasets. Think downloading MuST-C v1.2 amounts to ~ 1000GB... because there are 14 possible configs each around 60-70GB. I think users mostly will only use one of the 14 configs so that they would only need, in theory, will have to download ~60GB which is ok. But I think this functionality doesn't exist yet in `datasets` no? cc @lhoestq ",
"> Also cc @lhoestq - do you think we could mirror the dataset?\r\n\r\nYes we can mirror it if the authors are fine with it. You can create a dataset repo on huggingface.co (possibly under the relevant org) and add the mirrored data files.\r\n\r\n> I think users mostly will only use one of the 14 configs so that they would only need, in theory, will have to download ~60GB which is ok. But I think this functionality doesn't exist yet in datasets no? cc @lhoestq\r\n\r\nIf there are different download links for each configuration we can make the dataset builder download only the files related to the requested configuration.",
"I have written to the dataset authors, highlighting this issue. Waiting for their response. \r\n\r\nUpdate on 25th Feb: \r\nThe authors have replied back, they are updating the download link and will revert back shortly! \r\n\r\n```\r\nfirst of all thanks a lot for being interested in MuST-C and for building the data-loader.\r\n\r\nBefore answering your request, I'd like to clarify that the creation, maintenance, and expansion of MuST-c are not supported by any funded project, so this means that we need to find economic support for all these activities. This also includes permanently moving all the data to AWS or GCP. We are working at this with the goal of facilitating the use of MuST-C, but this is not something that can happen today. We hope to have some news ASAP and you will be among the first to be informed.\r\n\r\nI hope you understand our situation.\r\n```\r\n\r\n",
"Awesome, actually @lhoestq let's just ask the authors if we should host the dataset no? They could just use our links then as well for their website - what do you think? Is it fine to use our AWS dataset storage also as external links? ",
"Yes definitely. Shall we suggest them to create a dataset repository under their org on huggingface.co ? @julien-c \r\nThe dataset is around 1TB",
"Sounds good! \r\n\r\nOrder of magnitude is storage costs ~$20 per TB per month (not including bandwidth). \r\n\r\nHappy to provide this to the community as I feel this is an important dataset. Let us know what the authors want to do!\r\n\r\n",
"Great! @skyprince999, do you think you could ping the authors here or link to this thread? I think it could be a cool idea to host the dataset on our side then",
"Done. They replied back, and they want to have a call over a meet/ skype. Is that possible ? \r\nBtw @patrickvonplaten you are looped in that email (_pls check you gmail account_) ",
"Hello! Any news on this?",
"@gegallego there were some concerns regarding dataset usage & attribution by a for-profit company, so couldn't take it forward. Also the download links were unstable. \r\nBut I guess if you want to test the fairseq benchmarks, you can connect with them directly for downloading the dataset. ",
"Yes, that dataset is not easy to download... I had to copy it to my Google Drive and use `rsync` to be able to download it.\r\nHowever, we could add the dataset with a manual download, right?",
"yes that is possible. I couldn't unfortunately complete this PR, If you would like to add it, please feel free to do it. "
] | 2021-02-08T13:27:45
| 2021-05-14T14:53:34
| null |
CONTRIBUTOR
| null | null | null |
## Adding a Dataset
- **Name:** *IWSLT19*
- **Description:** *The Speech Translation Task addresses the translation of English audio into German and Portuguese text.*
- **Hompage:** *https://sites.google.com/view/iwslt-evaluation-2019/speech-translation*
- **Data:** *https://sites.google.com/view/iwslt-evaluation-2019/speech-translation* - all data under "Allowed Training Data" and "Development and Evalutaion Data for TED/How2"
- **Motivation:** Important speech dataset
If interested in tackling this issue, feel free to tag @patrickvonplaten
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1843/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1843/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1839
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1839/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1839/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1839/events
|
https://github.com/huggingface/datasets/issues/1839
| 803,559,164
|
MDU6SXNzdWU4MDM1NTkxNjQ=
| 1,839
|
Add Voxforge
|
{
"login": "patrickvonplaten",
"id": 23423619,
"node_id": "MDQ6VXNlcjIzNDIzNjE5",
"avatar_url": "https://avatars.githubusercontent.com/u/23423619?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/patrickvonplaten",
"html_url": "https://github.com/patrickvonplaten",
"followers_url": "https://api.github.com/users/patrickvonplaten/followers",
"following_url": "https://api.github.com/users/patrickvonplaten/following{/other_user}",
"gists_url": "https://api.github.com/users/patrickvonplaten/gists{/gist_id}",
"starred_url": "https://api.github.com/users/patrickvonplaten/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/patrickvonplaten/subscriptions",
"organizations_url": "https://api.github.com/users/patrickvonplaten/orgs",
"repos_url": "https://api.github.com/users/patrickvonplaten/repos",
"events_url": "https://api.github.com/users/patrickvonplaten/events{/privacy}",
"received_events_url": "https://api.github.com/users/patrickvonplaten/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
},
{
"id": 2725241052,
"node_id": "MDU6TGFiZWwyNzI1MjQxMDUy",
"url": "https://api.github.com/repos/huggingface/datasets/labels/speech",
"name": "speech",
"color": "d93f0b",
"default": false,
"description": ""
}
] |
open
| false
| null |
[] | null |
[] | 2021-02-08T13:19:56
| 2021-02-08T13:28:31
| null |
CONTRIBUTOR
| null | null | null |
## Adding a Dataset
- **Name:** *voxforge*
- **Description:** *VoxForge is a language classification dataset. It consists of user submitted audio clips submitted to the website. In this release, data from 6 languages is collected - English, Spanish, French, German, Russian, and Italian. Since the website is constantly updated, and for the sake of reproducibility, this release contains only recordings submitted prior to 2020-01-01. The samples are splitted between train, validation and testing so that samples from each speaker belongs to exactly one split.*
- **Paper:** *Homepage*: http://www.voxforge.org/
- **Data:** *http://www.voxforge.org/home/downloads*
- **Motivation:** Important speech dataset
- **TFDatasets Implementation**: https://www.tensorflow.org/datasets/catalog/voxforge
If interested in tackling this issue, feel free to tag @patrickvonplaten
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1839/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1839/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1835
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1835/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1835/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1835/events
|
https://github.com/huggingface/datasets/issues/1835
| 803,524,790
|
MDU6SXNzdWU4MDM1MjQ3OTA=
| 1,835
|
Add CHiME4 dataset
|
{
"login": "patrickvonplaten",
"id": 23423619,
"node_id": "MDQ6VXNlcjIzNDIzNjE5",
"avatar_url": "https://avatars.githubusercontent.com/u/23423619?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/patrickvonplaten",
"html_url": "https://github.com/patrickvonplaten",
"followers_url": "https://api.github.com/users/patrickvonplaten/followers",
"following_url": "https://api.github.com/users/patrickvonplaten/following{/other_user}",
"gists_url": "https://api.github.com/users/patrickvonplaten/gists{/gist_id}",
"starred_url": "https://api.github.com/users/patrickvonplaten/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/patrickvonplaten/subscriptions",
"organizations_url": "https://api.github.com/users/patrickvonplaten/orgs",
"repos_url": "https://api.github.com/users/patrickvonplaten/repos",
"events_url": "https://api.github.com/users/patrickvonplaten/events{/privacy}",
"received_events_url": "https://api.github.com/users/patrickvonplaten/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
},
{
"id": 2725241052,
"node_id": "MDU6TGFiZWwyNzI1MjQxMDUy",
"url": "https://api.github.com/repos/huggingface/datasets/labels/speech",
"name": "speech",
"color": "d93f0b",
"default": false,
"description": ""
}
] |
open
| false
| null |
[] | null |
[
"@patrickvonplaten not sure whether it is still needed, but willing to tackle this issue",
"Hey @patrickvonplaten, I have managed to download the zip on [here]( http://spandh.dcs.shef.ac.uk/chime_challenge/CHiME4/download.html) and successfully uploaded all the files on a hugging face dataset: \r\n\r\nhttps://huggingface.co/datasets/ksbai123/Chime4\r\n\r\nHowever I am getting this error when trying to use the dataset viewer:\r\n\r\n\r\n\r\nCan you take a look and let me know if I have missed any files please",
"@patrickvonplaten ?",
"Hi @KossaiSbai,\r\n\r\nThanks for your contribution.\r\n\r\nAs the issue is not strictly related to the `datasets` library, but to the specific implementation of the CHiME4 dataset, I have opened an issue in the Discussion tab of the dataset: https://huggingface.co/datasets/ksbai123/Chime4/discussions/2\r\nLet's continue the discussion there!"
] | 2021-02-08T12:36:38
| 2024-02-01T10:25:03
| null |
CONTRIBUTOR
| null | null | null |
## Adding a Dataset
- **Name:** Chime4
- **Description:** Chime4 is a dataset for automatic speech recognition. It is especially useful for evaluating models in a noisy environment and for multi-channel ASR
- **Paper:** Dataset comes from a channel: http://spandh.dcs.shef.ac.uk/chime_challenge/CHiME4/ . Results paper:
- **Data:** http://spandh.dcs.shef.ac.uk/chime_challenge/CHiME4/download.html
- **Motivation:** So far there are very little datasets for speech in `datasets`. Only `lbirispeech_asr` so far.
If interested in tackling this issue, feel free to tag @patrickvonplaten
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1835/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1835/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1830
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1830/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1830/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1830/events
|
https://github.com/huggingface/datasets/issues/1830
| 802,790,075
|
MDU6SXNzdWU4MDI3OTAwNzU=
| 1,830
|
using map on loaded Tokenizer 10x - 100x slower than default Tokenizer?
|
{
"login": "wumpusman",
"id": 7662740,
"node_id": "MDQ6VXNlcjc2NjI3NDA=",
"avatar_url": "https://avatars.githubusercontent.com/u/7662740?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/wumpusman",
"html_url": "https://github.com/wumpusman",
"followers_url": "https://api.github.com/users/wumpusman/followers",
"following_url": "https://api.github.com/users/wumpusman/following{/other_user}",
"gists_url": "https://api.github.com/users/wumpusman/gists{/gist_id}",
"starred_url": "https://api.github.com/users/wumpusman/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/wumpusman/subscriptions",
"organizations_url": "https://api.github.com/users/wumpusman/orgs",
"repos_url": "https://api.github.com/users/wumpusman/repos",
"events_url": "https://api.github.com/users/wumpusman/events{/privacy}",
"received_events_url": "https://api.github.com/users/wumpusman/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi @wumpusman \r\n`datasets` has a caching mechanism that allows to cache the results of `.map` so that when you want to re-run it later it doesn't recompute it again.\r\nSo when you do `.map`, what actually happens is:\r\n1. compute the hash used to identify your `map` for the cache\r\n2. apply your function on every batch\r\n\r\nThis can explain the time difference between your different experiments.\r\n\r\nThe hash computation time depends of how complex your function is. For a tokenizer, the hash computation scans the lists of the words in the tokenizer to identify this tokenizer. Usually it takes 2-3 seconds.\r\n\r\nAlso note that you can disable caching though using\r\n```python\r\nimport datasets\r\n\r\ndatasets.set_caching_enabled(False)\r\n```",
"Hi @lhoestq ,\r\n\r\nThanks for the reply. It's entirely possible that is the issue. Since it's a side project I won't be looking at it till later this week, but, I'll verify it by disabling caching and hopefully I'll see the same runtime. \r\n\r\nAppreciate the reference,\r\n\r\nMichael",
"I believe this is an actual issue, tokenizing a ~4GB txt file went from an hour and a half to ~10 minutes when I switched from my pre-trained tokenizer(on the same dataset) to the default gpt2 tokenizer.\r\nBoth were loaded using:\r\n```\r\nAutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)\r\n```\r\nI trained the tokenizer using ByteLevelBPETokenizer from the Tokenizers library and save it to a tokenizer.json file.\r\n\r\nI have tested the caching ideas above, changing the number of process, the TOKENIZERS_PARALLELISM env variable, keep_in_memory=True and batching with different sizes.\r\n\r\nApologies I can't really upload much code, but wanted to back up the finding and hopefully a fix/the problem can be found.\r\nI will comment back if I find a fix as well.",
"Hi @johncookds do you think this can come from one tokenizer being faster than the other one ? Can you try to compare their speed without using `datasets` just to make sure ?",
"Hi yes, I'm closing the loop here with some timings below. The issue seems to be at least somewhat/mainly with the tokenizer's themselves. Moreover legacy saves of the trainer tokenizer perform faster but differently than the new tokenizer.json saves(note nothing about the training process/adding of special tokens changed between the top two trained tokenizer tests, only the way it was saved). This is only a 3x slowdown vs like a 10x but I think the slowdown is most likely due to this.\r\n\r\n```\r\ntrained tokenizer - tokenizer.json save (same results for AutoTokenizer legacy_format=False):\r\nTokenizer time(seconds): 0.32767510414123535\r\nTokenized avg. length: 323.01\r\n\r\ntrained tokenizer - AutoTokenizer legacy_format=True:\r\nTokenizer time(seconds): 0.09258866310119629\r\nTokenized avg. length: 301.01\r\n\r\nGPT2 Tokenizer from huggingface\r\nTokenizer time(seconds): 0.1010282039642334\r\nTokenized avg. length: 461.21\r\n```",
"@lhoestq ,\r\n\r\nHi, which version of datasets has datasets.set_caching_enabled(False)? I get \r\nmodule 'datasets' has no attribute 'set_caching_enabled'. To hopefully get around this, I reran my code on a new set of data, and did so only once.\r\n\r\n@johncookds , thanks for chiming in, it looks this might be an issue of Tokenizer.\r\n\r\n**Tokenizer**: The runtime of GPT2TokenizerFast.from_pretrained(\"gpt2\") on 1000 chars is: **143 ms**\r\n**SlowTokenizer**: The runtime of a locally saved and loaded Tokenizer using the same vocab on 1000 chars is: **4.43 s**\r\n\r\nThat being said, I compared performance on the map function:\r\n\r\nRunning Tokenizer versus using it in the map function for 1000 chars goes from **141 ms** to **356 ms** \r\nRunning SlowTokenizer versus using it in the map function for 1000 chars with a single element goes from **4.43 s** to **9.76 s**\r\n\r\nI'm trying to figure out why the overhead of map would increase the time by double (figured it would be a fixed increase in time)? Though maybe this is expected behavior.\r\n\r\n@lhoestq, do you by chance know how I can redirect this issue to Tokenizer?\r\n\r\nRegards,\r\n\r\nMichael",
"Thanks for the experiments @johncookds and @wumpusman ! \r\n\r\n> Hi, which version of datasets has datasets.set_caching_enabled(False)?\r\n\r\nCurrently you have to install `datasets` from source to have this feature, but this will be available in the next release in a few days.\r\n\r\n> I'm trying to figure out why the overhead of map would increase the time by double (figured it would be a fixed increase in time)? Though maybe this is expected behavior.\r\n\r\nCould you also try with double the number of characters ? This should let us have an idea of the fixed cost (hashing) and the dynamic cost (actual tokenization, grows with the size of the input)\r\n\r\n> @lhoestq, do you by chance know how I can redirect this issue to Tokenizer?\r\n\r\nFeel free to post an issue on the `transformers` repo. Also I'm sure there should be related issues so you can also look for someone with the same concerns on the `transformers` repo.",
"@lhoestq,\r\n\r\nI just checked that previous run time was actually 3000 chars. I increased it to 6k chars, again, roughly double.\r\n\r\nSlowTokenizer **7.4 s** to **15.7 s**\r\nTokenizer: **276 ms** to **616 ms**\r\n\r\nI'll post this issue on Tokenizer, seems it hasn't quite been raised (albeit I noticed a similar issue that might relate).\r\n\r\nRegards,\r\n\r\nMichael",
"Hi, \r\nI'm following up here as I found my exact issue. It was with saving and re-loading the tokenizer. When I trained then processed the data without saving and reloading it, it was 10x-100x faster than when I saved and re-loaded it.\r\nBoth resulted in the exact same tokenized datasets as well. \r\nThere is additionally a bug where the older legacy tokenizer save does not preserve a learned tokenizing behavior if trained from scratch.\r\nUnderstand its not exactly Datasets related but hope it can help someone if they have the same issue.\r\nThanks!"
] | 2021-02-06T21:00:26
| 2021-02-24T21:56:14
| null |
NONE
| null | null | null |
This could total relate to me misunderstanding particular call functions, but I added words to a GPT2Tokenizer, and saved it to disk (note I'm only showing snippets but I can share more) and the map function ran much slower:
````
def save_tokenizer(original_tokenizer,text,path="simpledata/tokenizer"):
words_unique = set(text.split(" "))
for i in words_unique:
original_tokenizer.add_tokens(i)
original_tokenizer.save_pretrained(path)
tokenizer2 = GPT2Tokenizer.from_pretrained(os.path.join(experiment_path,experiment_name,"tokenizer_squad"))
train_set_baby=Dataset.from_dict({"text":[train_set["text"][0][0:50]]})
````
I then applied the dataset map function on a fairly small set of text:
```
%%time
train_set_baby = train_set_baby.map(lambda d:tokenizer2(d["text"]),batched=True)
```
The run time for train_set_baby.map was 6 seconds, and the batch itself was 2.6 seconds
**100% 1/1 [00:02<00:00, 2.60s/ba] CPU times: user 5.96 s, sys: 36 ms, total: 5.99 s Wall time: 5.99 s**
In comparison using (even after adding additional tokens):
`
tokenizer = GPT2TokenizerFast.from_pretrained("gpt2")`
```
%%time
train_set_baby = train_set_baby.map(lambda d:tokenizer2(d["text"]),batched=True)
```
The time is
**100% 1/1 [00:00<00:00, 34.09ba/s] CPU times: user 68.1 ms, sys: 16 µs, total: 68.1 ms Wall time: 62.9 ms**
It seems this might relate to the tokenizer save or load function, however, the issue appears to come up when I apply the loaded tokenizer to the map function.
I should also add that playing around with the amount of words I add to the tokenizer before I save it to disk and load it into memory appears to impact the time it takes to run the map function.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1830/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1830/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1810
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1810/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1810/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1810/events
|
https://github.com/huggingface/datasets/issues/1810
| 799,168,650
|
MDU6SXNzdWU3OTkxNjg2NTA=
| 1,810
|
Add Hateful Memes Dataset
|
{
"login": "gchhablani",
"id": 29076344,
"node_id": "MDQ6VXNlcjI5MDc2MzQ0",
"avatar_url": "https://avatars.githubusercontent.com/u/29076344?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/gchhablani",
"html_url": "https://github.com/gchhablani",
"followers_url": "https://api.github.com/users/gchhablani/followers",
"following_url": "https://api.github.com/users/gchhablani/following{/other_user}",
"gists_url": "https://api.github.com/users/gchhablani/gists{/gist_id}",
"starred_url": "https://api.github.com/users/gchhablani/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/gchhablani/subscriptions",
"organizations_url": "https://api.github.com/users/gchhablani/orgs",
"repos_url": "https://api.github.com/users/gchhablani/repos",
"events_url": "https://api.github.com/users/gchhablani/events{/privacy}",
"received_events_url": "https://api.github.com/users/gchhablani/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
},
{
"id": 3608941089,
"node_id": "LA_kwDODunzps7XHBIh",
"url": "https://api.github.com/repos/huggingface/datasets/labels/vision",
"name": "vision",
"color": "bfdadc",
"default": false,
"description": "Vision datasets"
}
] |
open
| false
| null |
[] | null |
[
"I am not sure, but would `datasets.Sequence(datasets.Sequence(datasets.Sequence(datasets.Value(\"int\")))` work?",
"Also, I found the information for loading only subsets of the data [here](https://github.com/huggingface/datasets/blob/master/docs/source/splits.rst).",
"Hi @lhoestq,\r\n\r\nRequest you to check this once.\r\n\r\nThanks,\r\nGunjan",
"Hi @gchhablani since Array2D doesn't support images of different sizes, I would suggest to store in the dataset the paths to the image file instead of the image data. This has the advantage of not decompressing the data (images are often compressed using jpeg, png etc.). Users can still apply `.map` to load the images if they want to. Though it would en up being Sequences features.\r\n\r\nIn the future we'll add support for ragged tensors for this case and update the relevant dataset with this feature."
] | 2021-02-02T10:53:59
| 2021-12-08T12:03:59
| null |
CONTRIBUTOR
| null | null | null |
## Add Hateful Memes Dataset
- **Name:** Hateful Memes
- **Description:** [https://ai.facebook.com/blog/hateful-memes-challenge-and-data-set]( https://ai.facebook.com/blog/hateful-memes-challenge-and-data-set)
- **Paper:** [https://arxiv.org/pdf/2005.04790.pdf](https://arxiv.org/pdf/2005.04790.pdf)
- **Data:** [This link](https://drivendata-competition-fb-hateful-memes-data.s3.amazonaws.com/XjiOc5ycDBRRNwbhRlgH.zip?AWSAccessKeyId=AKIARVBOBDCY4MWEDJKS&Signature=DaUuGgZWUgDHzEPPbyJ2PhSJ56Q%3D&Expires=1612816874)
- **Motivation:** Including multi-modal datasets to 🤗 datasets.
I will be adding this dataset. It requires the user to sign an agreement on DrivenData. So, it will be used with a manual download.
The issue with this dataset is that the images are of different sizes. The image datasets added so far (CIFAR-10 and MNIST) have a uniform shape throughout.
So something like
```python
datasets.Array2D(shape=(28, 28), dtype="uint8")
```
won't work for the images. How would I add image features then? I checked `datasets/features.py` but couldn't figure out the appropriate class for this. I'm assuming I would want to avoid re-sizing at all since we want the user to be able to access the original images.
Also, in case I want to load only a subset of the data, since the actual data is around 8.8GB, how would that be possible?
Thanks,
Gunjan
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1810/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1810/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1796
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1796/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1796/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1796/events
|
https://github.com/huggingface/datasets/issues/1796
| 797,329,905
|
MDU6SXNzdWU3OTczMjk5MDU=
| 1,796
|
Filter on dataset too much slowww
|
{
"login": "ayubSubhaniya",
"id": 20911334,
"node_id": "MDQ6VXNlcjIwOTExMzM0",
"avatar_url": "https://avatars.githubusercontent.com/u/20911334?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ayubSubhaniya",
"html_url": "https://github.com/ayubSubhaniya",
"followers_url": "https://api.github.com/users/ayubSubhaniya/followers",
"following_url": "https://api.github.com/users/ayubSubhaniya/following{/other_user}",
"gists_url": "https://api.github.com/users/ayubSubhaniya/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ayubSubhaniya/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ayubSubhaniya/subscriptions",
"organizations_url": "https://api.github.com/users/ayubSubhaniya/orgs",
"repos_url": "https://api.github.com/users/ayubSubhaniya/repos",
"events_url": "https://api.github.com/users/ayubSubhaniya/events{/privacy}",
"received_events_url": "https://api.github.com/users/ayubSubhaniya/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"When I use the filter on the arrow table directly, it works like butter. But I can't find a way to update the table in `Dataset` object.\r\n\r\n```\r\nds_table = dataset.data.filter(mask=dataset['flag'])\r\n```",
"@thomwolf @lhoestq can you guys please take a look and recommend some solution.",
"Hi ! Currently the filter method reads the dataset batch by batch to write a new, filtered, arrow file on disk. Therefore all the reading + writing can take some time.\r\nUsing a mask directly on the arrow table doesn't do any read or write operation therefore it's way quicker.\r\n\r\nReplacing the old table by the new one should do the job:\r\n```python\r\ndataset._data = dataset._data.filter(...)\r\n```\r\n\r\nNote: this is a **workaround** and in general users shouldn't have to do that. In particular if you did some `shuffle` or `select` before that then it would not work correctly since the indices mapping (index from `__getitem__` -> index in the table) would not be valid anymore. But if you haven't done any `shuffle`, `select`, `shard`, `train_test_split` etc. then it should work.\r\n\r\nIdeally it would be awesome to update the filter function to allow masking this way !\r\nIf you would like to give it a shot I will be happy to help :) ",
"Yes, would be happy to contribute. Thanks",
"Hi @lhoestq @ayubSubhaniya,\r\n\r\nIf there's no progress on this one, can I try working on it?\r\n\r\nThanks,\r\nGunjan",
"Sure @gchhablani feel free to start working on it, this would be very appreciated :)\r\nThis feature is would be really awesome, especially since arrow allows to mask really quickly and without having to rewrite the dataset on disk",
"Hi @lhoestq, any updates on this issue? The `filter` method is still veryyy slow 😕 ",
"No update so far, we haven't worked on this yet :/\r\n\r\nThough PyArrow is much more stable than 3 years ago so it would be a good time to dive into this",
"Hi @lhoestq, thanks a lot for the update! \r\n\r\nI would like to work on this(if possible). Could you please give me some steps regarding how should I approach this? Also any references would be great! ",
"I just played a bit with it to make sure using `table.filter()` is fine, but actually it seems to create a new table **in memory** :/\r\nThis is an issue since it can quickly fill the RAM, and `datasets`'s role is to make sure you can load bigger-than-memory datasets. Therefore I don't think it's a good idea in the end to use `table.filter()`\r\n\r\nAnyway I just ran OP's code an it runs in 20ms now on my side thanks to the I/O optimizations we did.\r\n\r\nAnother way to speed up `filter` is to add support pyarrow expressions though, using e.g. arrow formatting + dataset.filter (runs in 10ms on my side):\r\n\r\n```python\r\nimport pyarrow.dataset as pds\r\nimport pyarrow.compute as pc\r\n\r\nexpr = pc.field(\"flag\") == True\r\n\r\nfiltered = dataset.with_format(\"arrow\").filter(\r\n lambda t: pds.dataset(t).to_table(columns={\"mask\": expr})[0].to_numpy(),\r\n batched=True,\r\n).with_format(None)\r\n```"
] | 2021-01-30T04:09:19
| 2024-01-19T13:25:21
| null |
NONE
| null | null | null |
I have a dataset with 50M rows.
For pre-processing, I need to tokenize this and filter rows with the large sequence.
My tokenization took roughly 12mins. I used `map()` with batch size 1024 and multi-process with 96 processes.
When I applied the `filter()` function it is taking too much time. I need to filter sequences based on a boolean column.
Below are the variants I tried.
1. filter() with batch size 1024, single process (takes roughly 3 hr)
2. filter() with batch size 1024, 96 processes (takes 5-6 hrs ¯\\\_(ツ)\_/¯)
3. filter() with loading all data in memory, only a single boolean column (never ends).
Can someone please help?
Below is a sample code for small dataset.
```
from datasets import load_dataset
dataset = load_dataset('glue', 'mrpc', split='train')
dataset = dataset.map(lambda x: {'flag': random.randint(0,1)==1})
def _amplify(data):
return data
dataset = dataset.filter(_amplify, batch_size=1024, keep_in_memory=False, input_columns=['flag'])
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1796/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1796/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1790
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1790/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1790/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1790/events
|
https://github.com/huggingface/datasets/issues/1790
| 796,678,157
|
MDU6SXNzdWU3OTY2NzgxNTc=
| 1,790
|
ModuleNotFoundError: No module named 'apache_beam', when specific languages.
|
{
"login": "miyamonz",
"id": 6331508,
"node_id": "MDQ6VXNlcjYzMzE1MDg=",
"avatar_url": "https://avatars.githubusercontent.com/u/6331508?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/miyamonz",
"html_url": "https://github.com/miyamonz",
"followers_url": "https://api.github.com/users/miyamonz/followers",
"following_url": "https://api.github.com/users/miyamonz/following{/other_user}",
"gists_url": "https://api.github.com/users/miyamonz/gists{/gist_id}",
"starred_url": "https://api.github.com/users/miyamonz/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/miyamonz/subscriptions",
"organizations_url": "https://api.github.com/users/miyamonz/orgs",
"repos_url": "https://api.github.com/users/miyamonz/repos",
"events_url": "https://api.github.com/users/miyamonz/events{/privacy}",
"received_events_url": "https://api.github.com/users/miyamonz/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi !\r\n\r\nApache Beam is a framework used to define data transformation pipelines. These pipeline can then be run in many runtimes: DataFlow, Spark, Flink, etc. There also exist a local runner called the DirectRunner.\r\nWikipedia is a dataset that requires some parsing, so to allow the processing to be run on this kind of runtime we're using Apache Beam.\r\n\r\nAt Hugging Face we've already processed certain versions of wikipedia (the `20200501.en` one for example) so that users can directly download the processed version instead of using Apache Beam to process it.\r\nHowever for the japanese language we haven't processed it so you'll have to run the processing on your side.\r\nSo you do need Apache Beam to process `20200501.ja`.\r\n\r\nYou can install Apache Beam with\r\n```\r\npip install apache-beam\r\n```\r\n\r\nI think we can probably improve the error message to let users know of this subtlety.\r\nWhat #498 implied is that Apache Beam is not needed when you process a dataset that doesn't use Apache Beam.",
"Thanks for your reply! \r\nI understood.\r\n\r\nI tried again with installing apache-beam, add ` beam_runner=\"DirectRunner\"` and an anther `mwparserfromhell` is also required so I installed it.\r\nbut, it also failed. It exited 1 without error message.\r\n\r\n```py\r\nimport datasets\r\n# BTW, 20200501.ja doesn't exist at wikipedia, so I specified date argument\r\nwiki = datasets.load_dataset(\"wikipedia\", language=\"ja\", date=\"20210120\", cache_dir=\"./datasets\", beam_runner=\"DirectRunner\")\r\nprint(wiki)\r\n```\r\nand its log is below\r\n```\r\nUsing custom data configuration 20210120.ja\r\nDownloading and preparing dataset wikipedia/20210120.ja-date=20210120,language=ja (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to ./datasets/wikipedia/20210120.ja-date=20210120,language=ja/0.0.0/4021357e28509391eab2f8300d9b689e7e8f3a877ebb3d354b01577d497ebc63...\r\nKilled\r\n```\r\n\r\nI also tried on another machine because it may caused by insufficient resources.\r\n```\r\n$ python main.py\r\nUsing custom data configuration 20210120.ja\r\nDownloading and preparing dataset wikipedia/20210120.ja-date=20210120,language=ja (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to ./datasets/wikipedia/20210120.ja-date=20210120,language=ja/0.0.0/4021357e28509391eab2f8300d9b689e7e8f3a877ebb3d354b01577d497ebc63...\r\n\r\nTraceback (most recent call last):\r\n File \"main.py\", line 3, in <module>\r\n wiki = datasets.load_dataset(\"wikipedia\", language=\"ja\", date=\"20210120\", cache_dir=\"./datasets\", beam_runner=\"DirectRunner\")\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/datasets/load.py\", line 609, in load_dataset\r\n builder_instance.download_and_prepare(\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/datasets/builder.py\", line 526, in download_and_prepare\r\n self._download_and_prepare(\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/datasets/builder.py\", line 1069, in _download_and_prepare\r\n pipeline_results = pipeline.run()\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/pipeline.py\", line 561, in run\r\n return self.runner.run_pipeline(self, self._options)\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/direct/direct_runner.py\", line 126, in run_pipeline\r\n return runner.run_pipeline(pipeline, options)\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/portability/fn_api_runner/fn_runner.py\", line 182, in run_pipeline\r\n self._latest_run_result = self.run_via_runner_api(\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/portability/fn_api_runner/fn_runner.py\", line 193, in run_via_runner_api\r\n return self.run_stages(stage_context, stages)\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/portability/fn_api_runner/fn_runner.py\", line 358, in run_stages\r\n stage_results = self._run_stage(\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/portability/fn_api_runner/fn_runner.py\", line 549, in _run_stage\r\n last_result, deferred_inputs, fired_timers = self._run_bundle(\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/portability/fn_api_runner/fn_runner.py\", line 595, in _run_bundle\r\n result, splits = bundle_manager.process_bundle(\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/portability/fn_api_runner/fn_runner.py\", line 888, in process_bundle\r\n self._send_input_to_worker(process_bundle_id, transform_id, elements)\r\n File \"/home/miyamonz/.cache/pypoetry/virtualenvs/try-datasets-4t4JWXxu-py3.8/lib/python3.8/site-packages/apache_beam/runners/portability/fn_api_runner/fn_runner.py\", line 765, in _send_input_to_worker\r\n data_out.write(byte_stream)\r\n File \"apache_beam/coders/stream.pyx\", line 42, in apache_beam.coders.stream.OutputStream.write\r\n File \"apache_beam/coders/stream.pyx\", line 47, in apache_beam.coders.stream.OutputStream.write\r\n File \"apache_beam/coders/stream.pyx\", line 109, in apache_beam.coders.stream.OutputStream.extend\r\nAssertionError: OutputStream realloc failed.\r\n```\r\n\r\n",
"Hi @miyamonz,\r\n\r\nI tried replicating this issue using the same snippet used by you. I am able to download the dataset without any issues, although I stopped it in the middle because the dataset is huge.\r\n\r\nBased on a similar issue [here](https://github.com/google-research/fixmatch/issues/23), it could be related to your environment setup, although I am just guessing here. Can you share these details?",
"thanks for your reply and sorry for my late response.\r\n\r\n## environment\r\nmy local machine environment info\r\n- Ubuntu on WSL2\r\n\r\n`lsb_release -a`\r\n```\r\nNo LSB modules are available.\r\nDistributor ID: Ubuntu\r\nDescription: Ubuntu 20.04.2 LTS\r\nRelease: 20.04\r\nCodename: focal\r\n```\r\n\r\nRTX 2070 super\r\nInside WSL, there is no nvidia-msi command. I don't know why.\r\nBut, `torch.cuda.is_available()` is true and when I start something ML training code GPU usage is growing up, so I think it works.\r\n\r\nFrom PowerShell, there is nvidia-smi.exe and result is below.\r\n```\r\n+-----------------------------------------------------------------------------+\r\n| NVIDIA-SMI 470.05 Driver Version: 470.05 CUDA Version: 11.3 |\r\n|-------------------------------+----------------------+----------------------+\r\n| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |\r\n| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |\r\n| | | MIG M. |\r\n|===============================+======================+======================|\r\n| 0 NVIDIA GeForce ... WDDM | 00000000:09:00.0 On | N/A |\r\n| 0% 30C P8 19W / 175W | 523MiB / 8192MiB | 3% Default |\r\n| | | N/A |\r\n+-------------------------------+----------------------+----------------------+\r\n\r\n+-----------------------------------------------------------------------------+\r\n| Processes: |\r\n| GPU GI CI PID Type Process name GPU Memory |\r\n| ID ID Usage |\r\n|=============================================================================|\r\n| 0 N/A N/A 1728 C+G Insufficient Permissions N/A |\r\n| 0 N/A N/A 3672 C+G ...ekyb3d8bbwe\\YourPhone.exe N/A |\r\n| 0 N/A N/A 6304 C+G ...2txyewy\\TextInputHost.exe N/A |\r\n| 0 N/A N/A 8648 C+G C:\\Windows\\explorer.exe N/A |\r\n| 0 N/A N/A 9536 C+G ...y\\ShellExperienceHost.exe N/A |\r\n| 0 N/A N/A 10668 C+G ...5n1h2txyewy\\SearchApp.exe N/A |\r\n| 0 N/A N/A 10948 C+G ...artMenuExperienceHost.exe N/A |\r\n| 0 N/A N/A 11988 C+G ...8wekyb3d8bbwe\\Cortana.exe N/A |\r\n| 0 N/A N/A 12464 C+G ...cw5n1h2txyewy\\LockApp.exe N/A |\r\n| 0 N/A N/A 13280 C+G ...upport\\CEF\\Max Helper.exe N/A |\r\n| 0 N/A N/A 15948 C+G ...t\\GoogleIMEJaRenderer.exe N/A |\r\n| 0 N/A N/A 16128 C+G ...ram Files\\Slack\\Slack.exe N/A |\r\n| 0 N/A N/A 19096 C+G ...8bbwe\\WindowsTerminal.exe N/A |\r\n+-----------------------------------------------------------------------------+\r\n```\r\n\r\nI don't know what should I show in such a case. If it's not enough, please tell me some commands.\r\n\r\n---\r\n## what I did\r\nI surveyed more and I found 2 issues.\r\n\r\nAbout the first one, I wrote it as a new issue.\r\nhttps://github.com/huggingface/datasets/issues/2031\r\n\r\nThe error I mentioned in the previous comment above, which occurred on my local machine, is no longer occurring.\r\n\r\nBut, it still failed. In the previous comment, I wrote `AssertionError: OutputStream realloc failed.` happen on another machine. It also happens on my local machine.\r\n\r\nHere's what I've tried.\r\n\r\nthe wikipedia.py downloads these xml.bz2 files based on dumpstatus.json\r\nIn Japanese Wikipedia dataset that I specified, it will download these 6 files.\r\n\r\n\r\n`https://dumps.wikimedia.org/jawiki/20210120/dumpstatus.json`\r\nand filtered json based on wikipedia.py is below.\r\n```json\r\n {\r\n \"jobs\": {\r\n \"articlesmultistreamdump\": {\r\n \"files\": {\r\n \"jawiki-20210120-pages-articles-multistream1.xml-p1p114794.bz2\": {\r\n \"url\": \"/jawiki/20210120/jawiki-20210120-pages-articles-multistream1.xml-p1p114794.bz2\"\r\n },\r\n \"jawiki-20210120-pages-articles-multistream2.xml-p114795p390428.bz2\": {\r\n \"url\": \"/jawiki/20210120/jawiki-20210120-pages-articles-multistream2.xml-p114795p390428.bz2\"\r\n },\r\n \"jawiki-20210120-pages-articles-multistream3.xml-p390429p902407.bz2\": {\r\n \"url\": \"/jawiki/20210120/jawiki-20210120-pages-articles-multistream3.xml-p390429p902407.bz2\"\r\n },\r\n \"jawiki-20210120-pages-articles-multistream4.xml-p902408p1721646.bz2\": {\r\n \"url\": \"/jawiki/20210120/jawiki-20210120-pages-articles-multistream4.xml-p902408p1721646.bz2\"\r\n },\r\n \"jawiki-20210120-pages-articles-multistream5.xml-p1721647p2807947.bz2\": {\r\n \"url\": \"/jawiki/20210120/jawiki-20210120-pages-articles-multistream5.xml-p1721647p2807947.bz2\"\r\n },\r\n \"jawiki-20210120-pages-articles-multistream6.xml-p2807948p4290013.bz2\": {\r\n \"url\": \"/jawiki/20210120/jawiki-20210120-pages-articles-multistream6.xml-p2807948p4290013.bz2\"\r\n }\r\n }\r\n }\r\n }\r\n }\r\n```\r\n\r\nSo, I tried running with fewer resources by modifying this line.\r\nhttps://github.com/huggingface/datasets/blob/13a5b7db992ad5cf77895e4c0f76595314390418/datasets/wikipedia/wikipedia.py#L524\r\nI changed it like this. just change filepaths list.\r\n` | \"Initialize\" >> beam.Create(filepaths[:1])`\r\n\r\nand I added a print line inside for the loop of _extract_content.\r\nlike this `if(i % 100000 == 0): print(i)`\r\n\r\nfirst, without modification, it always stops after all _extract_content is done.\r\n\r\n- `filepaths[:1]` then it succeeded.\r\n- `filepaths[:2]` then it failed.\r\nI don't try all patterns because each pattern takes a long time.\r\n\r\n### my opinion\r\nIt seems it's successful when the entire file size is small.\r\n \r\nso, at least it doesn't file-specific issue.\r\n\r\n\r\nI don't know it's true but I think when beam_writter writes into a file, it consumes memory depends on its entire file.\r\nbut It's correct Apache Beam's behavior? I'm not familiar with this library.\r\n",
"I don't know if this is related, but there is this issue on the wikipedia processing that you reported at #2031 (open PR is at #2037 ) .\r\nDoes the fix your proposed at #2037 helps in your case ?\r\n\r\nAnd for information, the DirectRunner of Apache Beam is not optimized for memory intensive tasks, so you must be right when you say that it uses the memory for the entire file.",
"the #2037 doesn't solve my problem directly, but I found the point!\r\n\r\nhttps://github.com/huggingface/datasets/blob/349ac4398a3bcae6356f14c5754483383a60e8a4/datasets/wikipedia/wikipedia.py#L523\r\nthis `beam.transforms.Reshuffle()` cause the memory error.\r\n\r\nit makes sense if I consider the shuffle means. Beam's reshuffle seems need put all data in memory.\r\nPreviously I doubt that this line causes error, but at that time another bug showed in #2037 made error, so I can't found it.\r\n\r\nAnyway, I comment out this line, and run load_dataset, then it works!\r\n\r\n```python\r\nwiki = datasets.load_dataset(\r\n \"./wikipedia.py\",\r\n cache_dir=\"./datasets\",\r\n beam_runner=\"DirectRunner\",\r\n language=\"ja\",\r\n date=\"20210120\",\r\n)[\"train\"]\r\n```\r\n\r\n\r\nDataset has already shuffle function. https://github.com/huggingface/datasets/blob/349ac4398a3bcae6356f14c5754483383a60e8a4/src/datasets/arrow_dataset.py#L2069\r\nSo, though I don't know it's difference correctly, but I think Beam's reshuffle isn't be needed. How do you think?",
"The reshuffle is needed when you use parallelism.\r\nThe objective is to redistribute the articles evenly on the workers, since the `_extract_content` step generated many articles per file. By using reshuffle, we can split the processing of the articles of one file into several workers. Without reshuffle, all the articles of one file would be processed on the same worker that read the file, making the whole process take a very long time.",
"Maybe the reshuffle step can be added only if the runner is not a DirectRunner ?"
] | 2021-01-29T08:17:24
| 2021-03-25T12:10:51
| null |
CONTRIBUTOR
| null | null | null |
```py
import datasets
wiki = datasets.load_dataset('wikipedia', '20200501.ja', cache_dir='./datasets')
```
then `ModuleNotFoundError: No module named 'apache_beam'` happend.
The error doesn't appear when it's '20200501.en'.
I don't know Apache Beam, but according to #498 it isn't necessary when it's saved to local. is it correct?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1790/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1790/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1706
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1706/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1706/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1706/events
|
https://github.com/huggingface/datasets/issues/1706
| 781,494,476
|
MDU6SXNzdWU3ODE0OTQ0NzY=
| 1,706
|
Error when downloading a large dataset on slow connection.
|
{
"login": "lucadiliello",
"id": 23355969,
"node_id": "MDQ6VXNlcjIzMzU1OTY5",
"avatar_url": "https://avatars.githubusercontent.com/u/23355969?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lucadiliello",
"html_url": "https://github.com/lucadiliello",
"followers_url": "https://api.github.com/users/lucadiliello/followers",
"following_url": "https://api.github.com/users/lucadiliello/following{/other_user}",
"gists_url": "https://api.github.com/users/lucadiliello/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lucadiliello/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lucadiliello/subscriptions",
"organizations_url": "https://api.github.com/users/lucadiliello/orgs",
"repos_url": "https://api.github.com/users/lucadiliello/repos",
"events_url": "https://api.github.com/users/lucadiliello/events{/privacy}",
"received_events_url": "https://api.github.com/users/lucadiliello/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi ! Is this an issue you have with `openwebtext` specifically or also with other datasets ?\r\n\r\nIt looks like the downloaded file is corrupted and can't be extracted using `tarfile`.\r\nCould you try loading it again with \r\n```python\r\nimport datasets\r\ndatasets.load_dataset(\"openwebtext\", download_mode=\"force_redownload\")\r\n```"
] | 2021-01-07T17:48:15
| 2021-01-13T10:35:02
| null |
CONTRIBUTOR
| null | null | null |
I receive the following error after about an hour trying to download the `openwebtext` dataset.
The code used is:
```python
import datasets
datasets.load_dataset("openwebtext")
```
> Traceback (most recent call last): [4/28]
> File "<stdin>", line 1, in <module>
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/site-packages/datasets/load.py", line 610, in load_dataset
> ignore_verifications=ignore_verifications,
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/site-packages/datasets/builder.py", line 515, in download_and_prepare
> dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/site-packages/datasets/builder.py", line 570, in _download_and_prepare
> split_generators = self._split_generators(dl_manager, **split_generators_kwargs)
> File "/home/lucadiliello/.cache/huggingface/modules/datasets_modules/datasets/openwebtext/5c636399c7155da97c982d0d70ecdce30fbca66a4eb4fc768ad91f8331edac02/openwebtext.py", line 62, in _split_generators
> dl_dir = dl_manager.download_and_extract(_URL)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/site-packages/datasets/utils/download_manager.py", line 254, in download_and_extract
> return self.extract(self.download(url_or_urls))
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/site-packages/datasets/utils/download_manager.py", line 235, in extract
> num_proc=num_proc,
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/site-packages/datasets/utils/py_utils.py", line 225, in map_nested
> return function(data_struct)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/site-packages/datasets/utils/file_utils.py", line 343, in cached_path
> tar_file.extractall(output_path_extracted)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/tarfile.py", line 2000, in extractall
> numeric_owner=numeric_owner)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/tarfile.py", line 2042, in extract
> numeric_owner=numeric_owner)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/tarfile.py", line 2112, in _extract_member
> self.makefile(tarinfo, targetpath)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/tarfile.py", line 2161, in makefile
> copyfileobj(source, target, tarinfo.size, ReadError, bufsize)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/tarfile.py", line 253, in copyfileobj
> buf = src.read(remainder)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/lzma.py", line 200, in read
> return self._buffer.read(size)
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/_compression.py", line 68, in readinto
> data = self.read(len(byte_view))
> File "/home/lucadiliello/anaconda3/envs/nlp/lib/python3.7/_compression.py", line 99, in read
> raise EOFError("Compressed file ended before the "
> EOFError: Compressed file ended before the end-of-stream marker was reached
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1706/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1706/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1687
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1687/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1687/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1687/events
|
https://github.com/huggingface/datasets/issues/1687
| 779,004,894
|
MDU6SXNzdWU3NzkwMDQ4OTQ=
| 1,687
|
Question: Shouldn't .info be a part of DatasetDict?
|
{
"login": "KennethEnevoldsen",
"id": 23721977,
"node_id": "MDQ6VXNlcjIzNzIxOTc3",
"avatar_url": "https://avatars.githubusercontent.com/u/23721977?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/KennethEnevoldsen",
"html_url": "https://github.com/KennethEnevoldsen",
"followers_url": "https://api.github.com/users/KennethEnevoldsen/followers",
"following_url": "https://api.github.com/users/KennethEnevoldsen/following{/other_user}",
"gists_url": "https://api.github.com/users/KennethEnevoldsen/gists{/gist_id}",
"starred_url": "https://api.github.com/users/KennethEnevoldsen/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/KennethEnevoldsen/subscriptions",
"organizations_url": "https://api.github.com/users/KennethEnevoldsen/orgs",
"repos_url": "https://api.github.com/users/KennethEnevoldsen/repos",
"events_url": "https://api.github.com/users/KennethEnevoldsen/events{/privacy}",
"received_events_url": "https://api.github.com/users/KennethEnevoldsen/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"We could do something. There is a part of `.info` which is split specific (cache files, split instructions) but maybe if could be made to work.",
"Yes this was kinda the idea I was going for. DatasetDict.info would be the shared info amongs the datasets (maybe even some info on how they differ). "
] | 2021-01-05T13:08:41
| 2021-01-07T10:18:06
| null |
CONTRIBUTOR
| null | null | null |
Currently, only `Dataset` contains the .info or .features, but as many datasets contains standard splits (train, test) and thus the underlying information is the same (or at least should be) across the datasets.
For instance:
```
>>> ds = datasets.load_dataset("conll2002", "es")
>>> ds.info
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'DatasetDict' object has no attribute 'info'
```
I could imagine that this wouldn't work for datasets dicts which hold entirely different datasets (multimodal datasets), but it seems odd that splits of the same dataset is treated the same as what is essentially different datasets.
Intuitively it would also make sense that if a dataset is supplied via. the load_dataset that is have a common .info which covers the entire dataset.
It is entirely possible that I am missing another perspective
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1687/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1687/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1670
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1670/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1670/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1670/events
|
https://github.com/huggingface/datasets/issues/1670
| 776,608,579
|
MDU6SXNzdWU3NzY2MDg1Nzk=
| 1,670
|
wiki_dpr pre-processing performance
|
{
"login": "dbarnhart",
"id": 753898,
"node_id": "MDQ6VXNlcjc1Mzg5OA==",
"avatar_url": "https://avatars.githubusercontent.com/u/753898?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/dbarnhart",
"html_url": "https://github.com/dbarnhart",
"followers_url": "https://api.github.com/users/dbarnhart/followers",
"following_url": "https://api.github.com/users/dbarnhart/following{/other_user}",
"gists_url": "https://api.github.com/users/dbarnhart/gists{/gist_id}",
"starred_url": "https://api.github.com/users/dbarnhart/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/dbarnhart/subscriptions",
"organizations_url": "https://api.github.com/users/dbarnhart/orgs",
"repos_url": "https://api.github.com/users/dbarnhart/repos",
"events_url": "https://api.github.com/users/dbarnhart/events{/privacy}",
"received_events_url": "https://api.github.com/users/dbarnhart/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 2067401494,
"node_id": "MDU6TGFiZWwyMDY3NDAxNDk0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/Dataset%20discussion",
"name": "Dataset discussion",
"color": "72f99f",
"default": false,
"description": "Discussions on the datasets"
}
] |
open
| false
| null |
[] | null |
[
"Hi ! And thanks for the tips :) \r\n\r\nIndeed currently `wiki_dpr` takes some time to be processed.\r\nMultiprocessing for dataset generation is definitely going to speed up things.\r\n\r\nRegarding the index note that for the default configurations, the index is downloaded instead of being built, which avoid spending time on constructing the index. However in other cases it would be awesome to make the construction faster.\r\n\r\nAny contribution that can help things faster are welcome. In particular in you have some code that can build a wiki_dpr IVF PQ index in a sharded GPU setup and would like to share it, we can add it to an `examples` folder. In particular since faiss is becoming the library of reference for dataset indexing for tasks like Open Domain Question Answering.\r\n\r\n",
"I'd be happy to contribute something when I get the time, probably adding multiprocessing and / or cython support to wiki_dpr. I've written cythonized apache beam code before as well.\r\n\r\nFor sharded index building, I used the FAISS example code for indexing 1 billion vectors as a start. I'm sure you're aware that the documentation isn't great, but the source code is fairly easy to follow.",
"Nice thanks ! That would be awesome to make its construction faster :) "
] | 2020-12-30T19:41:43
| 2021-01-28T09:41:36
| null |
NONE
| null | null | null |
I've been working with wiki_dpr and noticed that the dataset processing is seriously impaired in performance [1]. It takes about 12h to process the entire dataset. Most of this time is simply loading and processing the data, but the actual indexing is also quite slow (3h).
I won't repeat the concerns around multiprocessing as they are addressed in other issues (#786), but this is the first obvious thing to do. Using cython to speed up the text manipulation may be also help. Loading and processing a dataset of this size in under 15 minutes does not seem unreasonable on a modern multi-core machine. I have hit such targets myself on similar tasks. Would love to see this improve.
The other issue is that it takes 3h to construct the FAISS index. If only we could use GPUs with HNSW, but we can't. My sharded GPU indexing code can build an IVF + PQ index in 10 minutes on 20 million vectors. Still, 3h seems slow even for the CPU.
It looks like HF is adding only 1000 vectors at a time by default [2], whereas the faiss benchmarks adds 1 million vectors at a time (effectively) [3]. It's possible the runtime could be reduced with a larger batch. Also, it looks like project dependencies ultimately use OpenBLAS, but this is known to have issues when combined with OpenMP, which HNSW does [3]. A workaround is to set the environment variable `OMP_WAIT_POLICY=PASSIVE` via `os.environ` or similar.
References:
[1] https://github.com/huggingface/datasets/blob/master/datasets/wiki_dpr/wiki_dpr.py
[2] https://github.com/huggingface/datasets/blob/master/src/datasets/search.py
[3] https://github.com/facebookresearch/faiss/blob/master/benchs/bench_hnsw.py
[4] https://github.com/facebookresearch/faiss/issues/422
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1670/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1670/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1615
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1615/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1615/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1615/events
|
https://github.com/huggingface/datasets/issues/1615
| 771,641,088
|
MDU6SXNzdWU3NzE2NDEwODg=
| 1,615
|
Bug: Can't download TriviaQA with `load_dataset` - custom `cache_dir`
|
{
"login": "SapirWeissbuch",
"id": 44585792,
"node_id": "MDQ6VXNlcjQ0NTg1Nzky",
"avatar_url": "https://avatars.githubusercontent.com/u/44585792?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/SapirWeissbuch",
"html_url": "https://github.com/SapirWeissbuch",
"followers_url": "https://api.github.com/users/SapirWeissbuch/followers",
"following_url": "https://api.github.com/users/SapirWeissbuch/following{/other_user}",
"gists_url": "https://api.github.com/users/SapirWeissbuch/gists{/gist_id}",
"starred_url": "https://api.github.com/users/SapirWeissbuch/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/SapirWeissbuch/subscriptions",
"organizations_url": "https://api.github.com/users/SapirWeissbuch/orgs",
"repos_url": "https://api.github.com/users/SapirWeissbuch/repos",
"events_url": "https://api.github.com/users/SapirWeissbuch/events{/privacy}",
"received_events_url": "https://api.github.com/users/SapirWeissbuch/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Hi @SapirWeissbuch,\r\nWhen you are saying it freezes, at that time it is unzipping the file from the zip file it downloaded. Since it's a very heavy file it'll take some time. It was taking ~11GB after unzipping when it started reading examples for me. Hope that helps!\r\n\r\n",
"Hi @bhavitvyamalik \r\nThanks for the reply!\r\nActually I let it run for 30 minutes before I killed the process. In this time, 30GB were extracted (much more than 11GB), I checked the size of the destination directory.\r\n\r\nWhat version of Datasets are you using?\r\n",
"I'm using datasets version: 1.1.3. I think you should drop `cache_dir` and use only\r\n`dataset = datasets.load_dataset(\"trivia_qa\", \"rc\")`\r\n\r\nTried that on colab and it's working there too\r\n\r\n",
"Train, Validation, and Test splits contain 138384, 18669, and 17210 samples respectively. It takes some time to read the samples. Even in your colab notebook it was reading the samples before you killed the process. Let me know if it works now!",
"Hi, it works on colab but it still doesn't work on my computer, same problem as before - overly large and long extraction process.\r\nI have to use a custom 'cache_dir' because I don't have any space left in my home directory where it is defaulted, maybe this could be the issue?",
"I tried running this again - More details of the problem:\r\nCode:\r\n```\r\ndatasets.load_dataset(\"trivia_qa\", \"rc\", cache_dir=\"/path/to/cache\")\r\n```\r\n\r\nThe output:\r\n```\r\nDownloading and preparing dataset trivia_qa/rc (download: 2.48 GiB, generated: 14.92 GiB, post-processed: Unknown size, total: 17.40 GiB) to path/to/cache/trivia_qa/rc/1.1.0/e734e28133f4d9a353af322aa52b9f266f6f27cbf2f072690a1694e577546b0d... \r\nDownloading: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2.67G/2.67G [03:38<00:00, 12.2MB/s]\r\n\r\n```\r\nThe process continues (no progress bar is visible).\r\nI tried `du -sh .` in `path/to/cache`, and the size keeps increasing, reached 35G before I killed the process.\r\n\r\nGoogle Colab with custom `cache_dir` has same issue.\r\nhttps://colab.research.google.com/drive/1nn1Lw02GhfGFylzbS2j6yksGjPo7kkN-?usp=sharing#scrollTo=2G2O0AeNIXan",
"1) You can clear the huggingface folder in your `.cache` directory to use default directory for datasets. Speed of extraction and loading of samples depends a lot on your machine's configurations too.\r\n\r\n2) I tried on colab `dataset = datasets.load_dataset(\"trivia_qa\", \"rc\", cache_dir = \"./datasets\")`. After memory usage reached around 42GB (starting from 32GB used already), the dataset was loaded in the memory. Even Your colab notebook shows \r\n\r\nwhich means it's loaded now.",
"Facing the same issue.\r\nI am able to download datasets without `cache_dir`, however, when I specify the `cache_dir`, the process hangs indefinitely after partial download. \r\nTried for `data = load_dataset(\"cnn_dailymail\", \"3.0.0\")`",
"Hi @ashutoshml,\r\nI tried this and it worked for me:\r\n`data = load_dataset(\"cnn_dailymail\", \"3.0.0\", cache_dir=\"./dummy\")`\r\n\r\nI'm using datasets==1.8.0. It took around 3-4 mins for dataset to unpack and start loading examples.",
"Ok. I waited for 20-30 mins, and it still is stuck.\r\nI am using datasets==1.8.0.\r\n\r\nIs there anyway to check what is happening? like a` --verbose` flag?\r\n\r\n\r\n"
] | 2020-12-20T17:27:38
| 2021-06-25T13:11:33
| null |
NONE
| null | null | null |
Hello,
I'm having issue downloading TriviaQA dataset with `load_dataset`.
## Environment info
- `datasets` version: 1.1.3
- Platform: Linux-4.19.129-aufs-1-x86_64-with-debian-10.1
- Python version: 3.7.3
## The code I'm running:
```python
import datasets
dataset = datasets.load_dataset("trivia_qa", "rc", cache_dir = "./datasets")
```
## The output:
1. Download begins:
```
Downloading and preparing dataset trivia_qa/rc (download: 2.48 GiB, generated: 14.92 GiB, post-processed: Unknown size, total: 17.40 GiB) to /cs/labs/gabis/sapirweissbuch/tr
ivia_qa/rc/1.1.0/e734e28133f4d9a353af322aa52b9f266f6f27cbf2f072690a1694e577546b0d...
Downloading: 17%|███████████████████▉ | 446M/2.67G [00:37<04:45, 7.77MB/s]
```
2. 100% is reached
3. It got stuck here for about an hour, and added additional 30G of data to "./datasets" directory. I killed the process eventually.
A similar issue can be observed in Google Colab:
https://colab.research.google.com/drive/1nn1Lw02GhfGFylzbS2j6yksGjPo7kkN-?usp=sharing
## Expected behaviour:
The dataset "TriviaQA" should be successfully downloaded.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1615/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1615/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/1474
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1474/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1474/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1474/events
|
https://github.com/huggingface/datasets/pull/1474
| 762,083,706
|
MDExOlB1bGxSZXF1ZXN0NTM2NjY4MjU3
| 1,474
|
Create JSON dummy data without loading all dataset in memory
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2020-12-11T08:44:23
| 2022-07-06T15:19:47
| null |
MEMBER
| null | false
|
{
"url": "https://api.github.com/repos/huggingface/datasets/pulls/1474",
"html_url": "https://github.com/huggingface/datasets/pull/1474",
"diff_url": "https://github.com/huggingface/datasets/pull/1474.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/1474.patch",
"merged_at": null
}
|
See #1442.
The statement `json.load()` loads **all the file content in memory**.
In order to avoid this, file content should be parsed **iteratively**, by using the library `ijson` e.g.
I have refactorized the code into a function `_create_json_dummy_data` and I have added some tests.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1474/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1474/timeline
| null | null | true
|
https://api.github.com/repos/huggingface/datasets/issues/1369
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1369/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1369/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1369/events
|
https://github.com/huggingface/datasets/pull/1369
| 760,227,776
|
MDExOlB1bGxSZXF1ZXN0NTM1MDk0NDk1
| 1,369
|
Use passed --cache_dir for modules cache
|
{
"login": "albertvillanova",
"id": 8515462,
"node_id": "MDQ6VXNlcjg1MTU0NjI=",
"avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/albertvillanova",
"html_url": "https://github.com/albertvillanova",
"followers_url": "https://api.github.com/users/albertvillanova/followers",
"following_url": "https://api.github.com/users/albertvillanova/following{/other_user}",
"gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}",
"starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions",
"organizations_url": "https://api.github.com/users/albertvillanova/orgs",
"repos_url": "https://api.github.com/users/albertvillanova/repos",
"events_url": "https://api.github.com/users/albertvillanova/events{/privacy}",
"received_events_url": "https://api.github.com/users/albertvillanova/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"I have a question: why not using a tmp dir instead, like the DummyDataGeneratorDownloadManager does?",
"Hi @lhoestq, I am trying to understand better the logic...\r\n\r\nWhy do we have a `dynamic_module_path` besides the modules cache path?\r\n```python\r\nDYNAMIC_MODULES_PATH = os.path.join(HF_MODULES_CACHE, \"datasets_modules\")\r\n```\r\nMoreover, 2 subdirectories (for datasets and for metrics) were created inside it:\r\n```python\r\nDATASETS_PATH = os.path.join(DYNAMIC_MODULES_PATH, \"datasets\")\r\nMETRICS_PATH = os.path.join(DYNAMIC_MODULES_PATH, \"metrics\")\r\n```",
"Hi :) \r\nThe modules cache path is the path added to `sys.path`.\r\nTherefore inside we need to have a folder that is going to be a package: `datasets_modules`.\r\nThis package will contain dynamic modules, i.e. datasets and metrics modules added on-the-fly.\r\nThen we have two sub-modules `datasets_modules.datasets` and `datasets_modules.metrics`.\r\n\r\nMaybe we can make things more explicit in the code with some comments explaining the structure, and maybe better variable naming as well..\r\n\r\nAlso I wanted to say that I started to work on offline loading of modules in #1726 and actually it lead to do similar changes to what you did to control the path where modules are stored.",
"Hi @lhoestq, I see...\r\n\r\nIndeed I was also creating a draft for test_load, to clarify the expected behavior... ;)\r\n\r\nSo, for the command line:\r\n```sh\r\npython datasets-cli test datasets/<my-dataset-folder> --save_infos --all_configs --cache_dir <my-cache-dir>\r\n```\r\nthe `cache_dir` argument refers to dataset cache dir. We do not have control over the modules cache dir, but we would like to have. And if I understand well, you suggest adding another argument `dynamic_module_path`. Am I right?",
"> So, for the command line:\r\n> \r\n> ```shell\r\n> python datasets-cli test datasets/<my-dataset-folder> --save_infos --all_configs --cache_dir <my-cache-dir>\r\n> ```\r\n> \r\n> the `cache_dir` argument refers to dataset cache dir. We do not have control over the modules cache dir, but we would like to have. And if I understand well, you suggest adding another argument `dynamic_module_path`. Am I right?\r\n\r\nYes the cache_dir is used to download files and also so save the dataset arrow files.\r\nThis is indeed different from the path for dynamic modules.\r\n\r\nI suggested to have `dynamic_module_path` as a parameter but actually this is the parent directory `hf_modules_cache` that we would need (it's the one that is passed to `init_dynamic_modules ` that we need to add to `sys.path`).\r\n\r\nCurrently it's already possible to override it using the env variable `HF_MODULES_CACHE` but we can imagine having it as a parameter as well.\r\n\r\nThis way the user controls both the `cache_dir` and the `hf_modules_cache` which are the two places used by the library to read/write stuff.\r\n\r\n",
"I think #1726 is going to be merged pretty soon. Maybe can work on this as soon as it's merged to avoid doing the same things twice and to avoid conflicts ?",
"I agree. Indeed I took some of your code in one of my last commit, to try to implement the logic you described."
] | 2020-12-09T10:59:59
| 2022-07-06T15:19:47
| null |
MEMBER
| null | false
|
{
"url": "https://api.github.com/repos/huggingface/datasets/pulls/1369",
"html_url": "https://github.com/huggingface/datasets/pull/1369",
"diff_url": "https://github.com/huggingface/datasets/pull/1369.diff",
"patch_url": "https://github.com/huggingface/datasets/pull/1369.patch",
"merged_at": null
}
|
When passed `--cache_dir` arg:
```shell
python datasets-cli test datasets/<my-dataset-folder> --save_infos --all_configs --cache_dir <my-cache-dir>
```
it is not used for caching the modules, which are cached in the default location at `.cache/huggingface/modules`.
With this fix, the modules will be cached at `<my-cache-dir>/modules`.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1369/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1369/timeline
| null | null | true
|
https://api.github.com/repos/huggingface/datasets/issues/1324
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/1324/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/1324/comments
|
https://api.github.com/repos/huggingface/datasets/issues/1324/events
|
https://github.com/huggingface/datasets/issues/1324
| 759,587,864
|
MDU6SXNzdWU3NTk1ODc4NjQ=
| 1,324
|
❓ Sharing ElasticSearch indexed dataset
|
{
"login": "pietrolesci",
"id": 61748653,
"node_id": "MDQ6VXNlcjYxNzQ4NjUz",
"avatar_url": "https://avatars.githubusercontent.com/u/61748653?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/pietrolesci",
"html_url": "https://github.com/pietrolesci",
"followers_url": "https://api.github.com/users/pietrolesci/followers",
"following_url": "https://api.github.com/users/pietrolesci/following{/other_user}",
"gists_url": "https://api.github.com/users/pietrolesci/gists{/gist_id}",
"starred_url": "https://api.github.com/users/pietrolesci/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/pietrolesci/subscriptions",
"organizations_url": "https://api.github.com/users/pietrolesci/orgs",
"repos_url": "https://api.github.com/users/pietrolesci/repos",
"events_url": "https://api.github.com/users/pietrolesci/events{/privacy}",
"received_events_url": "https://api.github.com/users/pietrolesci/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[
"Hello @pietrolesci , I am not sure to understand what you are trying to do here.\r\n\r\nIf you're looking for ways to save a dataset on disk, you can you the `save_to_disk` method:\r\n```python\r\n>>> import datasets\r\n>>> loaded_dataset = datasets.load(\"dataset_name\")\r\n>>> loaded_dataset.save_to_disk(\"/path/on/your/disk\")\r\n```\r\n\r\nThe saved dataset can later be retrieved using:\r\n```python\r\n>>> loaded_dataset = datasets.Dataset.load_from_disk(\"/path/on/your/disk\")\r\n```\r\n\r\nAlso, I'd recommend posting your question directly in the issue section of the [elasticsearch repo](https://github.com/elastic/elasticsearch)",
"Hi @SBrandeis,\n\nThanks a lot for picking up my request. \n\nMaybe I can clarify my use-case with a bit of context. Say I have the IMDb dataset. I create an ES index on it. Now I can save and reload the dataset from disk normally. Once I reload the dataset, it is easy to retrieve the ES index on my machine. I was wondering: is there a way I can share the (now) indexed version of the IMDb dataset with my colleagues without requiring them to re-index it?\n\nThanks a lot in advance for your consideration.\n\nBest,\n\nPietro",
"Thanks for the clarification.\r\n\r\nI am not familiar with ElasticSearch, but if I understand well you're trying to migrate your data along with the ES index.\r\nMy advice would be to check out ES documentation, for instance, this might help you: https://www.elastic.co/guide/en/cloud/current/ec-migrate-data.html\r\n\r\nLet me know if it helps"
] | 2020-12-08T16:25:58
| 2020-12-22T07:50:56
| null |
NONE
| null | null | null |
Hi there,
First of all, thank you very much for this amazing library. Datasets have become my preferred data structure for basically everything I am currently doing.
**Question:** I'm working with a dataset and I have an elasticsearch container running at localhost:9200. I added an elasticsearch index and I was wondering
- how can I know where it has been saved?
- how can I share the indexed dataset with others?
I tried to dig into the docs, but could not find anything about that.
Thank you very much for your help.
Best,
Pietro
Edit: apologies for the wrong label
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/1324/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/1324/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/915
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/915/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/915/comments
|
https://api.github.com/repos/huggingface/datasets/issues/915/events
|
https://github.com/huggingface/datasets/issues/915
| 753,118,481
|
MDU6SXNzdWU3NTMxMTg0ODE=
| 915
|
Shall we change the hashing to encoding to reduce potential replicated cache files?
|
{
"login": "zhuzilin",
"id": 10428324,
"node_id": "MDQ6VXNlcjEwNDI4MzI0",
"avatar_url": "https://avatars.githubusercontent.com/u/10428324?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/zhuzilin",
"html_url": "https://github.com/zhuzilin",
"followers_url": "https://api.github.com/users/zhuzilin/followers",
"following_url": "https://api.github.com/users/zhuzilin/following{/other_user}",
"gists_url": "https://api.github.com/users/zhuzilin/gists{/gist_id}",
"starred_url": "https://api.github.com/users/zhuzilin/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/zhuzilin/subscriptions",
"organizations_url": "https://api.github.com/users/zhuzilin/orgs",
"repos_url": "https://api.github.com/users/zhuzilin/repos",
"events_url": "https://api.github.com/users/zhuzilin/events{/privacy}",
"received_events_url": "https://api.github.com/users/zhuzilin/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 2067400324,
"node_id": "MDU6TGFiZWwyMDY3NDAwMzI0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/generic%20discussion",
"name": "generic discussion",
"color": "c5def5",
"default": false,
"description": "Generic discussion on the library"
}
] |
open
| false
| null |
[] | null |
[
"This is an interesting idea !\r\nDo you have ideas about how to approach the decoding and the normalization ?",
"@lhoestq\r\nI think we first need to save the transformation chain to a list in `self._fingerprint`. Then we can\r\n- decode all the current saved datasets to see if there is already one that is equivalent to the transformation we need now.\r\n- or, calculate all the possible hash value of the current chain for comparison so that we could continue to use hashing.\r\nIf we find one, we can adjust the list in `self._fingerprint` to it.\r\n\r\nAs for the transformation reordering rules, we can just start with some manual rules, like two sort on the same column should merge to one, filter and select can change orders.\r\n\r\nAnd for encoding and decoding, we can just manually specify `sort` is 0, `shuffling` is 2 and create a base-n number or use some general algorithm like `base64.urlsafe_b64encode`.\r\n\r\nBecause we are not doing lazy evaluation now, we may not be able to normalize the transformation to its minimal form. If we want to support that, we can provde a `Sequential` api and let user input a list or transformation, so that user would not use the intermediate datasets. This would look like tf.data.Dataset."
] | 2020-11-30T03:50:46
| 2020-12-24T05:11:49
| null |
NONE
| null | null | null |
Hi there. For now, we are using `xxhash` to hash the transformations to fingerprint and we will save a copy of the processed dataset to disk if there is a new hash value. However, there are some transformations that are idempotent or commutative to each other. I think that encoding the transformation chain as the fingerprint may help in those cases, for example, use `base64.urlsafe_b64encode`. In this way, before we want to save a new copy, we can decode the transformation chain and normalize it to prevent omit potential reuse. As the main targets of this project are the really large datasets that cannot be loaded entirely in memory, I believe it would save a lot of time if we can avoid some write.
If you have interest in this, I'd love to help :).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/915/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/915/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/887
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/887/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/887/comments
|
https://api.github.com/repos/huggingface/datasets/issues/887/events
|
https://github.com/huggingface/datasets/issues/887
| 750,868,831
|
MDU6SXNzdWU3NTA4Njg4MzE=
| 887
|
pyarrow.lib.ArrowNotImplementedError: MakeBuilder: cannot construct builder for type extension<arrow.py_extension_type>
|
{
"login": "AmitMY",
"id": 5757359,
"node_id": "MDQ6VXNlcjU3NTczNTk=",
"avatar_url": "https://avatars.githubusercontent.com/u/5757359?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/AmitMY",
"html_url": "https://github.com/AmitMY",
"followers_url": "https://api.github.com/users/AmitMY/followers",
"following_url": "https://api.github.com/users/AmitMY/following{/other_user}",
"gists_url": "https://api.github.com/users/AmitMY/gists{/gist_id}",
"starred_url": "https://api.github.com/users/AmitMY/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/AmitMY/subscriptions",
"organizations_url": "https://api.github.com/users/AmitMY/orgs",
"repos_url": "https://api.github.com/users/AmitMY/repos",
"events_url": "https://api.github.com/users/AmitMY/events{/privacy}",
"received_events_url": "https://api.github.com/users/AmitMY/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892857,
"node_id": "MDU6TGFiZWwxOTM1ODkyODU3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/bug",
"name": "bug",
"color": "d73a4a",
"default": true,
"description": "Something isn't working"
}
] |
open
| false
| null |
[] | null |
[
"Yes right now `ArrayXD` can only be used as a column feature type, not a subtype.\r\nWith the current Arrow limitations I don't think we'll be able to make it work as a subtype, however it should be possible to allow dimensions of dynamic sizes (`Array3D(shape=(None, 137, 2), dtype=\"float32\")` for example since the [underlying arrow type](https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L236) allows dynamic sizes.\r\n\r\nFor now I'd suggest the use of nested `Sequence` types. Once we have the dynamic sizes you can update the dataset.\r\nWhat do you think ?",
"> Yes right now ArrayXD can only be used as a column feature type, not a subtype. \r\n\r\nMeaning it can't be nested under `Sequence`?\r\nIf so, for now I'll just make it a python list and make it with the nested `Sequence` type you suggested.",
"Yea unfortunately..\r\nThat's a current limitation with Arrow ExtensionTypes that can't be used in the default Arrow Array objects.\r\nWe already have an ExtensionArray that allows us to use them as column types but not for subtypes.\r\nMaybe we can extend it, I haven't experimented with that yet",
"Cool\r\nSo please consider this issue as a feature request for:\r\n```\r\nArray3D(shape=(None, 137, 2), dtype=\"float32\")\r\n```\r\n\r\nits a way to represent videos, poses, and other cool sequences",
"@lhoestq well, so sequence of sequences doesn't work either...\r\n\r\n```\r\npyarrow.lib.ArrowCapacityError: List array cannot contain more than 2147483646 child elements, have 2147483648\r\n```\r\n\r\n\r\n",
"Working with Arrow can be quite fun sometimes.\r\nYou can fix this issue by trying to reduce the writer batch size (same trick than the one used to reduce the RAM usage in https://github.com/huggingface/datasets/issues/741).\r\n\r\nLet me know if it works.\r\nI haven't investigated yet on https://github.com/huggingface/datasets/issues/741 since I was preparing this week's sprint to add datasets but this is in my priority list for early next week.",
"The batch size fix doesn't work... not for #741 and not for this dataset I'm trying (DGS corpus)\r\nLoading the DGS corpus takes 400GB of RAM, which is fine with me as my machine is large enough\r\n",
"Sorry it doesn't work. Will let you know once I fixed it",
"Hi @lhoestq , any update on dynamic sized arrays?\r\n(`Array3D(shape=(None, 137, 2), dtype=\"float32\")`)",
"Not yet, I've been pretty busy with the dataset sprint lately but this is something that's been asked several times already. So I'll definitely work on this as soon as I'm done with the sprint and with the RAM issue you reported.",
"Hi @lhoestq,\r\nAny chance you have some updates on the supporting `ArrayXD` as a subtype or support of dynamic sized arrays?\r\n\r\ne.g.:\r\n`datasets.features.Sequence(datasets.features.Array2D(shape=(137, 2), dtype=\"float32\"))`\r\n`Array3D(shape=(None, 137, 2), dtype=\"float32\")`",
"Hi ! We haven't worked in this lately and it's not in our very short-term roadmap since it requires a bit a work to make it work with arrow. Though this will definitely be added at one point.",
"@lhoestq, thanks for the update.\r\n\r\nI actually tried to modify some piece of code to make it work. Can you please tell if I missing anything here?\r\nI think that for vast majority of cases it's enough to make first dimension of the array dynamic i.e. `shape=(None, 100, 100)`. For that, it's enough to modify class [ArrayExtensionArray](https://github.com/huggingface/datasets/blob/9ca24250ea44e7611c4dabd01ecf9415a7f0be6c/src/datasets/features.py#L397) to output list of arrays of different sizes instead of list of arrays of same sizes (current version)\r\nBelow are my modifications of this class.\r\n\r\n```\r\nclass ArrayExtensionArray(pa.ExtensionArray):\r\n def __array__(self):\r\n zero_copy_only = _is_zero_copy_only(self.storage.type)\r\n return self.to_numpy(zero_copy_only=zero_copy_only)\r\n\r\n def __getitem__(self, i):\r\n return self.storage[i]\r\n\r\n def to_numpy(self, zero_copy_only=True):\r\n storage: pa.ListArray = self.storage\r\n size = 1\r\n for i in range(self.type.ndims):\r\n size *= self.type.shape[i]\r\n storage = storage.flatten()\r\n numpy_arr = storage.to_numpy(zero_copy_only=zero_copy_only)\r\n numpy_arr = numpy_arr.reshape(len(self), *self.type.shape)\r\n return numpy_arr\r\n\r\n def to_list_of_numpy(self, zero_copy_only=True):\r\n storage: pa.ListArray = self.storage\r\n shape = self.type.shape\r\n arrays = []\r\n for dim in range(1, self.type.ndims):\r\n assert shape[dim] is not None, f\"Support only dynamic size on first dimension. Got: {shape}\"\r\n\r\n first_dim_offsets = np.array([off.as_py() for off in storage.offsets])\r\n for i in range(len(storage)):\r\n storage_el = storage[i:i+1]\r\n first_dim = first_dim_offsets[i+1] - first_dim_offsets[i]\r\n # flatten storage\r\n for dim in range(self.type.ndims):\r\n storage_el = storage_el.flatten()\r\n\r\n numpy_arr = storage_el.to_numpy(zero_copy_only=zero_copy_only)\r\n arrays.append(numpy_arr.reshape(first_dim, *shape[1:]))\r\n\r\n return arrays\r\n\r\n def to_pylist(self):\r\n zero_copy_only = _is_zero_copy_only(self.storage.type)\r\n if self.type.shape[0] is None:\r\n return self.to_list_of_numpy(zero_copy_only=zero_copy_only)\r\n else:\r\n return self.to_numpy(zero_copy_only=zero_copy_only).tolist()\r\n```\r\n\r\nI ran few tests and it works as expected. Let me know what you think.",
"Thanks for diving into this !\r\n\r\nIndeed focusing on making the first dimensions dynamic make total sense (and users could still re-order their dimensions to match this constraint).\r\nYour code looks great :) I think it can even be extended to support several dynamic dimensions if we want to.\r\n\r\nFeel free to open a PR to include these changes, then we can update our test suite to make sure it works in all use cases.\r\nIn particular I think we might need a few tweaks to allow it to be converted to pandas (though I haven't tested yet):\r\n\r\n```python\r\nfrom datasets import Dataset, Features, Array3D\r\n\r\n# this works\r\nmatrix = [[1, 0], [0, 1]]\r\nfeatures = Features({\"a\": Array3D(dtype=\"int32\", shape=(1, 2, 2))})\r\nd = Dataset.from_dict({\"a\": [[matrix], [matrix]]})\r\nprint(d.to_pandas())\r\n\r\n# this should work as well\r\nmatrix = [[1, 0], [0, 1]]\r\nfeatures = Features({\"a\": Array3D(dtype=\"int32\", shape=(None, 2, 2))})\r\nd = Dataset.from_dict({\"a\": [[matrix], [matrix] * 2]})\r\nprint(d.to_pandas())\r\n```\r\n\r\nI'll be happy to help you on this :)"
] | 2020-11-25T14:32:21
| 2021-09-09T17:03:40
| null |
CONTRIBUTOR
| null | null | null |
I set up a new dataset, with a sequence of arrays (really, I want to have an array of (None, 137, 2), and the first dimension is dynamic)
```python
def _info(self):
return datasets.DatasetInfo(
description=_DESCRIPTION,
# This defines the different columns of the dataset and their types
features=datasets.Features(
{
"pose": datasets.features.Sequence(datasets.features.Array2D(shape=(137, 2), dtype="float32"))
}
),
homepage=_HOMEPAGE,
citation=_CITATION,
)
def _generate_examples(self):
""" Yields examples. """
yield 1, {
"pose": [np.zeros(shape=(137, 2), dtype=np.float32)]
}
```
But this doesn't work -
> pyarrow.lib.ArrowNotImplementedError: MakeBuilder: cannot construct builder for type extension<arrow.py_extension_type>
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/887/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/887/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/883
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/883/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/883/comments
|
https://api.github.com/repos/huggingface/datasets/issues/883/events
|
https://github.com/huggingface/datasets/issues/883
| 749,750,801
|
MDU6SXNzdWU3NDk3NTA4MDE=
| 883
|
Downloading/caching only a part of a datasets' dataset.
|
{
"login": "SapirWeissbuch",
"id": 44585792,
"node_id": "MDQ6VXNlcjQ0NTg1Nzky",
"avatar_url": "https://avatars.githubusercontent.com/u/44585792?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/SapirWeissbuch",
"html_url": "https://github.com/SapirWeissbuch",
"followers_url": "https://api.github.com/users/SapirWeissbuch/followers",
"following_url": "https://api.github.com/users/SapirWeissbuch/following{/other_user}",
"gists_url": "https://api.github.com/users/SapirWeissbuch/gists{/gist_id}",
"starred_url": "https://api.github.com/users/SapirWeissbuch/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/SapirWeissbuch/subscriptions",
"organizations_url": "https://api.github.com/users/SapirWeissbuch/orgs",
"repos_url": "https://api.github.com/users/SapirWeissbuch/repos",
"events_url": "https://api.github.com/users/SapirWeissbuch/events{/privacy}",
"received_events_url": "https://api.github.com/users/SapirWeissbuch/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 1935892912,
"node_id": "MDU6TGFiZWwxOTM1ODkyOTEy",
"url": "https://api.github.com/repos/huggingface/datasets/labels/question",
"name": "question",
"color": "d876e3",
"default": true,
"description": "Further information is requested"
}
] |
open
| false
| null |
[] | null |
[
"Not at the moment but we could likely support this feature.",
"?",
"I think it would be a very helpful feature, because sometimes one only wants to evaluate models on the dev set, and the whole training data may be many times bigger.\r\nThis makes the task impossible with limited memory resources."
] | 2020-11-24T14:25:18
| 2020-11-27T13:51:55
| null |
NONE
| null | null | null |
Hi,
I want to use the validation data *only* (of natural question).
I don't want to have the whole dataset cached in my machine, just the dev set.
Is this possible? I can't find a way to do it in the docs.
Thank you,
Sapir
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/883/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/883/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/878
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/878/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/878/comments
|
https://api.github.com/repos/huggingface/datasets/issues/878/events
|
https://github.com/huggingface/datasets/issues/878
| 748,621,981
|
MDU6SXNzdWU3NDg2MjE5ODE=
| 878
|
Loading Data From S3 Path in Sagemaker
|
{
"login": "mahesh1amour",
"id": 42795522,
"node_id": "MDQ6VXNlcjQyNzk1NTIy",
"avatar_url": "https://avatars.githubusercontent.com/u/42795522?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/mahesh1amour",
"html_url": "https://github.com/mahesh1amour",
"followers_url": "https://api.github.com/users/mahesh1amour/followers",
"following_url": "https://api.github.com/users/mahesh1amour/following{/other_user}",
"gists_url": "https://api.github.com/users/mahesh1amour/gists{/gist_id}",
"starred_url": "https://api.github.com/users/mahesh1amour/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/mahesh1amour/subscriptions",
"organizations_url": "https://api.github.com/users/mahesh1amour/orgs",
"repos_url": "https://api.github.com/users/mahesh1amour/repos",
"events_url": "https://api.github.com/users/mahesh1amour/events{/privacy}",
"received_events_url": "https://api.github.com/users/mahesh1amour/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 1935892912,
"node_id": "MDU6TGFiZWwxOTM1ODkyOTEy",
"url": "https://api.github.com/repos/huggingface/datasets/labels/question",
"name": "question",
"color": "d876e3",
"default": true,
"description": "Further information is requested"
}
] |
open
| false
| null |
[] | null |
[
"This would be a neat feature",
"> neat feature\r\n\r\nI dint get these clearly, can you please elaborate like how to work on these ",
"It could maybe work almost out of the box just by using `cached_path` in the text/csv/json scripts, no?",
"Thanks thomwolf and julien-c\r\n\r\nI'm still confusion on what you guys said, \r\n\r\nI have solved the problem as follows:\r\n\r\n1. read the csv file using pandas from s3 \r\n2. Convert to dictionary key as column name and values as list column data\r\n3. convert it to Dataset using \r\n`from datasets import Dataset`\r\n`train_dataset = Dataset.from_dict(train_dict)`",
"We were brainstorming around your use-case.\r\n\r\nLet's keep the issue open for now, I think this is an interesting question to think about.",
"> We were brainstorming around your use-case.\r\n> \r\n> Let's keep the issue open for now, I think this is an interesting question to think about.\r\n\r\nSure thomwolf, Thanks for your concern ",
"I agree it would be cool to have that feature. Also that's good to know that pandas supports this.\r\nFor the moment I'd suggest to first download the files locally as thom suggested and then load the dataset by providing paths to the local files",
"Don't get\n",
"Any updates on this issue?\r\nI face a similar issue. I have many parquet files in S3 and I would like to train on them. \r\nTo be honest I even face issues with only getting the last layer embedding out of them.",
"Hi dorlavie, \r\nYou can find one solution that i have mentioned above, that can help you. \r\nAnd there is one more solution also which is downloading files locally\r\n",
"> Hi dorlavie,\r\n> You can find one solution that i have mentioned above, that can help you.\r\n> And there is one more solution also which is downloading files locally\r\n\r\nmahesh1amour, thanks for the fast reply\r\n\r\nUnfortunately, in my case I can not read with pandas. The dataset is too big (50GB). \r\nIn addition, due to security concerns I am not allowed to save the data locally",
"@dorlavie could use `boto3` to download the data to your local machine and then load it with `dataset`\r\n\r\nboto3 example [documentation](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/s3-example-download-file.html)\r\n```python\r\nimport boto3\r\n\r\ns3 = boto3.client('s3')\r\ns3.download_file('BUCKET_NAME', 'OBJECT_NAME', 'FILE_NAME')\r\n```\r\n\r\ndatasets example [documentation](https://huggingface.co/docs/datasets/loading_datasets.html)\r\n\r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', data_files=['my_file_1.csv', 'my_file_2.csv', 'my_file_3.csv'])\r\n```\r\n",
"Thanks @philschmid for the suggestion.\r\nAs I mentioned in the previous comment, due to security issues I can not save the data locally.\r\nI need to read it from S3 and process it directly.\r\n\r\nI guess that many other people try to train / fit those models on huge datasets (e.g entire Wiki), what is the best practice in those cases?",
"If I understand correctly you're not allowed to write data on disk that you downloaded from S3 for example ?\r\nOr is it the use of the `boto3` library that is not allowed in your case ?",
"@lhoestq yes you are correct.\r\nI am not allowed to save the \"raw text\" locally - The \"raw text\" must be saved only on S3.\r\nI am allowed to save the output of any model locally. \r\nIt doesn't matter how I do it boto3/pandas/pyarrow, it is forbidden",
"@dorlavie are you using sagemaker for training too? Then you could use S3 URI, for example `s3://my-bucket/my-training-data` and pass it within the `.fit()` function when you start the sagemaker training job. Sagemaker would then download the data from s3 into the training runtime and you could load it from disk\r\n\r\n**sagemaker start training job**\r\n```python\r\npytorch_estimator.fit({'train':'s3://my-bucket/my-training-data','eval':'s3://my-bucket/my-evaluation-data'})\r\n```\r\n\r\n**in the train.py script**\r\n```python\r\nfrom datasets import load_from_disk\r\n\r\ntrain_dataset = load_from_disk(os.environ['SM_CHANNEL_TRAIN'])\r\n```\r\n\r\nI have created an example of how to use transformers and datasets with sagemaker. \r\nhttps://github.com/philschmid/huggingface-sagemaker-example/tree/main/03_huggingface_sagemaker_trainer_with_data_from_s3\r\n\r\nThe example contains a jupyter notebook `sagemaker-example.ipynb` and an `src/` folder. The sagemaker-example is a jupyter notebook that is used to create the training job on AWS Sagemaker. The `src/` folder contains the `train.py`, our training script, and `requirements.txt` for additional dependencies.\r\n\r\n"
] | 2020-11-23T09:17:22
| 2020-12-23T09:53:08
| null |
NONE
| null | null | null |
In Sagemaker Im tring to load the data set from S3 path as follows
`train_path = 's3://xxxxxxxxxx/xxxxxxxxxx/train.csv'
valid_path = 's3://xxxxxxxxxx/xxxxxxxxxx/validation.csv'
test_path = 's3://xxxxxxxxxx/xxxxxxxxxx/test.csv'
data_files = {}
data_files["train"] = train_path
data_files["validation"] = valid_path
data_files["test"] = test_path
extension = train_path.split(".")[-1]
datasets = load_dataset(extension, data_files=data_files, s3_enabled=True)
print(datasets)`
I getting an error of
`algo-1-7plil_1 | File "main.py", line 21, in <module>
algo-1-7plil_1 | datasets = load_dataset(extension, data_files=data_files)
algo-1-7plil_1 | File "/opt/conda/lib/python3.6/site-packages/datasets/load.py", line 603, in load_dataset
algo-1-7plil_1 | **config_kwargs,
algo-1-7plil_1 | File "/opt/conda/lib/python3.6/site-packages/datasets/builder.py", line 155, in __init__
algo-1-7plil_1 | **config_kwargs,
algo-1-7plil_1 | File "/opt/conda/lib/python3.6/site-packages/datasets/builder.py", line 305, in _create_builder_config
algo-1-7plil_1 | m.update(str(os.path.getmtime(data_file)))
algo-1-7plil_1 | File "/opt/conda/lib/python3.6/genericpath.py", line 55, in getmtime
algo-1-7plil_1 | return os.stat(filename).st_mtime
algo-1-7plil_1 | FileNotFoundError: [Errno 2] No such file or directory: 's3://lsmv-sagemaker/pubmedbert/test.csv`
But when im trying with pandas , it is able to load from S3
Does the datasets library support S3 path to load
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/878/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/878/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/842
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/842/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/842/comments
|
https://api.github.com/repos/huggingface/datasets/issues/842/events
|
https://github.com/huggingface/datasets/issues/842
| 741,208,428
|
MDU6SXNzdWU3NDEyMDg0Mjg=
| 842
|
How to enable `.map()` pre-processing pipelines to support multi-node parallelism?
|
{
"login": "shangw-nvidia",
"id": 66387198,
"node_id": "MDQ6VXNlcjY2Mzg3MTk4",
"avatar_url": "https://avatars.githubusercontent.com/u/66387198?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/shangw-nvidia",
"html_url": "https://github.com/shangw-nvidia",
"followers_url": "https://api.github.com/users/shangw-nvidia/followers",
"following_url": "https://api.github.com/users/shangw-nvidia/following{/other_user}",
"gists_url": "https://api.github.com/users/shangw-nvidia/gists{/gist_id}",
"starred_url": "https://api.github.com/users/shangw-nvidia/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/shangw-nvidia/subscriptions",
"organizations_url": "https://api.github.com/users/shangw-nvidia/orgs",
"repos_url": "https://api.github.com/users/shangw-nvidia/repos",
"events_url": "https://api.github.com/users/shangw-nvidia/events{/privacy}",
"received_events_url": "https://api.github.com/users/shangw-nvidia/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Right now multiprocessing only runs on single node.\r\n\r\nHowever it's probably possible to extend it to support multi nodes. Indeed we're using the `multiprocess` library from the `pathos` project to do multiprocessing in `datasets`, and `pathos` is made to support parallelism on several nodes. More info about pathos [on the pathos repo](https://github.com/uqfoundation/pathos).\r\n\r\nIf you're familiar with pathos or if you want to give it a try, it could be a nice addition to the library :)",
"Curious to hear if anything on that side changed or if you suggestions to do it changed @lhoestq :)\r\n\r\nFor our use-case, we are entering the regime where trading a few more instances to save a few days would be nice :)",
"Currently for multi-node setups we're mostly going towards a nice integration with Dask. But I wouldn't exclude exploring `pathos` more at one point"
] | 2020-11-12T02:04:38
| 2022-10-12T16:10:51
| null |
NONE
| null | null | null |
Hi,
Currently, multiprocessing can be enabled for the `.map()` stages on a single node. However, in the case of multi-node training, (since more than one node would be available) I'm wondering if it's possible to extend the parallel processing among nodes, instead of only 1 node running the `.map()` while the other node is waiting for it to finish?
Thanks!
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/842/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/842/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/839
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/839/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/839/comments
|
https://api.github.com/repos/huggingface/datasets/issues/839/events
|
https://github.com/huggingface/datasets/issues/839
| 740,355,270
|
MDU6SXNzdWU3NDAzNTUyNzA=
| 839
|
XSum dataset missing spaces between sentences
|
{
"login": "loganlebanoff",
"id": 10007282,
"node_id": "MDQ6VXNlcjEwMDA3Mjgy",
"avatar_url": "https://avatars.githubusercontent.com/u/10007282?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/loganlebanoff",
"html_url": "https://github.com/loganlebanoff",
"followers_url": "https://api.github.com/users/loganlebanoff/followers",
"following_url": "https://api.github.com/users/loganlebanoff/following{/other_user}",
"gists_url": "https://api.github.com/users/loganlebanoff/gists{/gist_id}",
"starred_url": "https://api.github.com/users/loganlebanoff/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/loganlebanoff/subscriptions",
"organizations_url": "https://api.github.com/users/loganlebanoff/orgs",
"repos_url": "https://api.github.com/users/loganlebanoff/repos",
"events_url": "https://api.github.com/users/loganlebanoff/events{/privacy}",
"received_events_url": "https://api.github.com/users/loganlebanoff/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2020-11-11T00:34:43
| 2020-11-11T00:34:43
| null |
NONE
| null | null | null |
I noticed that the XSum dataset has no space between sentences. This could lead to worse results for anyone training or testing on it. Here's an example (0th entry in the test set):
`The London trio are up for best UK act and best album, as well as getting two nominations in the best song category."We got told like this morning 'Oh I think you're nominated'", said Dappy."And I was like 'Oh yeah, which one?' And now we've got nominated for four awards. I mean, wow!"Bandmate Fazer added: "We thought it's best of us to come down and mingle with everyone and say hello to the cameras. And now we find we've got four nominations."The band have two shots at the best song prize, getting the nod for their Tynchy Stryder collaboration Number One, and single Strong Again.Their album Uncle B will also go up against records by the likes of Beyonce and Kanye West.N-Dubz picked up the best newcomer Mobo in 2007, but female member Tulisa said they wouldn't be too disappointed if they didn't win this time around."At the end of the day we're grateful to be where we are in our careers."If it don't happen then it don't happen - live to fight another day and keep on making albums and hits for the fans."Dappy also revealed they could be performing live several times on the night.The group will be doing Number One and also a possible rendition of the War Child single, I Got Soul.The charity song is a re-working of The Killers' All These Things That I've Done and is set to feature artists like Chipmunk, Ironik and Pixie Lott.This year's Mobos will be held outside of London for the first time, in Glasgow on 30 September.N-Dubz said they were looking forward to performing for their Scottish fans and boasted about their recent shows north of the border."We just done Edinburgh the other day," said Dappy."We smashed up an N-Dubz show over there. We done Aberdeen about three or four months ago - we smashed up that show over there! Everywhere we go we smash it up!"`
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/839/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/839/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/768
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/768/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/768/comments
|
https://api.github.com/repos/huggingface/datasets/issues/768/events
|
https://github.com/huggingface/datasets/issues/768
| 730,908,060
|
MDU6SXNzdWU3MzA5MDgwNjA=
| 768
|
Add a `lazy_map` method to `Dataset` and `DatasetDict`
|
{
"login": "sgugger",
"id": 35901082,
"node_id": "MDQ6VXNlcjM1OTAxMDgy",
"avatar_url": "https://avatars.githubusercontent.com/u/35901082?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sgugger",
"html_url": "https://github.com/sgugger",
"followers_url": "https://api.github.com/users/sgugger/followers",
"following_url": "https://api.github.com/users/sgugger/following{/other_user}",
"gists_url": "https://api.github.com/users/sgugger/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sgugger/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sgugger/subscriptions",
"organizations_url": "https://api.github.com/users/sgugger/orgs",
"repos_url": "https://api.github.com/users/sgugger/repos",
"events_url": "https://api.github.com/users/sgugger/events{/privacy}",
"received_events_url": "https://api.github.com/users/sgugger/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"This is cool! I think some aspects to think about and decide in terms of API are:\r\n- do we allow several methods (chained i guess)\r\n- how do we inspect the currently set method(s)\r\n- how do we control/reset them"
] | 2020-10-27T22:33:03
| 2020-10-28T08:58:13
| null |
CONTRIBUTOR
| null | null | null |
The library is great, but it would be even more awesome with a `lazy_map` method implemented on `Dataset` and `DatasetDict`. This would apply a function on a give item but when the item is requested. Two use cases:
1. load image on the fly
2. apply a random function and get different outputs at each epoch (like data augmentation or randomly masking a part of a sentence for BERT-like objectives).
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/768/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/768/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/767
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/767/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/767/comments
|
https://api.github.com/repos/huggingface/datasets/issues/767/events
|
https://github.com/huggingface/datasets/issues/767
| 730,771,610
|
MDU6SXNzdWU3MzA3NzE2MTA=
| 767
|
Add option for named splits when using ds.train_test_split
|
{
"login": "nateraw",
"id": 32437151,
"node_id": "MDQ6VXNlcjMyNDM3MTUx",
"avatar_url": "https://avatars.githubusercontent.com/u/32437151?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/nateraw",
"html_url": "https://github.com/nateraw",
"followers_url": "https://api.github.com/users/nateraw/followers",
"following_url": "https://api.github.com/users/nateraw/following{/other_user}",
"gists_url": "https://api.github.com/users/nateraw/gists{/gist_id}",
"starred_url": "https://api.github.com/users/nateraw/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/nateraw/subscriptions",
"organizations_url": "https://api.github.com/users/nateraw/orgs",
"repos_url": "https://api.github.com/users/nateraw/repos",
"events_url": "https://api.github.com/users/nateraw/events{/privacy}",
"received_events_url": "https://api.github.com/users/nateraw/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
}
] |
open
| false
| null |
[] | null |
[
"Yes definitely we should give more flexibility to control the name of the splits outputted by `train_test_split`.\r\n\r\nRelated is the very interesting feedback from @bramvanroy on how we should improve this method: https://discuss.huggingface.co/t/how-to-split-main-dataset-into-train-dev-test-as-datasetdict/1090/5\r\n\r\nAnd in particular that it should advantageously be able to split in 3 splits as well instead of just 2 like we copied from sklearn."
] | 2020-10-27T19:59:44
| 2020-11-10T14:05:21
| null |
CONTRIBUTOR
| null | null | null |
### Feature Request 🚀
Can we add a way to name your splits when using the `.train_test_split` function?
In almost every use case I've come across, I have a `train` and a `test` split in my `DatasetDict`, and I want to create a `validation` split. Therefore, its kinda useless to get a `test` split back from `train_test_split`, as it'll just overwrite my real `test` split that I intended to keep.
### Workaround
this is my hack for dealin with this, for now :slightly_smiling_face:
```python
from datasets import load_dataset
ds = load_dataset('imdb')
ds['train'], ds['validation'] = ds['train'].train_test_split(.1).values()
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/767/reactions",
"total_count": 5,
"+1": 5,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/767/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/743
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/743/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/743/comments
|
https://api.github.com/repos/huggingface/datasets/issues/743/events
|
https://github.com/huggingface/datasets/issues/743
| 724,703,980
|
MDU6SXNzdWU3MjQ3MDM5ODA=
| 743
|
load_dataset for CSV files not working
|
{
"login": "iliemihai",
"id": 2815308,
"node_id": "MDQ6VXNlcjI4MTUzMDg=",
"avatar_url": "https://avatars.githubusercontent.com/u/2815308?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/iliemihai",
"html_url": "https://github.com/iliemihai",
"followers_url": "https://api.github.com/users/iliemihai/followers",
"following_url": "https://api.github.com/users/iliemihai/following{/other_user}",
"gists_url": "https://api.github.com/users/iliemihai/gists{/gist_id}",
"starred_url": "https://api.github.com/users/iliemihai/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/iliemihai/subscriptions",
"organizations_url": "https://api.github.com/users/iliemihai/orgs",
"repos_url": "https://api.github.com/users/iliemihai/repos",
"events_url": "https://api.github.com/users/iliemihai/events{/privacy}",
"received_events_url": "https://api.github.com/users/iliemihai/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Thank you !\r\nCould you provide a csv file that reproduces the error ?\r\nIt doesn't have to be one of your dataset. As long as it reproduces the error\r\nThat would help a lot !",
"I think another good example is the following:\r\n`\r\nfrom datasets import load_dataset\r\n`\r\n`\r\ndataset = load_dataset(\"csv\", data_files=[\"./sts-dev.csv\"], delimiter=\"\\t\", column_names=[\"one\", \"two\", \"three\", \"four\", \"score\", \"sentence1\", \"sentence2\"], script_version=\"master\")`\r\n`\r\n\r\nDisplayed error `CSV parse error: Expected 7 columns, got 6` even tough I put 7 columns. First four columns from the csv don't have a name, so I've named them by default. The csv file is the .dev file from STSb benchmark dataset.\r\n\r\n",
"Hi, seems I also can't read csv file. I was trying with a dummy csv with only three rows.\r\n\r\n```\r\ntext,label\r\nI hate google,negative\r\nI love Microsoft,positive\r\nI don't like you,negative\r\n```\r\nI was using the HuggingFace image in Paperspace Gradient (datasets==1.1.3). The following code doesn't work:\r\n\r\n```\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', script_version=\"master\", data_files=['test_data.csv'], delimiter=\",\")\r\n```\r\nIt outputs the following:\r\n```\r\nUsing custom data configuration default\r\nDownloading and preparing dataset csv/default-3b6254ff4dd403e5 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /root/.cache/huggingface/datasets/csv/default-3b6254ff4dd403e5/0.0.0/2960f95a26e85d40ca41a230ac88787f715ee3003edaacb8b1f0891e9f04dda2...\r\nDataset csv downloaded and prepared to /root/.cache/huggingface/datasets/csv/default-3b6254ff4dd403e5/0.0.0/2960f95a26e85d40ca41a230ac88787f715ee3003edaacb8b1f0891e9f04dda2. Subsequent calls will reuse this data.\r\n```\r\nBut `len(dataset)` gives `1` and I can't access rows with indexing `dataset[0]` (it gives `KeyError: 0`).\r\n\r\nHowever, loading from pandas dataframe is working.\r\n```\r\nfrom datasets import Dataset\r\nimport pandas as pd\r\ndf = pd.read_csv('test_data.csv')\r\ndataset = Dataset.from_pandas(df)\r\n```\r\n\r\n",
"This is because load_dataset without `split=` returns a dictionary of split names (train/validation/test) to dataset.\r\nYou can do\r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', script_version=\"master\", data_files=['test_data.csv'], delimiter=\",\")\r\nprint(dataset[\"train\"][0])\r\n```\r\n\r\nOr if you want to directly get the train split:\r\n\r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', script_version=\"master\", data_files=['test_data.csv'], delimiter=\",\", split=\"train\")\r\nprint(dataset[0])\r\n```\r\n",
"Good point\r\n\r\nDesign question for us, though: should `load_dataset` when no split is specified and only one split is present in the dataset (common use case with CSV/text/JSON datasets) return a `Dataset` instead of a `DatsetDict`? I feel like it's often what the user is expecting. I break a bit the paradigm of a unique return type but since this library is designed for widespread DS people more than CS people usage I would tend to think that UX should take precedence over CS reasons. What do you think?",
"In this case the user expects to get only one dataset object instead of the dictionary of datasets since only one csv file was specified without any split specifications.\r\nI'm ok with returning the dataset object if no split specifications are given for text/json/csv/pandas.\r\n\r\nFor the other datasets ton the other hand the user doesn't know in advance the splits so I would keep the dictionary by default. What do you think ?",
"Thanks for your quick response! I'm fine with specifying the split as @lhoestq suggested. My only concern is when I'm loading from python dict or pandas, the library returns a dataset instead of a dictionary of datasets when no split is specified. I know that they use a different function `Dataset.from_dict` or `Dataset.from_pandas` but the text/csv files use `load_dataset()`. However, to the user, they do the same task and we probably expect them to have the same behavior.",
"```\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', data_files='./amazon_data/Video_Games_5.csv', delimiter=\",\", split=['train', 'test'])\r\n```\r\nI was running the above line, but got this error.\r\n\r\n```ValueError: Unknown split \"test\". Should be one of ['train'].```\r\n\r\nThe data is amazon product data. I load the Video_Games_5.json.gz data into pandas and save it as csv file. and then load the csv file using the above code. I thought, ```split=['train', 'test']``` would split the data into train and test. did I misunderstood?\r\n\r\nThank you!\r\n\r\n",
"Hi ! the `split` argument in `load_dataset` is used to select the splits you want among the available splits.\r\nHowever when loading a csv with a single file as you did, only a `train` split is available by default.\r\n\r\nIndeed since `data_files='./amazon_data/Video_Games_5.csv'` is equivalent to `data_files={\"train\": './amazon_data/Video_Games_5.csv'}`, you can get a dataset with \r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', data_files='./amazon_data/Video_Games_5.csv', delimiter=\",\", split=\"train\")\r\n```\r\n\r\nAnd then to get both a train and test split you can do\r\n```python\r\ndataset = dataset.train_test_split()\r\nprint(dataset.keys())\r\n# ['train', 'test']\r\n```\r\n\r\n\r\nAlso note that a csv dataset may have several available splits if it is defined this way:\r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', data_files={\r\n \"train\": './amazon_data/Video_Games_5_train.csv',\r\n \"test\": './amazon_data/Video_Games_5_test.csv'\r\n})\r\nprint(dataset.keys())\r\n# ['train', 'test']\r\n```\r\n",
"> In this case the user expects to get only one dataset object instead of the dictionary of datasets since only one csv file was specified without any split specifications.\r\n> I'm ok with returning the dataset object if no split specifications are given for text/json/csv/pandas.\r\n> \r\n> For the other datasets ton the other hand the user doesn't know in advance the splits so I would keep the dictionary by default. What do you think ?\r\n\r\nYes maybe this would be good. I think having to select 'train' from the resulting object why the user gave no split information is a confusing and not intuitive behavior.",
"> Similar to #622, I've noticed there is a problem when trying to load a CSV file with datasets.\r\n> \r\n> `from datasets import load_dataset`\r\n> `dataset = load_dataset(\"csv\", data_files=[\"./sample_data.csv\"], delimiter=\"\\t\", column_names=[\"title\", \"text\"], script_version=\"master\")`\r\n> \r\n> Displayed error:\r\n> `... ArrowInvalid: CSV parse error: Expected 2 columns, got 1`\r\n\r\nI'm also facing the same issue when trying to load from a csv file locally:\r\n\r\n```python\r\nfrom nlp import load_dataset\r\ndataset = load_dataset('csv', data_files='sample_data.csv')\r\n```\r\n\r\nError when executed from Google Colab:\r\n```python\r\nArrowInvalid Traceback (most recent call last)\r\n<ipython-input-34-79a8d4f65ed6> in <module>()\r\n 1 from nlp import load_dataset\r\n----> 2 dataset = load_dataset('csv', data_files='sample_data.csv')\r\n\r\n9 frames\r\n/usr/local/lib/python3.7/dist-packages/nlp/load.py in load_dataset(path, name, version, data_dir, data_files, split, cache_dir, features, download_config, download_mode, ignore_verifications, save_infos, **config_kwargs)\r\n 547 # Download and prepare data\r\n 548 builder_instance.download_and_prepare(\r\n--> 549 download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,\r\n 550 )\r\n 551 \r\n\r\n/usr/local/lib/python3.7/dist-packages/nlp/builder.py in download_and_prepare(self, download_config, download_mode, ignore_verifications, try_from_hf_gcs, dl_manager, **download_and_prepare_kwargs)\r\n 461 if not downloaded_from_gcs:\r\n 462 self._download_and_prepare(\r\n--> 463 dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs\r\n 464 )\r\n 465 # Sync info\r\n\r\n/usr/local/lib/python3.7/dist-packages/nlp/builder.py in _download_and_prepare(self, dl_manager, verify_infos, **prepare_split_kwargs)\r\n 535 try:\r\n 536 # Prepare split will record examples associated to the split\r\n--> 537 self._prepare_split(split_generator, **prepare_split_kwargs)\r\n 538 except OSError:\r\n 539 raise OSError(\"Cannot find data file. \" + (self.manual_download_instructions or \"\"))\r\n\r\n/usr/local/lib/python3.7/dist-packages/nlp/builder.py in _prepare_split(self, split_generator)\r\n 863 \r\n 864 generator = self._generate_tables(**split_generator.gen_kwargs)\r\n--> 865 for key, table in utils.tqdm(generator, unit=\" tables\", leave=False):\r\n 866 writer.write_table(table)\r\n 867 num_examples, num_bytes = writer.finalize()\r\n\r\n/usr/local/lib/python3.7/dist-packages/tqdm/notebook.py in __iter__(self, *args, **kwargs)\r\n 213 def __iter__(self, *args, **kwargs):\r\n 214 try:\r\n--> 215 for obj in super(tqdm_notebook, self).__iter__(*args, **kwargs):\r\n 216 # return super(tqdm...) will not catch exception\r\n 217 yield obj\r\n\r\n/usr/local/lib/python3.7/dist-packages/tqdm/std.py in __iter__(self)\r\n 1102 fp_write=getattr(self.fp, 'write', sys.stderr.write))\r\n 1103 \r\n-> 1104 for obj in iterable:\r\n 1105 yield obj\r\n 1106 # Update and possibly print the progressbar.\r\n\r\n/usr/local/lib/python3.7/dist-packages/nlp/datasets/csv/ede98314803c971fef04bcee45d660c62f3332e8a74491e0b876106f3d99bd9b/csv.py in _generate_tables(self, files)\r\n 78 read_options=self.config.pa_read_options,\r\n 79 parse_options=self.config.pa_parse_options,\r\n---> 80 convert_options=self.config.convert_options,\r\n 81 )\r\n 82 yield i, pa_table\r\n\r\n/usr/local/lib/python3.7/dist-packages/pyarrow/_csv.pyx in pyarrow._csv.read_csv()\r\n\r\n/usr/local/lib/python3.7/dist-packages/pyarrow/error.pxi in pyarrow.lib.pyarrow_internal_check_status()\r\n\r\n/usr/local/lib/python3.7/dist-packages/pyarrow/error.pxi in pyarrow.lib.check_status()\r\n\r\nArrowInvalid: CSV parse error: Expected 1 columns, got 8\r\n```\r\n\r\nVersion:\r\n```\r\nnlp==0.4.0\r\n```",
"Hi @kauvinlucas\r\n\r\nYou can use the latest versions of `datasets` to do this.\r\nTo do so, just `pip install datasets` instead of `nlp` (the library was renamed) and then\r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset('csv', data_files='sample_data.csv')",
"Hi \r\nI'm having a different problem with loading local csv. \r\n```Python\r\nfrom datasets import load_dataset \r\ndataset = load_dataset('csv', data_files='sample.csv') \r\n``` \r\n\r\ngives `ValueError: Specified named and prefix; you can only specify one.` error \r\n\r\nversions: \r\n- datasets: 1.1.3 \r\n- python: 3.9.6 \r\n- pyarrow: 2.0.0 ",
"Oh.. I figured it out. According to issue #[42387](https://github.com/pandas-dev/pandas/issues/42387) from pandas, this new version does not accept None for both parameters (which was being done by the repo I'm testing). Dowgrading Pandas==1.0.4 and Python==3.8 worked",
"Hi, \r\nI got an `OSError: Cannot find data file. ` when I tried to use load_dataset with tsv files. I have checked the paths, and they are correct. \r\n\r\nversions\r\n- python: 3.7.9\r\n- datasets: 1.1.3\r\n- pyarrow: 2.0.0\r\n- transformers: 4.2.2\r\n\r\n~~~\r\ndata_files = {\"train\": \"train.tsv\", \"test\",: \"test.tsv\"}\r\ndatasets = load_dataset(\"csv\", data_files=data_files, delimiter=\"\\t\")\r\n~~~\r\n\r\nThe entire Error message is on below:\r\n\r\n```08/14/2021 16:55:44 - INFO - __main__ - load a local file for train: /project/media-framing/transformer4/data/0/val/label1.tsv\r\n08/14/2021 16:55:44 - INFO - __main__ - load a local file for test: /project/media-framing/transformer4/data/unlabel/test.tsv\r\nUsing custom data configuration default\r\nDownloading and preparing dataset csv/default-00a4200ae8507533 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /usr4/cs542sp/hey1/.cache/huggingface/datasets/csv/default-00a4200ae8507533/0.0.0/2960f95a26e85d40ca41a230ac88787f715ee3003edaacb8b1f0891e9f04dda2...\r\nTraceback (most recent call last):\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 592, in _download_and_prepare\r\n self._prepare_split(split_generator, **prepare_split_kwargs)\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 944, in _prepare_split\r\n num_examples, num_bytes = writer.finalize()\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/arrow_writer.py\", line 307, in finalize\r\n self.stream.close()\r\n File \"pyarrow/io.pxi\", line 132, in pyarrow.lib.NativeFile.close\r\n File \"pyarrow/error.pxi\", line 99, in pyarrow.lib.check_status\r\nOSError: error closing file\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nTraceback (most recent call last):\r\n File \"run_glue.py\", line 484, in <module>\r\n main()\r\n File \"run_glue.py\", line 243, in main\r\n datasets = load_dataset(\"csv\", data_files=data_files, delimiter=\"\\t\")\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/load.py\", line 610, in load_dataset\r\n ignore_verifications=ignore_verifications,\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 515, in download_and_prepare\r\n dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 594, in _download_and_prepare\r\n raise OSError(\"Cannot find data file. \" + (self.manual_download_instructions or \"\"))\r\nOSError: Cannot find data file. ```",
"Hi ! It looks like the error stacktrace doesn't match with your code snippet.\r\n\r\nWhat error do you get when running this ?\r\n```\r\ndata_files = {\"train\": \"train.tsv\", \"test\",: \"test.tsv\"}\r\ndatasets = load_dataset(\"csv\", data_files=data_files, delimiter=\"\\t\")\r\n```\r\ncan you check that both tsv files are in the same folder as the current working directory of your shell ?",
"Hi @lhoestq, Below is the entire error message after I move both tsv files to the same directory. It's the same with I got before.\r\n```\r\n/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/torch/cuda/__init__.py:52: UserWarning: CUDA initialization: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx (Triggered internally at /pytorch/c10/cuda/CUDAFunctions.cpp:100.)\r\n return torch._C._cuda_getDeviceCount() > 0\r\n08/29/2021 22:56:43 - WARNING - __main__ - Process rank: -1, device: cpu, n_gpu: 0distributed training: False, 16-bits training: False\r\n08/29/2021 22:56:43 - INFO - __main__ - Training/evaluation parameters TrainingArguments(output_dir=/projectnb/media-framing/pred_result/label1/, overwrite_output_dir=True, do_train=True, do_eval=False, do_predict=True, evaluation_strategy=EvaluationStrategy.NO, prediction_loss_only=False, per_device_train_batch_size=8, per_device_eval_batch_size=8, gradient_accumulation_steps=1, eval_accumulation_steps=None, learning_rate=5e-05, weight_decay=0.0, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=8.0, max_steps=-1, lr_scheduler_type=SchedulerType.LINEAR, warmup_steps=0, logging_dir=runs/Aug29_22-56-43_scc1, logging_first_step=False, logging_steps=500, save_steps=3000, save_total_limit=None, no_cuda=False, seed=42, fp16=False, fp16_opt_level=O1, fp16_backend=auto, local_rank=-1, tpu_num_cores=None, tpu_metrics_debug=False, debug=False, dataloader_drop_last=False, eval_steps=500, dataloader_num_workers=0, past_index=-1, run_name=/projectnb/media-framing/pred_result/label1/, disable_tqdm=False, remove_unused_columns=True, label_names=None, load_best_model_at_end=False, metric_for_best_model=None, greater_is_better=None, ignore_data_skip=False, sharded_ddp=False, deepspeed=None, label_smoothing_factor=0.0, adafactor=False, _n_gpu=0)\r\n08/29/2021 22:56:43 - INFO - __main__ - load a local file for train: /project/media-framing/transformer4/temp_train.tsv\r\n08/29/2021 22:56:43 - INFO - __main__ - load a local file for test: /project/media-framing/transformer4/temp_test.tsv\r\n08/29/2021 22:56:43 - WARNING - datasets.builder - Using custom data configuration default-df627c23ac0e98ec\r\nDownloading and preparing dataset csv/default (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /usr4/cs542sp/hey1/.cache/huggingface/datasets/csv/default-df627c23ac0e98ec/0.0.0/9144e0a4e8435090117cea53e6c7537173ef2304525df4a077c435d8ee7828ff...\r\nTraceback (most recent call last):\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 693, in _download_and_prepare\r\n self._prepare_split(split_generator, **prepare_split_kwargs)\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 1166, in _prepare_split\r\n num_examples, num_bytes = writer.finalize()\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/arrow_writer.py\", line 428, in finalize\r\n self.stream.close()\r\n File \"pyarrow/io.pxi\", line 132, in pyarrow.lib.NativeFile.close\r\n File \"pyarrow/error.pxi\", line 99, in pyarrow.lib.check_status\r\nOSError: error closing file\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nTraceback (most recent call last):\r\n File \"run_glue.py\", line 487, in <module>\r\n main()\r\n File \"run_glue.py\", line 244, in main\r\n datasets = load_dataset(\"csv\", data_files=data_files, delimiter=\"\\t\")\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/load.py\", line 852, in load_dataset\r\n use_auth_token=use_auth_token,\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 616, in download_and_prepare\r\n dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 699, in _download_and_prepare\r\n + str(e)\r\nOSError: Cannot find data file. \r\nOriginal error:\r\nerror closing file\r\n```",
"Hi !\r\nCan you try running this into a python shell directly ?\r\n\r\n```python\r\nimport os\r\nfrom datasets import load_dataset\r\n\r\ndata_files = {\"train\": \"train.tsv\", \"test\": \"test.tsv\"}\r\nassert all(os.path.isfile(data_file) for data_file in data_files.values()), \"Couln't find files\"\r\n\r\ndatasets = load_dataset(\"csv\", data_files=data_files, delimiter=\"\\t\")\r\nprint(\"success !\")\r\n```\r\n\r\nThis way all the code from `run_glue.py` doesn't interfere with our tests :)",
"Hi @lhoestq, \r\n\r\nBelow is what I got from terminal after I copied and run your code. I think the files themselves are good since there is no assertion error. \r\n\r\n```\r\nUsing custom data configuration default-df627c23ac0e98ec\r\nDownloading and preparing dataset csv/default (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /usr4/cs542sp/hey1/.cache/huggingface/datasets/csv/default-df627c23ac0e98ec/0.0.0/9144e0a4e8435090117cea53e6c7537173ef2304525df4a077c435d8ee7828ff...\r\nTraceback (most recent call last):\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 693, in _download_and_prepare\r\n self._prepare_split(split_generator, **prepare_split_kwargs)\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 1166, in _prepare_split\r\n num_examples, num_bytes = writer.finalize()\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/arrow_writer.py\", line 428, in finalize\r\n self.stream.close()\r\n File \"pyarrow/io.pxi\", line 132, in pyarrow.lib.NativeFile.close\r\n File \"pyarrow/error.pxi\", line 99, in pyarrow.lib.check_status\r\nOSError: error closing file\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nTraceback (most recent call last):\r\n File \"test.py\", line 7, in <module>\r\n datasets = load_dataset(\"csv\", data_files=data_files, delimiter=\"\\t\")\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/load.py\", line 852, in load_dataset\r\n use_auth_token=use_auth_token,\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 616, in download_and_prepare\r\n dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs\r\n File \"/projectnb2/media-framing/env-trans4/lib/python3.7/site-packages/datasets/builder.py\", line 699, in _download_and_prepare\r\n + str(e)\r\nOSError: Cannot find data file. \r\nOriginal error:\r\nerror closing file\r\n```",
"Hi, could this be a permission error ? I think it fails to close the arrow file that contains the data from your CSVs in the cache.\r\n\r\nBy default datasets are cached in `~/.cache/huggingface/datasets`, could you check that you have the right permissions ?\r\nYou can also try to change the cache directory by passing `cache_dir=\"path/to/my/cache/dir\"` to `load_dataset`.",
"Thank you!! @lhoestq\r\n\r\nFor some reason, I don't have the default path for datasets to cache, maybe because I work from a remote system. The issue solved after I pass the `cache_dir` argument to the function. Thank you very much!!",
"> Hi, could this be a permission error ? I think it fails to close the arrow file that contains the data from your CSVs in the cache.\r\n> \r\n> By default datasets are cached in `~/.cache/huggingface/datasets`, could you check that you have the right permissions ? You can also try to change the cache directory by passing `cache_dir=\"path/to/my/cache/dir\"` to `load_dataset`.\r\n\r\nThis is the exact solution I have been finding for the whole afternoon. Thanks a lot!\r\nI tried to do a training on a cluster computing system. The user's home directory is shared between nodes.\r\nIt always gets **stuck** at dataset loading...\r\nThe reason might be, the node (with GPU) can't read/write data in the default cache folder (in my home directory).\r\nAfter using an intermediate cache folder, this issue is resolved for me."
] | 2020-10-19T14:53:51
| 2022-11-28T16:59:36
| null |
CONTRIBUTOR
| null | null | null |
Similar to #622, I've noticed there is a problem when trying to load a CSV file with datasets.
`
from datasets import load_dataset
`
`
dataset = load_dataset("csv", data_files=["./sample_data.csv"], delimiter="\t", column_names=["title", "text"], script_version="master")
`
Displayed error:
`
...
ArrowInvalid: CSV parse error: Expected 2 columns, got 1
`
I should mention that when I've tried to read data from `https://github.com/lhoestq/transformers/tree/custom-dataset-in-rag-retriever/examples/rag/test_data/my_knowledge_dataset.csv` it worked without a problem. I've read that there might be some problems with /r character, so I've removed them from the custom dataset, but the problem still remains.
I've added a colab reproducing the bug, but unfortunately I cannot provide the dataset.
https://colab.research.google.com/drive/1Qzu7sC-frZVeniiWOwzoCe_UHZsrlxu8?usp=sharing
Are there any work around for it ?
Thank you
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/743/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/743/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/727
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/727/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/727/comments
|
https://api.github.com/repos/huggingface/datasets/issues/727/events
|
https://github.com/huggingface/datasets/issues/727
| 719,386,366
|
MDU6SXNzdWU3MTkzODYzNjY=
| 727
|
Parallel downloads progress bar flickers
|
{
"login": "lhoestq",
"id": 42851186,
"node_id": "MDQ6VXNlcjQyODUxMTg2",
"avatar_url": "https://avatars.githubusercontent.com/u/42851186?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/lhoestq",
"html_url": "https://github.com/lhoestq",
"followers_url": "https://api.github.com/users/lhoestq/followers",
"following_url": "https://api.github.com/users/lhoestq/following{/other_user}",
"gists_url": "https://api.github.com/users/lhoestq/gists{/gist_id}",
"starred_url": "https://api.github.com/users/lhoestq/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/lhoestq/subscriptions",
"organizations_url": "https://api.github.com/users/lhoestq/orgs",
"repos_url": "https://api.github.com/users/lhoestq/repos",
"events_url": "https://api.github.com/users/lhoestq/events{/privacy}",
"received_events_url": "https://api.github.com/users/lhoestq/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[] | 2020-10-12T13:36:05
| 2020-10-12T13:36:05
| null |
MEMBER
| null | null | null |
When there are parallel downloads using the download manager, the tqdm progress bar flickers since all the progress bars are on the same line.
To fix that we could simply specify `position=i` for i=0 to n the number of files to download when instantiating the tqdm progress bar.
Another way would be to have one "master" progress bar that tracks the number of finished downloads, and then one progress bar per process that show the current downloads.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/727/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/727/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/651
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/651/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/651/comments
|
https://api.github.com/repos/huggingface/datasets/issues/651/events
|
https://github.com/huggingface/datasets/issues/651
| 705,212,034
|
MDU6SXNzdWU3MDUyMTIwMzQ=
| 651
|
Problem with JSON dataset format
|
{
"login": "vikigenius",
"id": 12724810,
"node_id": "MDQ6VXNlcjEyNzI0ODEw",
"avatar_url": "https://avatars.githubusercontent.com/u/12724810?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/vikigenius",
"html_url": "https://github.com/vikigenius",
"followers_url": "https://api.github.com/users/vikigenius/followers",
"following_url": "https://api.github.com/users/vikigenius/following{/other_user}",
"gists_url": "https://api.github.com/users/vikigenius/gists{/gist_id}",
"starred_url": "https://api.github.com/users/vikigenius/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/vikigenius/subscriptions",
"organizations_url": "https://api.github.com/users/vikigenius/orgs",
"repos_url": "https://api.github.com/users/vikigenius/repos",
"events_url": "https://api.github.com/users/vikigenius/events{/privacy}",
"received_events_url": "https://api.github.com/users/vikigenius/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Currently the `json` dataset doesn't support this format unfortunately.\r\nHowever you could load it with\r\n```python\r\nfrom datasets import Dataset\r\nimport pandas as pd\r\n\r\ndf = pd.read_json(\"path_to_local.json\", orient=\"index\")\r\ndataset = Dataset.from_pandas(df)\r\n```",
"or you can make a custom dataset script as explained in doc here: https://huggingface.co/docs/datasets/add_dataset.html"
] | 2020-09-20T23:57:14
| 2020-09-21T12:14:24
| null |
NONE
| null | null | null |
I have a local json dataset with the following form.
{
'id01234': {'key1': value1, 'key2': value2, 'key3': value3},
'id01235': {'key1': value1, 'key2': value2, 'key3': value3},
.
.
.
'id09999': {'key1': value1, 'key2': value2, 'key3': value3}
}
Note that instead of a list of records it's basically a dictionary of key value pairs with the keys being the record_ids and the values being the corresponding record.
Reading this with json:
```
data = datasets.load('json', data_files='path_to_local.json')
```
Throws an error and asks me to chose a field. What's the right way to handle this?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/651/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/651/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/633
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/633/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/633/comments
|
https://api.github.com/repos/huggingface/datasets/issues/633/events
|
https://github.com/huggingface/datasets/issues/633
| 702,440,484
|
MDU6SXNzdWU3MDI0NDA0ODQ=
| 633
|
Load large text file for LM pre-training resulting in OOM
|
{
"login": "leethu2012",
"id": 29704017,
"node_id": "MDQ6VXNlcjI5NzA0MDE3",
"avatar_url": "https://avatars.githubusercontent.com/u/29704017?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/leethu2012",
"html_url": "https://github.com/leethu2012",
"followers_url": "https://api.github.com/users/leethu2012/followers",
"following_url": "https://api.github.com/users/leethu2012/following{/other_user}",
"gists_url": "https://api.github.com/users/leethu2012/gists{/gist_id}",
"starred_url": "https://api.github.com/users/leethu2012/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/leethu2012/subscriptions",
"organizations_url": "https://api.github.com/users/leethu2012/orgs",
"repos_url": "https://api.github.com/users/leethu2012/repos",
"events_url": "https://api.github.com/users/leethu2012/events{/privacy}",
"received_events_url": "https://api.github.com/users/leethu2012/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"Not sure what could cause that on the `datasets` side. Could this be a `Trainer` issue ? cc @julien-c @sgugger ?",
"There was a memory leak issue fixed recently in master. You should install from source and see if it fixes your problem.",
"@lhoestq @sgugger Thanks for your comments. I have install from source code as you told, but the problem is still there.\r\nTo reproduce the issue, just replace [these lines](https://github.com/huggingface/transformers/blob/master/examples/language-modeling/run_language_modeling.py#L241-L258) with: \r\n(load_dataset and DataCollatorForDatasetsLanguageModeling as [above mentioned](https://github.com/huggingface/datasets/issues/633#issue-702440484))\r\n```python\r\n dataset = load_dataset(\"bookcorpus\")\r\n dataset = dataset.train_test_split(test_size=0.1)\r\n train_dataset = dataset['train']\r\n eval_dataset = dataset['test'] if training_args.do_eval else None\r\n\r\n data_collator = DataCollatorForDatasetsLanguageModeling(\r\n tokenizer=tokenizer,\r\n mlm=data_args.mlm,\r\n mlm_probability=data_args.mlm_probability,\r\n block_size=data_args.block_size\r\n )\r\n```\r\nand run by:\r\n```bash\r\npython run_language_modeling.py\r\n--output_dir=output \\\r\n--model_type=bert \\\r\n--model_name_or_path=bert-base-uncased \\\r\n--do_train \\\r\n--do_eval \\\r\n--mlm \r\n```",
"Same here. Pre-training on wikitext-103 to do some test. At the end of the training it takes 32GB of RAM + ~30GB of SWAP. I installed dataset==1.1.0, not built from source. I will try uninstalling and building from source when it finish.",
"This seems to be on the `transformers` library side.\r\n\r\nIf you have more informations (pip env) or even better, a colab reproducing the error we can investigate.",
"It seems like it's solved with freshed versions of transformers. I have tried to replicate the error doing a fresh pip install transformers & datasets on colab and the error doesn't continue. On colab it keeps stable on 5GB! (Y)\r\n\r\nEdit: **Thanks for your great work**. Have a good day.",
"@gaceladri witch version transformers and datasets are you using now? I want to try again. Thanks.",
"transformers==3.3.1\r\ndatasets==1.1.0\r\ntokenizers==0.8.1rc2\r\n",
"doing some modifications to mobilebert\r\nhttps://colab.research.google.com/drive/1ba09ZOpyHGAOQLcsxiQAHRXl10qnMU5o?usp=sharing ",
"It does not happen to me anymore. Can we close? @leethu2012 ",
"It's happening to me again. After 4 hours of pre-training, my ram memory gets full and the kernel dies. I am using the last transformers version as today. 4.4.0 and the last version of datasets 1.2.1, both installed from master. The memory consumption keeps increasing.",
"It looks like it is something from pytorch/python itself :face_with_head_bandage: https://github.com/pytorch/pytorch/issues/13246 ",
"Thanks for the investigation @gaceladri \r\n\r\nApparently this happens when `num_workers>0` and has to do with objects being copied-on-write.\r\nDid you try setting num_workers to 0 @gaceladri ?\r\nIf the issue doesn't happen with `num_workers=0` then this would confirm that it's indeed related to this python/pytorch issue.\r\n\r\nSince a `Dataset` object is a wrapper of a pyarrow Table, we should investigate if the data being copied comes from the Table itself or from metadata in the `Dataset` object. If it comes from the metadata in the `Dataset` object, we should be able to implement a workaround. But if it comes from the Table, we'll need to see with the pyarrow team what we can do... ",
"@lhoestq I have tried and it keeps increasing also with `dataloader_num_workers=0`",
"Hmmm so this might come from another issue...\r\nSince it doesn't seem to be related to multiprocessing it should be easier to investigate though.\r\nDo you have some ideas @gaceladri ?",
"@lhoestq I looked quickly to a previously spoted bug in my env wandb /sdk/interface/interface.py, because sometimes when I load the dataset I got a multiprocessing error at line 510 in wandb...interface.py\r\n\r\nThis bug is reported here https://github.com/huggingface/datasets/issues/847\r\n\r\n```\r\n---------------------------------------------------------------------------\r\nAssertionError Traceback (most recent call last)\r\n<timed eval> in <module>\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/transformers/trainer.py in train(self, model_path, trial)\r\n 877 print(len(epoch_iterator))\r\n 878 \r\n--> 879 for step, inputs in enumerate(epoch_iterator):\r\n 880 \r\n 881 start_step = time.time()\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/dataloader.py in __next__(self)\r\n 433 if self._sampler_iter is None:\r\n 434 self._reset()\r\n--> 435 data = self._next_data()\r\n 436 self._num_yielded += 1\r\n 437 if self._dataset_kind == _DatasetKind.Iterable and \\\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/dataloader.py in _next_data(self)\r\n 1083 else:\r\n 1084 del self._task_info[idx]\r\n-> 1085 return self._process_data(data)\r\n 1086 \r\n 1087 def _try_put_index(self):\r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/dataloader.py in _process_data(self, data)\r\n 1109 self._try_put_index()\r\n 1110 if isinstance(data, ExceptionWrapper):\r\n-> 1111 data.reraise()\r\n 1112 return data\r\n 1113 \r\n\r\n~/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/_utils.py in reraise(self)\r\n 426 # have message field\r\n 427 raise self.exc_type(message=msg)\r\n--> 428 raise self.exc_type(msg)\r\n 429 \r\n 430 \r\n\r\nAssertionError: Caught AssertionError in DataLoader worker process 0.\r\nOriginal Traceback (most recent call last):\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py\", line 198, in _worker_loop\r\n data = fetcher.fetch(index)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py\", line 44, in fetch\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py\", line 44, in <listcomp>\r\n data = [self.dataset[idx] for idx in possibly_batched_index]\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 1083, in __getitem__\r\n format_kwargs=self._format_kwargs,\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 1070, in _getitem\r\n format_kwargs=format_kwargs,\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 886, in _convert_outputs\r\n v = map_nested(command, v, **map_nested_kwargs)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/utils/py_utils.py\", line 216, in map_nested\r\n return function(data_struct)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/datasets/arrow_dataset.py\", line 847, in command\r\n return torch.tensor(x, **format_kwargs)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/warnings.py\", line 101, in _showwarnmsg\r\n _showwarnmsg_impl(msg)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/warnings.py\", line 30, in _showwarnmsg_impl\r\n file.write(text)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/lib/redirect.py\", line 100, in new_write\r\n cb(name, data)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/wandb_run.py\", line 729, in _console_callback\r\n self._backend.interface.publish_output(name, data)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/interface/interface.py\", line 186, in publish_output\r\n self._publish_output(o)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/interface/interface.py\", line 191, in _publish_output\r\n self._publish(rec)\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/site-packages/wandb/sdk/interface/interface.py\", line 510, in _publish\r\n if self._process and not self._process.is_alive():\r\n File \"/home/ad/anaconda3/envs/tfm/lib/python3.6/multiprocessing/process.py\", line 134, in is_alive\r\n assert self._parent_pid == os.getpid(), 'can only test a child process'\r\nAssertionError: can only test a child process\r\n```\r\n\r\nMy workaround was to just comment those lines without looking to much into consecuences:\r\n\r\n```\r\ndef _publish(self, record: pb.Record, local: bool = None) -> None:\r\n #if self._process and not self._process.is_alive():\r\n # raise Exception(\"The wandb backend process has shutdown\")\r\n```\r\n\r\nIt worked so far... I need to try running without wandb and see if it could be causing something wrong with multiprocessing. I am going to try to launch the training setting wandb to false and I will let you know again.",
"@lhoestq But despite this, I got lost into the [class Dataset()](https://huggingface.co/docs/datasets/_modules/datasets/arrow_dataset.html#Dataset) reading the pyarrow files.\r\n\r\nEdit: but you should be rigth, that it does not have to be related to multiprocessing since it keeps happening when `num_workers=0` ",
"Or maybe wandb uses multiprocessing ? One process for wandb logging and one for actual training ? If this is the case then even setting `num_workers=0` would cause the process to be forked for wandb and therefore cause the memory issue.",
"@lhoestq could be, but if we set wandb to false this should not happen. I am going to try.",
"@lhoestq It keeps happening. I have uninstalled wandb from my env, setted `%env WANDB_DISABLED=true` on my notebook, and commented this func:\r\n\r\n```\r\ndef get_available_reporting_integrations():\r\n integrations = []\r\n if is_azureml_available():\r\n integrations.append(\"azure_ml\")\r\n if is_comet_available():\r\n integrations.append(\"comet_ml\")\r\n if is_mlflow_available():\r\n integrations.append(\"mlflow\")\r\n if is_tensorboard_available():\r\n integrations.append(\"tensorboard\")\r\n # if is_wandb_available():\r\n # integrations.append(\"wandb\")\r\n return integrations\r\n```\r\nAs a fast test and it keeps increasing the ram memory. Wandb could not be the blameworthy here.",
"Thanks for checking @gaceladri . Let's investigate the single process setting then.\r\nIf you have some sort of colab notebook with a minimal code example that shows this behavior feel free to share it @gaceladri so that we can play around with it to find what causes this. Otherwise I'll probably try to reproduce on my side at one point",
"@lhoestq sure. Here you have https://colab.research.google.com/drive/1ba09ZOpyHGAOQLcsxiQAHRXl10qnMU5o?usp=sharing let me know if the link works and it reproduces the issue. To me, it reproduces the issue, since if you start the training the ram memory keeps increasing.\r\n\r\nLet me know. Thanks!",
"Could the bug be comming from tokenizers?\r\n\r\nI got this warning at the terminal from my jupyter notebook: \r\n```\r\nhuggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...\r\nTo disable this warning, you can either:\r\n\t- Avoid using `tokenizers` before the fork if possible\r\n\t- Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)\r\n```",
"I've never experienced memory issues with tokenizers so I don't know\r\nCc @n1t0 are you aware of any issue that would cause memory to keep increasing when the tokenizer is used in the Data Collator for language modeling ?",
"@lhoestq Thanks for pointing to n1t0, just to clarify. That warning was doing fine-tuning, without collator:\r\n```\r\n\r\nfrom datasets import load_dataset, load_metric\r\nimport numpy as np\r\n\r\nGLUE_TASKS = [\r\n \"cola\",\r\n \"mnli\",\r\n \"mnli-mm\",\r\n \"mrpc\",\r\n \"qnli\",\r\n \"qqp\",\r\n \"rte\",\r\n \"sst2\",\r\n \"stsb\",\r\n \"wnli\",\r\n]\r\ntask = \"mnli\"\r\nactual_task = \"mnli\" if task == \"mnli-mm\" else task\r\ndataset = load_dataset(\"glue\", actual_task)\r\nmetric = load_metric(\"glue\", actual_task)\r\nbatch_size = 16\r\nattention_type = \"linear\"\r\n\r\nfrom transformers.models.mobilebert_mod import (\r\n MobileBertForSequenceClassification,\r\n MobileBertTokenizerFast,\r\n)\r\nfrom transformers.models.mobilebert_mod.configuration_mobilebert import (\r\n MobileBertConfigMod,\r\n)\r\nfrom transformers import TrainingArguments, Trainer\r\n\r\nnum_labels = 3 if task.startswith(\"mnli\") else 1 if task == \"stsb\" else 2\r\ntokenizer = MobileBertTokenizerFast.from_pretrained(\r\n \"/media/ad/00b5422b-9d54-4449-8b5d-08eab5cdac8c/training_trfm/big_linear_layerdrop_shared/checkpoint-23000/\",\r\n max_len=512,\r\n)\r\nmodel = MobileBertForSequenceClassification.from_pretrained(\r\n \"/media/ad/00b5422b-9d54-4449-8b5d-08eab5cdac8c/training_trfm/big_linear_layerdrop_shared/checkpoint-23000/\",\r\n num_labels=num_labels,\r\n)\r\nprint(model.num_parameters())\r\n\r\ntask_to_keys = {\r\n \"cola\": (\"sentence\", None),\r\n \"mnli\": (\"premise\", \"hypothesis\"),\r\n \"mnli-mm\": (\"premise\", \"hypothesis\"),\r\n \"mrpc\": (\"sentence1\", \"sentence2\"),\r\n \"qnli\": (\"question\", \"sentence\"),\r\n \"qqp\": (\"question1\", \"question2\"),\r\n \"rte\": (\"sentence1\", \"sentence2\"),\r\n \"sst2\": (\"sentence\", None),\r\n \"stsb\": (\"sentence1\", \"sentence2\"),\r\n \"wnli\": (\"sentence1\", \"sentence2\"),\r\n}\r\n\r\nsentence1_key, sentence2_key = task_to_keys[task]\r\nif sentence2_key is None:\r\n print(f\"Sentence: {dataset['train'][0][sentence1_key]}\")\r\nelse:\r\n print(f\"Sentence 1: {dataset['train'][0][sentence1_key]}\")\r\n print(f\"Sentence 2: {dataset['train'][0][sentence2_key]}\")\r\n\r\n\r\ndef preprocess_function(examples):\r\n if sentence2_key is None:\r\n return tokenizer(examples[sentence1_key], truncation=True)\r\n return tokenizer(examples[sentence1_key], examples[sentence2_key], truncation=True)\r\n\r\n\r\nencoded_dataset = dataset.map(preprocess_function, batched=True)\r\nmetric_name = (\r\n \"pearson\"\r\n if task == \"stsb\"\r\n else \"matthews_correlation\"\r\n if task == \"cola\"\r\n else \"accuracy\"\r\n)\r\n\r\nargs = TrainingArguments(\r\n f\"test-glue/{task}_{attention_type}\",\r\n evaluation_strategy=\"steps\",\r\n learning_rate=1e-5,\r\n per_device_train_batch_size=batch_size,\r\n per_device_eval_batch_size=batch_size,\r\n logging_steps=200,\r\n num_train_epochs=5,\r\n gradient_accumulation_steps=1,\r\n warmup_steps=10000,\r\n fp16=True,\r\n dataloader_num_workers=10,\r\n weight_decay=0.1,\r\n load_best_model_at_end=True,\r\n metric_for_best_model=metric_name,\r\n)\r\n\r\n\r\ndef compute_metrics(eval_pred):\r\n predictions, labels = eval_pred\r\n if task != \"stsb\":\r\n predictions = np.argmax(predictions, axis=1)\r\n else:\r\n predictions = predictions[:, 0]\r\n return metric.compute(predictions=predictions, references=labels)\r\n\r\n\r\nvalidation_key = (\r\n \"validation_mismatched\"\r\n if task == \"mnli-mm\"\r\n else \"validation_matched\"\r\n if task == \"mnli\"\r\n else \"validation\"\r\n)\r\n\r\ntrainer = Trainer(\r\n model,\r\n args,\r\n train_dataset=encoded_dataset[\"train\"],\r\n eval_dataset=encoded_dataset[validation_key],\r\n tokenizer=tokenizer,\r\n compute_metrics=compute_metrics,\r\n)\r\n\r\ntrainer.train()\r\n```\r\n\r\nNow, I have come back to pre-training. The changes that I think I have done are: not formatting the dataset to torch: ~~`big_dataset.set_format(type='torch', columns=[\"text\", \"input_ids\", \"attention_mask\", \"token_type_ids\"])`~~ so maybe some column is dropped and not freezed in memory and now I have not setted any validation dataset in the trainer. \r\n\r\nMy validation dataset before:\r\n```\r\nbook_corpus_eval = load_dataset(\r\n \"bookcorpus\",\r\n \"plain_text\",\r\n cache_dir=\"/home/ad/Desktop/bookcorpus\",\r\n split=\"train[98:99%]\",\r\n)\r\nbook_corpus_eval = book_corpus_eval.map(encode, batched=True)\r\nbook_corpus_eval.set_format(\r\n type=\"torch\", columns=[\"text\", \"input_ids\", \"attention_mask\", \"token_type_ids\"]\r\n)\r\n**book_corpus_eval = book_corpus_eval.select([i for i in range(1500)])**\r\n```\r\nMaybe _selecting_ or indexing the dataset before feeding it to the trainer, do something strange.\r\n\r\nMy trainer now:\r\n```\r\n\r\nbig_dataset = load_from_disk(\"/home/ad/Desktop/35percent_data.arrow/\")\r\n\r\nfrom transformers import DataCollatorForWholeWordMask\r\n\r\ndata_collator = DataCollatorForWholeWordMask(\r\n tokenizer=tokenizer, mlm=True, mlm_probability=0.15)\r\n\r\nfrom transformers import Trainer, TrainingArguments\r\n\r\ntraining_args = TrainingArguments(\r\n output_dir=\"./big_linear_layerdrop_shared_silu_secondtry\",\r\n overwrite_output_dir=True,\r\n per_device_train_batch_size=60,\r\n per_device_eval_batch_size=60,\r\n save_steps=500,\r\n save_total_limit=10,\r\n logging_first_step=True,\r\n logging_steps=100,\r\n# evaluation_strategy='steps',\r\n# eval_steps=250,\r\n gradient_accumulation_steps=8,\r\n fp16=True,\r\n dataloader_num_workers=10,\r\n warmup_steps=15000,\r\n learning_rate=6e-4,\r\n adam_epsilon=1e-6,\r\n adam_beta2=0.98,\r\n weight_decay=0.01,\r\n max_grad_norm=1.0,\r\n max_steps=500000, \r\n)\r\n\r\ntrainer = Trainer(\r\n model=model,\r\n args=training_args,\r\n data_collator=data_collator,\r\n train_dataset=big_dataset,\r\n# eval_dataset=book_corpus_eval,\r\n tokenizer=tokenizer)\r\n\r\nimport wandb\r\nwandb.login()\r\n\r\ntrainer.train()\r\n```\r\n\r\nAnd surprisingly, the ram now keeps going up and down. The training is up now for 12h without collapse the ram. I don't know what could cause the leakage. :mag: \r\n\r\nEdit: I didn't see the swap memory, that keeps increasing. So the problem persist. ",
"Thanks for sharing your results.\r\nSo you still had the issue for fine-tuning ?\r\nAnd the issue still appears with a bare-bone dataset from an arrow file...",
"Yes, on both cases. Fine-tuning a pre-trained model and pre-training from scratch with a local arrow file already pre-processed."
] | 2020-09-16T04:33:15
| 2021-02-16T12:02:01
| null |
NONE
| null | null | null |
I tried to pretrain Longformer using transformers and datasets. But I got OOM issues with loading a large text file. My script is almost like this:
```python
from datasets import load_dataset
@dataclass
class DataCollatorForDatasetsLanguageModeling(DataCollatorForLanguageModeling):
"""
Data collator used for language modeling based on DataCollatorForLazyLanguageModeling
- collates batches of tensors, honoring their tokenizer's pad_token
- preprocesses batches for masked language modeling
"""
block_size: int = 512
def __call__(self, examples: List[dict]) -> Dict[str, torch.Tensor]:
examples = [example['text'] for example in examples]
batch, attention_mask = self._tensorize_batch(examples)
if self.mlm:
inputs, labels = self.mask_tokens(batch)
return {"input_ids": inputs, "labels": labels}
else:
labels = batch.clone().detach()
if self.tokenizer.pad_token_id is not None:
labels[labels == self.tokenizer.pad_token_id] = -100
return {"input_ids": batch, "labels": labels}
def _tensorize_batch(self, examples: List[str]) -> Tuple[torch.Tensor, torch.Tensor]:
if self.tokenizer._pad_token is None:
raise ValueError(
"You are attempting to pad samples but the tokenizer you are using"
f" ({self.tokenizer.__class__.__name__}) does not have one."
)
tensor_examples = self.tokenizer.batch_encode_plus(
[ex for ex in examples if ex],
max_length=self.block_size,
return_tensors="pt",
pad_to_max_length=True,
return_attention_mask=True,
truncation=True,
)
input_ids, attention_mask = tensor_examples["input_ids"], tensor_examples["attention_mask"]
return input_ids, attention_mask
dataset = load_dataset('text', data_files='train.txt',cache_dir="./", , split='train')
data_collator = DataCollatorForDatasetsLanguageModeling(tokenizer=tokenizer, mlm=True,
mlm_probability=0.15, block_size=tokenizer.max_len)
trainer = Trainer(model=model, args=args, data_collator=data_collator,
train_dataset=train_dataset, prediction_loss_only=True, )
trainer.train(model_path=model_path)
```
This train.txt is about 1.1GB and has 90k lines where each line is a sequence of 4k words.
During training, the memory usage increased fast as the following graph and resulted in OOM before the finish of training.
Could you please give me any suggestions on why this happened and how to fix it?
Thanks.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/633/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/633/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/624
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/624/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/624/comments
|
https://api.github.com/repos/huggingface/datasets/issues/624/events
|
https://github.com/huggingface/datasets/issues/624
| 700,541,628
|
MDU6SXNzdWU3MDA1NDE2Mjg=
| 624
|
Add learningq dataset
|
{
"login": "krrishdholakia",
"id": 17561003,
"node_id": "MDQ6VXNlcjE3NTYxMDAz",
"avatar_url": "https://avatars.githubusercontent.com/u/17561003?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/krrishdholakia",
"html_url": "https://github.com/krrishdholakia",
"followers_url": "https://api.github.com/users/krrishdholakia/followers",
"following_url": "https://api.github.com/users/krrishdholakia/following{/other_user}",
"gists_url": "https://api.github.com/users/krrishdholakia/gists{/gist_id}",
"starred_url": "https://api.github.com/users/krrishdholakia/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/krrishdholakia/subscriptions",
"organizations_url": "https://api.github.com/users/krrishdholakia/orgs",
"repos_url": "https://api.github.com/users/krrishdholakia/repos",
"events_url": "https://api.github.com/users/krrishdholakia/events{/privacy}",
"received_events_url": "https://api.github.com/users/krrishdholakia/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[] | 2020-09-13T10:20:27
| 2020-09-14T09:50:02
| null |
NONE
| null | null | null |
Hi,
Thank you again for this amazing repo.
Would it be possible for y'all to add the LearningQ dataset - https://github.com/AngusGLChen/LearningQ ?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/624/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/624/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/616
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/616/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/616/comments
|
https://api.github.com/repos/huggingface/datasets/issues/616/events
|
https://github.com/huggingface/datasets/issues/616
| 699,462,293
|
MDU6SXNzdWU2OTk0NjIyOTM=
| 616
|
UserWarning: The given NumPy array is not writeable, and PyTorch does not support non-writeable tensors
|
{
"login": "BramVanroy",
"id": 2779410,
"node_id": "MDQ6VXNlcjI3Nzk0MTA=",
"avatar_url": "https://avatars.githubusercontent.com/u/2779410?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/BramVanroy",
"html_url": "https://github.com/BramVanroy",
"followers_url": "https://api.github.com/users/BramVanroy/followers",
"following_url": "https://api.github.com/users/BramVanroy/following{/other_user}",
"gists_url": "https://api.github.com/users/BramVanroy/gists{/gist_id}",
"starred_url": "https://api.github.com/users/BramVanroy/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/BramVanroy/subscriptions",
"organizations_url": "https://api.github.com/users/BramVanroy/orgs",
"repos_url": "https://api.github.com/users/BramVanroy/repos",
"events_url": "https://api.github.com/users/BramVanroy/events{/privacy}",
"received_events_url": "https://api.github.com/users/BramVanroy/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"I have the same issue",
"Same issue here when Trying to load a dataset from disk.",
"I am also experiencing this issue, and don't know if it's affecting my training.",
"Same here. I hope the dataset is not being modified in-place.",
"I think the only way to avoid this warning would be to do a copy of the numpy array before providing it.\r\n\r\nThis would slow down a bit the iteration over the dataset but maybe it would be safer. We could disable the copy with a flag on the `set_format` command.\r\n\r\nIn most typical cases of training a NLP model, PyTorch shouldn't modify the input so it's ok to have a non-writable array but I can understand the warning is a bit scary so maybe we could choose the side of non-warning/slower by default and have an option to speedup.\r\n\r\nWhat do you think @lhoestq ? ",
"@thomwolf Would it be possible to have the array look writeable, but raise an error if it is actually written to?\r\n\r\nI would like to keep my code free of warning, but I also wouldn't like to slow down the program because of unnecessary copy operations. ",
"@AndreasMadsen probably not I would guess (no free lunch hahah)",
"@thomwolf Why not? Writable is checked with `arr.flags.writeable`, and writing is done via magic methods.",
"Well because I don't know the internal of numpy as well as you I guess hahahah, do you want to try to open a PR proposing a solution?",
"@thomwolf @AndreasMadsen I think this is a terrible idea, n/o, and I am very much against it. Modifying internals of an array in such a hacky way is bound to run into other (user) issues down the line. To users it would not be clear at all what is going on e.g. when they check for write access (which will return True) but then they get a warning that the array is not writeable. That's extremely confusing. \r\n\r\nIf your only goal is to get rid of warnings in your code, then you can just use a [simplefilter](https://docs.python.org/3.8/library/warnings.html#temporarily-suppressing-warnings) for UserWarnings in your own code. Changing the code-base in such an intuitive way does not seem like a good way to go and sets a bad precedent, imo. \r\n\r\n(Feel free to disagree, of course.)\r\n\r\nIMO a warning can stay (as they can be filtered by users anyway), but it can be clarified why the warning takes place.",
"> To users it would not be clear at all what is going on e.g. when they check for write access (which will return True) but then they get a warning that the array is not writeable. That's extremely confusing.\r\n\r\nConfusion can be resolved with a helpful error message. In this case, that error message can be controlled by huggingface/datasets. The right argument here is that if code depends on `.flags.writable` being truthful (not just for warnings), then it will cause unavoidable errors. Although, I can't imagine such a use-case.\r\n\r\n> If your only goal is to get rid of warnings in your code, then you can just use a simplefilter for UserWarnings in your own code. Changing the code-base in such an intuitive way does not seem like a good way to go and sets a bad precedent, imo.\r\n\r\nI don't want to ignore all `UserWarnings`, nor all warnings regarding non-writable arrays. Ignoring warnings leads to hard to debug issues.\r\n\r\n> IMO a warning can stay (as they can be filtered by users anyway), but it can be clarified why the warning takes place.\r\n\r\nPlain use cases should really not generate warnings. It teaches developers to ignore warnings which is a terrible practice.\r\n\r\n---\r\n\r\nThe best solution would be to allow non-writable arrays in `DataLoader`, but that is a PyTorch issue.",
"> The right argument here is that if code depends on `.flags.writable` being truthful (not just for warnings), then it will cause unavoidable errors. Although, I can't imagine such a use-case.\r\n\r\nThat's exactly the argument in my first sentence. Too often someone \"cannot think of a use-case\", but you can not foresee the use-cases of a whole research community.\r\n \r\n> I don't want to ignore all `UserWarnings`, nor all warnings regarding non-writable arrays. Ignoring warnings leads to hard to debug issues.\r\n\r\nThat's fair.\r\n\r\n> Plain use cases should really not generate warnings. It teaches developers to ignore warnings which is a terrible practice.\r\n\r\nBut this is not a plain use-case (because Pytorch does not support these read-only tensors). Manually setting the flag to writable will solve the issue on the surface but is basically just a hack to compensate for something that is not allowed in another library. \r\n\r\nWhat about an \"ignore_warnings\" flag in `set_format` that when True wraps the offending code in a block to ignore userwarnings at that specific step in [_convert_outputs](https://github.com/huggingface/datasets/blob/880c2c76a8223a00c303eab2909371e857113063/src/datasets/arrow_dataset.py#L821)? Something like:\r\n\r\n```python\r\ndef _convert_outputs(..., ignore_warnings=True):\r\n ...\r\n with warnings.catch_warnings():\r\n if ignore_warnings:\r\n warnings.simplefilter(\"ignore\", UserWarning)\r\n return torch.tensor(...)\r\n# continues without warning filter after context manager...\r\n```",
"> But this is not a plain use-case (because Pytorch does not support these read-only tensors).\r\n\r\nBy \"plain\", I mean the recommended way to use `datasets` with PyTorch according to the `datasets` documentation.",
"This error is what I see when I run the first lines of the Pytorch Quickstart. It should also say that it should be ignored and/or how to fix it. BTW, this is a Pytorch error message -- not a Huggingface error message. My code runs anyway."
] | 2020-09-11T15:39:16
| 2021-07-22T21:12:21
| null |
CONTRIBUTOR
| null | null | null |
I am trying out the library and want to load in pickled data with `from_dict`. In that dict, one column `text` should be tokenized and the other (an embedding vector) should be retained. All other columns should be removed. When I eventually try to set the format for the columns with `set_format` I am getting this strange Userwarning without a stack trace:
> Set __getitem__(key) output type to torch for ['input_ids', 'sembedding'] columns (when key is int or slice) and don't output other (un-formatted) columns.
> C:\Users\bramv\.virtualenvs\dutch-simplification-nbNdqK9u\lib\site-packages\datasets\arrow_dataset.py:835: UserWarning: The given NumPy array is not writeable, and PyTorch does not support non-writeable tensors. This means you can write to the underlying (supposedly non-writeable) NumPy array using the tensor. You may want to copy the array to protect its data or make it writeable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at ..\torch\csrc\utils\tensor_numpy.cpp:141.)
> return torch.tensor(x, **format_kwargs)
The first one might not be related to the warning, but it is odd that it is shown, too. It is unclear whether that is something that I should do or something that that the program is doing at that moment.
Snippet:
```
dataset = Dataset.from_dict(torch.load("data/dummy.pt.pt"))
print(dataset)
tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
keys_to_retain = {"input_ids", "sembedding"}
dataset = dataset.map(lambda example: tokenizer(example["text"], padding='max_length'), batched=True)
dataset.remove_columns_(set(dataset.column_names) - keys_to_retain)
dataset.set_format(type="torch", columns=["input_ids", "sembedding"])
dataloader = torch.utils.data.DataLoader(dataset, batch_size=2)
print(next(iter(dataloader)))
```
PS: the input type for `remove_columns_` should probably be an Iterable rather than just a List.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/616/reactions",
"total_count": 4,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 4
}
|
https://api.github.com/repos/huggingface/datasets/issues/616/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/532
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/532/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/532/comments
|
https://api.github.com/repos/huggingface/datasets/issues/532/events
|
https://github.com/huggingface/datasets/issues/532
| 685,540,614
|
MDU6SXNzdWU2ODU1NDA2MTQ=
| 532
|
File exists error when used with TPU
|
{
"login": "go-inoue",
"id": 20531705,
"node_id": "MDQ6VXNlcjIwNTMxNzA1",
"avatar_url": "https://avatars.githubusercontent.com/u/20531705?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/go-inoue",
"html_url": "https://github.com/go-inoue",
"followers_url": "https://api.github.com/users/go-inoue/followers",
"following_url": "https://api.github.com/users/go-inoue/following{/other_user}",
"gists_url": "https://api.github.com/users/go-inoue/gists{/gist_id}",
"starred_url": "https://api.github.com/users/go-inoue/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/go-inoue/subscriptions",
"organizations_url": "https://api.github.com/users/go-inoue/orgs",
"repos_url": "https://api.github.com/users/go-inoue/repos",
"events_url": "https://api.github.com/users/go-inoue/events{/privacy}",
"received_events_url": "https://api.github.com/users/go-inoue/received_events",
"type": "User",
"site_admin": false
}
|
[] |
open
| false
| null |
[] | null |
[
"I am facing probably facing similar issues with \r\n\r\n`wiki40b_en_100_0`",
"Could you try to run `dataset = load_dataset(\"text\", data_files=file_path, split=\"train\")` once before calling the script ?\r\n\r\nIt looks like several processes try to create the dataset in arrow format at the same time. If the dataset is already created it should be fine",
"Thanks! I tested on 328MB text data on `n1-standard-8 (8 vCPUs, 30 GB memory)`. The main script ran without any issue, but it seems to require a huge space in the drive.\r\n\r\nAs suggested, I ran the following script before running the pre-training command with `xla_spawn.py`.\r\n\r\n```python\r\nfrom nlp import load_dataset\r\n\r\nfile_path=\"your_file_name\"\r\nload_dataset(\"text\", data_files=file_path, split=\"train\")\r\n```\r\nThis will create `text-train.arrow` under the default cache directory. Then, I run the script with `xla_spawn.py`. It will load data from the cached file. My understanding is that there's no other way but to do this two-step process with the current version (0.4) of `nlp`.\r\n\r\nDuring another caching process that happens in the main script:\r\n\r\n```\r\n08/26/2020 09:19:51 - INFO - nlp.utils.info_utils - All the checksums matched successfully for post processing resources\r\n08/26/2020 09:19:53 - INFO - nlp.arrow_dataset - Caching processed dataset at /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d/cache-f90f341e5308a7469\r\n8d872bcc88f9c0e.arrow\r\n```\r\n\r\n`nlp` generates a temporary file per core, each of which is three times larger than the original text data. If each process is actually writing on the disk, you will need a huge amount of space in your drive. (Maybe I'm missing something.)\r\n\r\n```\r\n-rw-r--r-- 1 ***** ***** 674 Aug 26 09:19 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 26 09:19 LICENSE\r\n-rw-r--r-- 1 ***** ***** 332M Aug 26 09:10 text-train.arrow\r\n-rw------- 1 ***** ***** 940M Aug 26 09:31 tmp0k43sazw\r\n-rw------- 1 ***** ***** 940M Aug 26 09:31 tmp7sxs9mj5\r\n-rw------- 1 ***** ***** 939M Aug 26 09:31 tmpbbiqw2vp\r\n-rw------- 1 ***** ***** 937M Aug 26 09:31 tmpjxb5ptyu\r\n-rw------- 1 ***** ***** 933M Aug 26 09:31 tmpk3hkdh0e\r\n-rw------- 1 ***** ***** 944M Aug 26 09:31 tmpnoalwftz\r\n-rw------- 1 ***** ***** 931M Aug 26 09:31 tmpuxdr_dz3\r\n-rw------- 1 ***** ***** 945M Aug 26 09:31 tmpxjyuy6dk\r\n```\r\nAfter the caching process, they seem to be merged into one file.\r\n\r\n```\r\n-rw------- 1 ***** ***** 989M Aug 26 09:32 cache-f90f341e5308a74698d872bcc88f9c0e.arrow\r\n-rw-r--r-- 1 ***** ***** 674 Aug 26 09:19 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 26 09:19 LICENSE\r\n-rw-r--r-- 1 ***** ***** 332M Aug 26 09:10 text-train.arrow\r\n```",
"Again it looks like every process tries to tokenize the full dataset at the same time.\r\nIf you do the tokenization before calling `xla_spawn.py` once, then each process will then use the tokenized cached file `cache-f90f341e5308a74698d872bcc88f9c0e.arrow` and not recompute it.\r\n\r\nNot sure if there's a better way to do that cc @julien-c @thomwolf ",
"I wrote a separate script just for preparing a cached file, including tokenization. Each process did use the tokenized cached file.\r\n\r\nCurrently I'm testing the pipeline on 24GB text data. It took about 1.5 hour to create a cached file on `n1-highmem-16 (16 vCPUs, 104 GB memory)`. I assume loading this cached file in the main script with `xla_spawn.py` won't be an issue (even if there are 8 processes).\r\n\r\n```\r\ntotal 98G\r\ndrwxr-xr-x 2 ***** ***** 4.0K Aug 26 13:38 .\r\ndrwxr-xr-x 3 ***** ***** 4.0K Aug 26 12:24 ..\r\n-rw------- 1 ***** ***** 74G Aug 26 13:38 cache-a7aa04134ba7b1aff5d9710f14a4e334.arrow\r\n-rw-r--r-- 1 ***** ***** 681 Aug 26 12:24 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 26 12:24 LICENSE\r\n-rw-r--r-- 1 ***** ***** 25G Aug 26 12:24 text-train.arrow\r\n```",
"Yes loading the cached file should be fine from different processes",
"Sorry, I thought it was working, but actually the second call doesn't use the cached file that was generated separately, and it will generate another cache-****.arrorw file with a different name. If I run the training script again (with `xla_spawn.py`), it will use the second cached file, which was generated by the training script itself in the previous run.\r\n\r\n```\r\ndrwxr-xr-x 2 ***** ***** 4.0K Aug 26 15:35 .\r\ndrwxr-xr-x 3 ***** ***** 4.0K Aug 26 15:29 ..\r\n-rw------- 1 ***** ***** 99M Aug 26 15:35 cache-0d77dfce704493dbe63f071eed6a5431.arrow\r\n-rw------- 1 ***** ***** 99M Aug 26 15:29 cache-69633651476e943b93c89ace715f9487.arrow\r\n-rw-r--r-- 1 ***** ***** 670 Aug 26 15:33 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 26 15:33 LICENSE\r\n-rw-r--r-- 1 ***** ***** 33M Aug 26 15:29 text-train.arrow\r\n```",
"So if I understand correctly it means that the cached file generated by your separated script is different by the one used by the training script ?",
"Yes.\r\n\r\n1. `cache-69633651476e943b93c89ace715f9487.arrow` was generated with a separate script. \r\n2. I ran the entire script with `xla_spawn.py`.\r\n3. `cache-69633651476e943b93c89ace715f9487.arrow` is not used.\r\n4. `cache-0d77dfce704493dbe63f071eed6a5431.arrow` is created.\r\n5. training starts...\r\n\r\nNow, if I kill the process at step 5, and do the step 2 again, it will use `cache-0d77dfce704493dbe63f071eed6a5431.arrow` (cached file created at step 4) without any issue.\r\n\r\nI used the following to generate the first cached file.\r\n```python\r\ndataset = load_dataset(\"text\", data_files=file_path, split=\"train\")\r\ndataset = dataset.map(lambda ex: tokenizer(ex[\"text\"], add_special_tokens=True,\r\n truncation=True, max_length=args.block_size), batched=True)\r\ndataset.set_format(type='torch', columns=['input_ids'])\r\n```",
"1. Here's the log from the first step.\r\n```\r\nDownloading and preparing dataset text/default-e84dd29acc4ad9ef (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/\r\n447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...\r\nDataset text downloaded and prepared to /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d. Subsequent calls will reuse this data.\r\n```\r\nThere's a file named `cache-7b1440ba7077af0f0d9035b5a55d01fc.arrow`, so it did create a cached file.\r\n```\r\ndrwxr-xr-x 2 ***** ***** 4.0K Aug 26 15:59 .\r\ndrwxr-xr-x 3 ***** ***** 4.0K Aug 26 15:58 ..\r\n-rw------- 1 ***** ***** 99M Aug 26 15:59 cache-7b1440ba7077af0f0d9035b5a55d01fc.arrow\r\n-rw-r--r-- 1 ***** ***** 670 Aug 26 15:58 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 26 15:58 LICENSE\r\n-rw-r--r-- 1 ***** ***** 33M Aug 26 15:58 text-train.arrow\r\n```\r\n2. Ideally, `cache-7b1440ba7077af0f0d9035b5a55d01fc.arrow` should be used in `run_language_modeling.py` (modified version using `nlp`) with `xla_spawn.py`. But it looks like it's creating a new cached file.\r\n\r\n```\r\n08/26/2020 16:13:03 - INFO - filelock - Lock 139635836351096 released on /home/*****/.cache/huggingface/datasets/3e34209a2741375a1db1ff03bf1abba1a9bd0e6016912d3ead0114b9d1ca2685.202fa4f84f552bff1f5400ae012663839c61efb3de068c6c8722d34ac0ea6192\r\n.py.lock\r\n08/26/2020 16:13:03 - WARNING - nlp.builder - Using custom data configuration default\r\n08/26/2020 16:13:03 - INFO - nlp.builder - Overwrite dataset info from restored data version.\r\n08/26/2020 16:13:03 - INFO - nlp.info - Loading Dataset info from /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\n08/26/2020 16:13:03 - INFO - nlp.builder - Reusing dataset text (/home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)\r\n08/26/2020 16:13:03 - INFO - nlp.builder - Constructing Dataset for split train, from /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\n08/26/2020 16:13:03 - INFO - nlp.utils.info_utils - All the checksums matched successfully for post processing resources\r\n08/26/2020 16:13:03 - INFO - nlp.builder - Overwrite dataset info from restored data version.\r\n08/26/2020 16:13:03 - INFO - nlp.info - Loading Dataset info from /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\n08/26/2020 16:13:03 - INFO - nlp.builder - Reusing dataset text (/home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)\r\n08/26/2020 16:13:03 - INFO - nlp.builder - Constructing Dataset for split train, from /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\n08/26/2020 16:13:03 - INFO - nlp.utils.info_utils - All the checksums matched successfully for post processing resources\r\n08/26/2020 16:13:05 - INFO - nlp.arrow_dataset - Caching processed dataset at /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d/cache-0d77dfce704493dbe\r\n63f071eed6a5431.arrow\r\n^M 0%| | 0/100 [00:00<?, ?it/s]08/26/2020 16:13:05 - INFO - nlp.arrow_dataset - Caching processed dataset at /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6\r\nfe661fe4d070d380d/cache-0d77dfce704493dbe63f071eed6a5431.arrow\r\n```\r\n\r\nThere are two cached files in the directory:\r\n\r\n```\r\ndrwxr-xr-x 2 ***** ***** 4.0K Aug 26 16:14 .\r\ndrwxr-xr-x 3 ***** ***** 4.0K Aug 26 15:58 ..\r\n-rw------- 1 ***** ***** 99M Aug 26 16:14 cache-0d77dfce704493dbe63f071eed6a5431.arrow\r\n-rw------- 1 ***** ***** 99M Aug 26 15:59 cache-7b1440ba7077af0f0d9035b5a55d01fc.arrow\r\n-rw-r--r-- 1 ***** ***** 670 Aug 26 16:13 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 26 16:13 LICENSE\r\n-rw-r--r-- 1 ***** ***** 33M Aug 26 15:58 text-train.arrow\r\n```\r\n\r\nIf I kill the process, and run it again, it will use the second cached file.\r\n\r\n```\r\n08/26/2020 16:19:52 - WARNING - nlp.builder - Using custom data configuration default\r\n08/26/2020 16:19:52 - INFO - nlp.builder - Overwrite dataset info from restored data version.\r\n08/26/2020 16:19:52 - INFO - nlp.info - Loading Dataset info from /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\n08/26/2020 16:19:52 - INFO - nlp.builder - Reusing dataset text (/home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)\r\n08/26/2020 16:19:52 - INFO - nlp.builder - Constructing Dataset for split train, from /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\n08/26/2020 16:19:52 - INFO - nlp.utils.info_utils - All the checksums matched successfully for post processing resources\r\n08/26/2020 16:19:53 - INFO - nlp.arrow_dataset - Loading cached processed dataset at /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d/cache-0d77dfce70\r\n4493dbe63f071eed6a5431.arrow\r\n08/26/2020 16:19:53 - INFO - nlp.arrow_dataset - Set __getitem__(key) output type to torch for ['input_ids'] columns (when key is int or slice) and don't output other (un-formatted) columns.\r\n```",
"Thanks for all the details.\r\nThe two cached files are supposed to be the same. I suspect that the caching has a problem with the tokenizer.\r\nWhich tokenizer did you use ?",
"I trained a byte-level BPE tokenizer on my data with `tokenziers` library following this [example](https://github.com/huggingface/tokenizers/blob/master/bindings/python/examples/train_bytelevel_bpe.py).\r\n\r\nAnd I put these model files in a directory named `\"model_name\"`. I also put config.json, which is the original RoBERTa config file.\r\n\r\n```bash\r\n%ls model_name\r\nconfig.json merges.txt vocab.json\r\n```\r\n\r\n[This](https://github.com/huggingface/transformers/blob/4bd7be9a4268221d2a0000c7e8033aaeb365c03b/examples/language-modeling/run_language_modeling.py#L196) is the line where `run_language_modeling.py` loads the tokenier.\r\n\r\n```python\r\ntokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name, cache_dir=model_args.cache_dir)\r\n```\r\n\r\nI use `\"model_name\"` for `model_args.tokenizer_name`. I don't specify `model_args.cache_dir`. It is 'None' by default.",
"In my separated script for caching, I'm using `use_fast=True` when initializing a tokenizer.\r\n\r\n```python\r\ntokenizer = AutoTokenizer.from_pretrained(args.config_name, use_fast=True)\r\n```\r\nI wasn't using that option in the main script. That could be the reason...",
"Yea it could definitely explain why you have two different cache files.\r\nLet me know if using the same tokenizers on both sides fixes the issue",
"It still creates a new file even if I remove `use_fast=True`... \r\n\r\nHere's the script used to create a cached file.\r\n```python \r\n#!/usr/bin/env python3\r\n\r\nimport argparse\r\n\r\nfrom transformers import AutoTokenizer\r\n\r\nfrom nlp import load_dataset\r\n\r\n\r\ndef main():\r\n parser = argparse.ArgumentParser(description='description')\r\n parser.add_argument('--config_name', type=str, help='Pretrained config name or path if not the same as model_name')\r\n parser.add_argument('--data_file', type=str, help='The input data file (a text file).')\r\n parser.add_argument('--block_size', type=int, default=-1, help='The training dataset will be truncated in block of this size for training')\r\n args = parser.parse_args()\r\n\r\n tokenizer = AutoTokenizer.from_pretrained(args.config_name)\r\n\r\n dataset = load_dataset(\"text\", data_files=args.data_file, split=\"train\")\r\n dataset = dataset.map(lambda ex: tokenizer(ex[\"text\"], add_special_tokens=True,\r\n truncation=True, max_length=args.block_size), batched=True)\r\n dataset.set_format(type='torch', columns=['input_ids'])\r\n\r\n\r\nif __name__ == \"__main__\":\r\n main()\r\n```\r\n\r\nHere's how the data is loaded in the modified `run_language_modeling.py`. [[original function](https://github.com/huggingface/transformers/blob/971d1802d009d9996b36a34a34477cee849ef39f/examples/language-modeling/run_language_modeling.py#L128-L135)]\r\n\r\n```python\r\ndef get_dataset(args: DataTrainingArguments, tokenizer: PreTrainedTokenizer, evaluate=False):\r\n file_path = args.eval_data_file if evaluate else args.train_data_file\r\n split = \"validation\" if evaluate else \"train\"\r\n if args.line_by_line:\r\n # return LineByLineTextDataset(tokenizer=tokenizer, file_path=file_path, block_size=args.block_size)\r\n dataset = load_dataset(\"text\", data_files=file_path, split=\"train\")\r\n dataset = dataset.map(lambda ex: tokenizer(ex[\"text\"], add_special_tokens=True,\r\n truncation=True, max_length=args.block_size), batched=True)\r\n dataset.set_format(type='torch', columns=['input_ids'])\r\n return dataset\r\n\r\n else:\r\n return TextDataset(\r\n tokenizer=tokenizer, file_path=file_path, block_size=args.block_size, overwrite_cache=args.overwrite_cache\r\n )\r\n```\r\n\r\nProbably I don't need this part in the main script,\r\n\r\n```python\r\ndataset = dataset.map(lambda ex: tokenizer(ex[\"text\"], add_special_tokens=True,\r\n truncation=True, max_length=args.block_size), batched=True)\r\n dataset.set_format(type='torch', columns=['input_ids'])\r\n```\r\nand simply do this?\r\n```python\r\ndataset = load_dataset(\"text\", data_files=file_path, split=\"train\")\r\nreturn dataset\r\n```",
"You need this part in the main script or it will use the dataset that is not tokenized\r\n\r\n",
"I can see that the tokenizer in `run_language_modeling.py` is not instantiated the same way as in your separated script.\r\nIndeed we can see L196:\r\n```python\r\ntokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name, cache_dir=model_args.cache_dir)\r\n```\r\nCould you try to make it so they are instantiated the exact same way please ?",
"I updated my separated script, but it's creating a cached file again. If I don't use the `model_args.cache_dir`, both will get `None`, so they should be the same.\r\n\r\n```python\r\n#!/usr/bin/env python3\r\nimport argparse\r\n\r\nfrom transformers import AutoTokenizer\r\nfrom nlp import load_dataset\r\n\r\ndef main():\r\n parser = argparse.ArgumentParser(description='description')\r\n parser.add_argument('--tokenizer_name', type=str, help='Pretrained tokenizer name or path if not the same as model_name')\r\n parser.add_argument('--data_file', type=str, help='The input data file (a text file).')\r\n parser.add_argument('--cache_dir', type=str, default=None, help='Where do you want to store the pretrained models downloaded from s3')\r\n parser.add_argument('--block_size', type=int, default=-1, help='The training dataset will be truncated in block of this size for training')\r\n\r\n model_args = parser.parse_args()\r\n\r\n tokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name, cache_dir=model_args.cache_dir)\r\n\r\n dataset = load_dataset(\"text\", data_files=model_args.data_file, split=\"train\")\r\n dataset = dataset.map(lambda ex: tokenizer(ex[\"text\"], add_special_tokens=True,\r\n truncation=True, max_length=model_args.block_size), batched=True)\r\n dataset.set_format(type='torch', columns=['input_ids'])\r\n\r\nif __name__ == \"__main__\":\r\n main()\r\n```\r\n\r\nIs there a way to specify the cache file to load, and skip the re-computation?",
"Could you also check that the `args.block_size` used in the lambda function is the same as well ?",
"Here's a minimal working example to reproduce this issue.\r\n\r\nAssumption:\r\n- You have access to TPU.\r\n- You have installed `transformers` and `nlp`.\r\n- You have tokenizer files (`config.json`, `merges.txt`, `vocab.json`) under the directory named `model_name`.\r\n- You have `xla_spawn.py` (Download from https://github.com/huggingface/transformers/blob/master/examples/xla_spawn.py).\r\n- You have saved the following script as `prepare_cached_dataset.py`.\r\n\r\n```python\r\n#!/usr/bin/env python3\r\nimport argparse\r\nfrom transformers import AutoTokenizer\r\nfrom nlp import load_dataset\r\n\r\ndef main():\r\n parser = argparse.ArgumentParser(description='description')\r\n parser.add_argument('--tokenizer_name', type=str, help='Pretrained tokenizer name or path if not the same as model_name')\r\n parser.add_argument('--data_file', type=str, help='The input data file (a text file).')\r\n parser.add_argument('--cache_dir', type=str, default=None, help='Where do you want to store the pretrained models downloaded from s3')\r\n parser.add_argument('--block_size', type=int, default=-1, help='The training dataset will be truncated in block of this size for training')\r\n parser.add_argument('--tpu_num_cores', type=int, default=1, help='Number of TPU cores to use (1 or 8). For xla_apwan.py')\r\n model_args = parser.parse_args()\r\n \r\n tokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name, cache_dir=model_args.cache_dir, use_fast=True)\r\n \r\n dataset = load_dataset(\"text\", data_files=model_args.data_file, split=\"train\")\r\n dataset = dataset.map(lambda ex: tokenizer(ex[\"text\"], add_special_tokens=True,\r\n truncation=True, max_length=model_args.block_size), batched=True)\r\n dataset.set_format(type='torch', columns=['input_ids'])\r\n\r\ndef _mp_fn(index):\r\n # For xla_spawn (TPUs)\r\n main()\r\n\r\nif __name__ == \"__main__\":\r\n main()\r\n```\r\n\r\n- Run the following command. Replace `your_training_data` with some text file.\r\n\r\n```bash\r\nexport TRAIN_DATA=your_training_data\r\n\r\npython prepare_cached_dataset.py \\\r\n--tokenizer_name=model_name \\\r\n--block_size=512 \\\r\n--data_file=$TRAIN_DATA\r\n```\r\n- Check the cached directory.\r\n```bash\r\nls -lha /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\ntotal 132M\r\ndrwxr-xr-x 2 ***** ***** 4.0K Aug 28 13:08 .\r\ndrwxr-xr-x 3 ***** ***** 4.0K Aug 28 13:08 ..\r\n-rw------- 1 ***** ***** 99M Aug 28 13:08 cache-bfc7cb0702426d19242db5e8c079f04b.arrow\r\n-rw-r--r-- 1 ***** ***** 670 Aug 28 13:08 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 28 13:08 LICENSE\r\n-rw-r--r-- 1 ***** ***** 33M Aug 28 13:08 text-train.arrow\r\n```\r\n\r\n- Run the same script again. (The output should be just `Using custom data configuration default`.)\r\n```\r\npython prepare_cached_dataset.py \\\r\n--tokenizer_name=model_name \\\r\n--block_size=512 \\\r\n--data_file=$TRAIN_DATA\r\n```\r\n- Check the cached directory.\r\n```bash\r\nls -lha /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\ntotal 132M\r\ndrwxr-xr-x 2 ***** ***** 4.0K Aug 28 13:08 .\r\ndrwxr-xr-x 3 ***** ***** 4.0K Aug 28 13:08 ..\r\n-rw------- 1 ***** ***** 99M Aug 28 13:08 cache-bfc7cb0702426d19242db5e8c079f04b.arrow\r\n-rw-r--r-- 1 ***** ***** 670 Aug 28 13:20 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 28 13:20 LICENSE\r\n-rw-r--r-- 1 ***** ***** 33M Aug 28 13:08 text-train.arrow\r\n```\r\n- The cached file (`cache-bfc7cb0702426d19242db5e8c079f04b.arrow`) is reused.\r\n- Now, run this script with `xla_spawn.py`. Ideally, it should reuse the cached file, however, you will see each process is creating a cache file again.\r\n\r\n```bash\r\npython xla_spawn.py --num_cores 8 \\\r\nprepare_cached_dataset.py \\\r\n--tokenizer_name=model_name \\\r\n--block_size=512 \\\r\n--data_file=$TRAIN_DATA\r\n```\r\n\r\n- Check the cached directory. There are two arrrow files.\r\n```bash\r\nls -lha /home/*****/.cache/huggingface/datasets/text/default-e84dd29acc4ad9ef/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d\r\ntotal 230M\r\ndrwxr-xr-x 2 ***** ***** 4.0K Aug 28 13:25 .\r\ndrwxr-xr-x 3 ***** ***** 4.0K Aug 28 13:08 ..\r\n-rw------- 1 ***** ***** 99M Aug 28 13:08 cache-bfc7cb0702426d19242db5e8c079f04b.arrow\r\n-rw------- 1 ***** ***** 99M Aug 28 13:25 cache-e0e2313e49c8a110aafcc8133154c19a.arrow\r\n-rw-r--r-- 1 ***** ***** 670 Aug 28 13:24 dataset_info.json\r\n-rw-r--r-- 1 ***** ***** 0 Aug 28 13:24 LICENSE\r\n-rw-r--r-- 1 ***** ***** 33M Aug 28 13:08 text-train.arrow\r\n```\r\n",
"I ended up specifying the `cache_file_name` argument when I call `map` function.\r\n\r\n```python\r\ndataset = dataset.map(lambda ex: tokenizer(ex[\"text\"], add_special_tokens=True, truncation=True, max_length=args.block_size),\r\n batched=True,\r\n cache_file_name=cache_file_name)\r\n```\r\n\r\nNote:\r\n- `text` dataset in `nlp` does not strip `\"\\n\"`. If you want the same output as in [`LineByLineTextDataset`](https://github.com/huggingface/transformers/blob/afc4ece462ad83a090af620ff4da099a0272e171/src/transformers/data/datasets/language_modeling.py#L88-L111), you would need to create your own dataset class where you replace `line` to `line.strip()` [here](https://github.com/huggingface/nlp/blob/master/datasets/text/text.py#L35).\r\n"
] | 2020-08-25T14:36:38
| 2020-09-01T12:14:56
| null |
NONE
| null | null | null |
Hi,
I'm getting a "File exists" error when I use [text dataset](https://github.com/huggingface/nlp/tree/master/datasets/text) for pre-training a RoBERTa model using `transformers` (3.0.2) and `nlp`(0.4.0) on a VM with TPU (v3-8).
I modified [line 131 in the original `run_language_modeling.py`](https://github.com/huggingface/transformers/blob/master/examples/language-modeling/run_language_modeling.py#L131) as follows:
```python
# line 131: return LineByLineTextDataset(tokenizer=tokenizer, file_path=file_path, block_size=args.block_size)
dataset = load_dataset("text", data_files=file_path, split="train")
dataset = dataset.map(lambda ex: tokenizer(ex["text"], add_special_tokens=True,
truncation=True, max_length=args.block_size), batched=True)
dataset.set_format(type='torch', columns=['input_ids'])
return dataset
```
When I run this with [`xla_spawn.py`](https://github.com/huggingface/transformers/blob/master/examples/xla_spawn.py), I get the following error (it produces one message per core in TPU, which I believe is fine).
It seems the current version doesn't take into account distributed training processes as in [this example](https://github.com/huggingface/transformers/blob/a573777901e662ec2e565be312ffaeedef6effec/src/transformers/data/datasets/language_modeling.py#L35-L38)?
```
08/25/2020 13:59:41 - WARNING - nlp.builder - Using custom data configuration default
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
08/25/2020 13:59:43 - INFO - nlp.builder - Generating dataset text (/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d)
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Exception in device=TPU:6: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Exception in device=TPU:4: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Exception in device=TPU:1: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Exception in device=TPU:7: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Exception in device=TPU:3: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Downloading and preparing dataset text/default-b0932b2bdbb63283 (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/
447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d...
Exception in device=TPU:2: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Exception in device=TPU:0: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Traceback (most recent call last):
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 231, in _start_fn
fn(gindex, *args)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 231, in _start_fn
fn(gindex, *args)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 231, in _start_fn
fn(gindex, *args)
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 300, in _mp_fn
main()
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 300, in _mp_fn
main()
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 300, in _mp_fn
main()
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 240, in main
train_dataset = get_dataset(data_args, tokenizer=tokenizer) if training_args.do_train else None
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 240, in main
train_dataset = get_dataset(data_args, tokenizer=tokenizer) if training_args.do_train else None
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 240, in main
train_dataset = get_dataset(data_args, tokenizer=tokenizer) if training_args.do_train else None
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 134, in get_dataset
dataset = load_dataset("text", data_files=file_path, split="train")
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/load.py", line 546, in load_dataset
download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 134, in get_dataset
dataset = load_dataset("text", data_files=file_path, split="train")
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 134, in get_dataset
dataset = load_dataset("text", data_files=file_path, split="train")
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 450, in download_and_prepare
with incomplete_dir(self._cache_dir) as tmp_data_dir:
Traceback (most recent call last):
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/load.py", line 546, in load_dataset
download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/contextlib.py", line 81, in __enter__
return next(self.gen)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/load.py", line 546, in load_dataset
download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 231, in _start_fn
fn(gindex, *args)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 450, in download_and_prepare
with incomplete_dir(self._cache_dir) as tmp_data_dir:
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 422, in incomplete_dir
os.makedirs(tmp_dir)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 450, in download_and_prepare
with incomplete_dir(self._cache_dir) as tmp_data_dir:
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 300, in _mp_fn
main()
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/contextlib.py", line 81, in __enter__
return next(self.gen)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/os.py", line 220, in makedirs
mkdir(name, mode)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/contextlib.py", line 81, in __enter__
return next(self.gen)
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 240, in main
train_dataset = get_dataset(data_args, tokenizer=tokenizer) if training_args.do_train else None
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 422, in incomplete_dir
os.makedirs(tmp_dir)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 231, in _start_fn
fn(gindex, *args)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 422, in incomplete_dir
os.makedirs(tmp_dir)
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 134, in get_dataset
dataset = load_dataset("text", data_files=file_path, split="train")
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/os.py", line 220, in makedirs
mkdir(name, mode)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/load.py", line 546, in load_dataset
download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,
FileExistsError: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 300, in _mp_fn
main()
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 450, in download_and_prepare
with incomplete_dir(self._cache_dir) as tmp_data_dir:
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/os.py", line 220, in makedirs
mkdir(name, mode)
FileExistsError: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 240, in main
train_dataset = get_dataset(data_args, tokenizer=tokenizer) if training_args.do_train else None
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/contextlib.py", line 81, in __enter__
return next(self.gen)
FileExistsError: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 134, in get_dataset
dataset = load_dataset("text", data_files=file_path, split="train")
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 422, in incomplete_dir
os.makedirs(tmp_dir)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/load.py", line 546, in load_dataset
download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/os.py", line 220, in makedirs
mkdir(name, mode)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 450, in download_and_prepare
with incomplete_dir(self._cache_dir) as tmp_data_dir:
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/contextlib.py", line 81, in __enter__
return next(self.gen)
FileExistsError: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 422, in incomplete_dir
os.makedirs(tmp_dir)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/os.py", line 220, in makedirs
mkdir(name, mode)
Traceback (most recent call last):
FileExistsError: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
Traceback (most recent call last):
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 231, in _start_fn
fn(gindex, *args)
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 300, in _mp_fn
main()
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 240, in main
train_dataset = get_dataset(data_args, tokenizer=tokenizer) if training_args.do_train else None
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 134, in get_dataset
dataset = load_dataset("text", data_files=file_path, split="train")
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/torch_xla/distributed/xla_multiprocessing.py", line 231, in _start_fn
fn(gindex, *args)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/load.py", line 546, in load_dataset
download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 450, in download_and_prepare
with incomplete_dir(self._cache_dir) as tmp_data_dir:
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 300, in _mp_fn
main()
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 240, in main
train_dataset = get_dataset(data_args, tokenizer=tokenizer) if training_args.do_train else None
File "/home/*****/huggingface_roberta/run_language_modeling.py", line 134, in get_dataset
dataset = load_dataset("text", data_files=file_path, split="train")
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/load.py", line 546, in load_dataset
download_config=download_config, download_mode=download_mode, ignore_verifications=ignore_verifications,
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 450, in download_and_prepare
with incomplete_dir(self._cache_dir) as tmp_data_dir:
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/contextlib.py", line 81, in __enter__
return next(self.gen)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/site-packages/nlp/builder.py", line 422, in incomplete_dir
os.makedirs(tmp_dir)
File "/anaconda3/envs/torch-xla-1.6/lib/python3.6/os.py", line 220, in makedirs
mkdir(name, mode)
FileExistsError: [Errno 17] File exists: '/home/*****/.cache/huggingface/datasets/text/default-b0932b2bdbb63283/0.0.0/447f2bcfa2a721a37bc8fdf23800eade1523cf07f7eada6fe661fe4d070d380d.incomplete'
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/532/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/532/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/517
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/517/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/517/comments
|
https://api.github.com/repos/huggingface/datasets/issues/517/events
|
https://github.com/huggingface/datasets/issues/517
| 681,896,944
|
MDU6SXNzdWU2ODE4OTY5NDQ=
| 517
|
add MLDoc dataset
|
{
"login": "jxmorris12",
"id": 13238952,
"node_id": "MDQ6VXNlcjEzMjM4OTUy",
"avatar_url": "https://avatars.githubusercontent.com/u/13238952?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jxmorris12",
"html_url": "https://github.com/jxmorris12",
"followers_url": "https://api.github.com/users/jxmorris12/followers",
"following_url": "https://api.github.com/users/jxmorris12/following{/other_user}",
"gists_url": "https://api.github.com/users/jxmorris12/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jxmorris12/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jxmorris12/subscriptions",
"organizations_url": "https://api.github.com/users/jxmorris12/orgs",
"repos_url": "https://api.github.com/users/jxmorris12/repos",
"events_url": "https://api.github.com/users/jxmorris12/events{/privacy}",
"received_events_url": "https://api.github.com/users/jxmorris12/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[
"Any updates on this?",
"This request is still an open issue waiting to be addressed by any community member, @GuillemGSubies."
] | 2020-08-19T14:41:59
| 2021-08-03T05:59:33
| null |
CONTRIBUTOR
| null | null | null |
Hi,
I am recommending that someone add MLDoc, a multilingual news topic classification dataset.
- Here's a link to the Github: https://github.com/facebookresearch/MLDoc
- and the paper: http://www.lrec-conf.org/proceedings/lrec2018/pdf/658.pdf
Looks like the dataset contains news stories in multiple languages that can be classified into four hierarchical groups: CCAT (Corporate/Industrial), ECAT (Economics), GCAT (Government/Social) and MCAT (Markets). There are 13 languages: Dutch, French, German, Chinese, Japanese, Russian, Portuguese, Spanish, Latin American Spanish, Italian, Danish, Norwegian, and Swedish
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/517/reactions",
"total_count": 4,
"+1": 4,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/517/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/442
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/442/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/442/comments
|
https://api.github.com/repos/huggingface/datasets/issues/442/events
|
https://github.com/huggingface/datasets/issues/442
| 666,201,810
|
MDU6SXNzdWU2NjYyMDE4MTA=
| 442
|
[Suggestion] Glue Diagnostic Data with Labels
|
{
"login": "ggbetz",
"id": 3662782,
"node_id": "MDQ6VXNlcjM2NjI3ODI=",
"avatar_url": "https://avatars.githubusercontent.com/u/3662782?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ggbetz",
"html_url": "https://github.com/ggbetz",
"followers_url": "https://api.github.com/users/ggbetz/followers",
"following_url": "https://api.github.com/users/ggbetz/following{/other_user}",
"gists_url": "https://api.github.com/users/ggbetz/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ggbetz/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ggbetz/subscriptions",
"organizations_url": "https://api.github.com/users/ggbetz/orgs",
"repos_url": "https://api.github.com/users/ggbetz/repos",
"events_url": "https://api.github.com/users/ggbetz/events{/privacy}",
"received_events_url": "https://api.github.com/users/ggbetz/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067401494,
"node_id": "MDU6TGFiZWwyMDY3NDAxNDk0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/Dataset%20discussion",
"name": "Dataset discussion",
"color": "72f99f",
"default": false,
"description": "Discussions on the datasets"
}
] |
open
| false
| null |
[] | null |
[] | 2020-07-27T10:59:58
| 2020-08-24T15:13:20
| null |
NONE
| null | null | null |
Hello! First of all, thanks for setting up this useful project!
I've just realised you provide the the [Glue Diagnostics Data](https://huggingface.co/nlp/viewer/?dataset=glue&config=ax) without labels, indicating in the `GlueConfig` that you've only a test set.
Yet, the data with labels is available, too (see also [here](https://gluebenchmark.com/diagnostics#introduction)):
https://www.dropbox.com/s/ju7d95ifb072q9f/diagnostic-full.tsv?dl=1
Have you considered incorporating it?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/442/reactions",
"total_count": 1,
"+1": 1,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/442/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/438
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/438/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/438/comments
|
https://api.github.com/repos/huggingface/datasets/issues/438/events
|
https://github.com/huggingface/datasets/issues/438
| 665,865,490
|
MDU6SXNzdWU2NjU4NjU0OTA=
| 438
|
New Datasets: IWSLT15+, ITTB
|
{
"login": "sshleifer",
"id": 6045025,
"node_id": "MDQ6VXNlcjYwNDUwMjU=",
"avatar_url": "https://avatars.githubusercontent.com/u/6045025?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/sshleifer",
"html_url": "https://github.com/sshleifer",
"followers_url": "https://api.github.com/users/sshleifer/followers",
"following_url": "https://api.github.com/users/sshleifer/following{/other_user}",
"gists_url": "https://api.github.com/users/sshleifer/gists{/gist_id}",
"starred_url": "https://api.github.com/users/sshleifer/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/sshleifer/subscriptions",
"organizations_url": "https://api.github.com/users/sshleifer/orgs",
"repos_url": "https://api.github.com/users/sshleifer/repos",
"events_url": "https://api.github.com/users/sshleifer/events{/privacy}",
"received_events_url": "https://api.github.com/users/sshleifer/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[
"Thanks Sam, we now have a very detailed tutorial and template on how to add a new dataset to the library. It typically take 1-2 hours to add one. Do you want to give it a try ?\r\nThe tutorial on writing a new dataset loading script is here: https://huggingface.co/nlp/add_dataset.html\r\nAnd the part on how to share a new dataset is here: https://huggingface.co/nlp/share_dataset.html",
"Hi @sshleifer, I'm trying to add IWSLT using the link you provided but the download urls are not working. Only `[en, de]` pair is working. For others language pairs it throws a `404` error.\r\n\r\n"
] | 2020-07-26T21:43:04
| 2020-08-24T15:12:15
| null |
CONTRIBUTOR
| null | null | null |
**Links:**
[iwslt](https://pytorchnlp.readthedocs.io/en/latest/_modules/torchnlp/datasets/iwslt.html)
Don't know if that link is up to date.
[ittb](http://www.cfilt.iitb.ac.in/iitb_parallel/)
**Motivation**: replicate mbart finetuning results (table below)
For future readers, we already have the following language pairs in the wmt namespaces:
```
wmt14: ['cs-en', 'de-en', 'fr-en', 'hi-en', 'ru-en']
wmt15: ['cs-en', 'de-en', 'fi-en', 'fr-en', 'ru-en']
wmt16: ['cs-en', 'de-en', 'fi-en', 'ro-en', 'ru-en', 'tr-en']
wmt17: ['cs-en', 'de-en', 'fi-en', 'lv-en', 'ru-en', 'tr-en', 'zh-en']
wmt18: ['cs-en', 'de-en', 'et-en', 'fi-en', 'kk-en', 'ru-en', 'tr-en', 'zh-en']
wmt19: ['cs-en', 'de-en', 'fi-en', 'gu-en', 'kk-en', 'lt-en', 'ru-en', 'zh-en', 'fr-de']
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/438/reactions",
"total_count": 0,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/438/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/415
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/415/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/415/comments
|
https://api.github.com/repos/huggingface/datasets/issues/415/events
|
https://github.com/huggingface/datasets/issues/415
| 660,687,076
|
MDU6SXNzdWU2NjA2ODcwNzY=
| 415
|
Something is wrong with WMT 19 kk-en dataset
|
{
"login": "ChenghaoMou",
"id": 32014649,
"node_id": "MDQ6VXNlcjMyMDE0NjQ5",
"avatar_url": "https://avatars.githubusercontent.com/u/32014649?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ChenghaoMou",
"html_url": "https://github.com/ChenghaoMou",
"followers_url": "https://api.github.com/users/ChenghaoMou/followers",
"following_url": "https://api.github.com/users/ChenghaoMou/following{/other_user}",
"gists_url": "https://api.github.com/users/ChenghaoMou/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ChenghaoMou/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ChenghaoMou/subscriptions",
"organizations_url": "https://api.github.com/users/ChenghaoMou/orgs",
"repos_url": "https://api.github.com/users/ChenghaoMou/repos",
"events_url": "https://api.github.com/users/ChenghaoMou/events{/privacy}",
"received_events_url": "https://api.github.com/users/ChenghaoMou/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067388877,
"node_id": "MDU6TGFiZWwyMDY3Mzg4ODc3",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20bug",
"name": "dataset bug",
"color": "2edb81",
"default": false,
"description": "A bug in a dataset script provided in the library"
}
] |
open
| false
| null |
[] | null |
[] | 2020-07-19T08:18:51
| 2020-07-20T09:54:26
| null |
NONE
| null | null | null |
The translation in the `train` set does not look right:
```
>>>import nlp
>>>from nlp import load_dataset
>>>dataset = load_dataset('wmt19', 'kk-en')
>>>dataset["train"]["translation"][0]
{'kk': 'Trumpian Uncertainty', 'en': 'Трамптық белгісіздік'}
>>>dataset["validation"]["translation"][0]
{'kk': 'Ақша-несие саясатының сценарийін қайта жазсақ', 'en': 'Rewriting the Monetary-Policy Script'}
```
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/415/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 1,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/415/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/353
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/353/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/353/comments
|
https://api.github.com/repos/huggingface/datasets/issues/353/events
|
https://github.com/huggingface/datasets/issues/353
| 653,250,611
|
MDU6SXNzdWU2NTMyNTA2MTE=
| 353
|
[Dataset requests] New datasets for Text Classification
|
{
"login": "thomwolf",
"id": 7353373,
"node_id": "MDQ6VXNlcjczNTMzNzM=",
"avatar_url": "https://avatars.githubusercontent.com/u/7353373?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/thomwolf",
"html_url": "https://github.com/thomwolf",
"followers_url": "https://api.github.com/users/thomwolf/followers",
"following_url": "https://api.github.com/users/thomwolf/following{/other_user}",
"gists_url": "https://api.github.com/users/thomwolf/gists{/gist_id}",
"starred_url": "https://api.github.com/users/thomwolf/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/thomwolf/subscriptions",
"organizations_url": "https://api.github.com/users/thomwolf/orgs",
"repos_url": "https://api.github.com/users/thomwolf/repos",
"events_url": "https://api.github.com/users/thomwolf/events{/privacy}",
"received_events_url": "https://api.github.com/users/thomwolf/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892884,
"node_id": "MDU6TGFiZWwxOTM1ODkyODg0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/help%20wanted",
"name": "help wanted",
"color": "008672",
"default": true,
"description": "Extra attention is needed"
},
{
"id": 2067376369,
"node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5",
"url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request",
"name": "dataset request",
"color": "e99695",
"default": false,
"description": "Requesting to add a new dataset"
}
] |
open
| false
| null |
[] | null |
[
"Pinging @mariamabarham as well",
"- `nlp` has MR! It's called `rotten_tomatoes`\r\n- SST is part of GLUE, or is that just SST-2?\r\n- `nlp` also has `ag_news`, a popular news classification dataset\r\n\r\nI'd also like to see:\r\n- the Yahoo Answers topic classification dataset\r\n- the Kaggle Fake News classification dataset",
"Thanks @jxmorris12 for pointing this out. \r\n\r\nIn glue we only have SST-2 maybe we can add separately SST-1.\r\n",
"This is the homepage for the Amazon dataset: https://www.kaggle.com/datafiniti/consumer-reviews-of-amazon-products\r\n\r\nIs there an easy way to download kaggle datasets programmatically? If so, I can add this one!",
"Hi @jxmorris12 for now I think our `dl_manager` does not download from Kaggle.\r\n@thomwolf , @lhoestq",
"Pretty sure the quora dataset is the same one I implemented here: https://github.com/huggingface/nlp/pull/366",
"Great list. Any idea if Amazon Reviews has been added?\r\n\r\n- ~40 GB of text (sadly no emoji)\r\n- popular MLM pre-training dataset before bigger datasets like WebText https://arxiv.org/abs/1808.01371\r\n- turns out that binarizing the 1-5 star rating leads to great Pos/Neg/Neutral dataset, T5 paper claims to get very high accuracy (98%!) on this with small amount of finetuning https://arxiv.org/abs/2004.14546\r\n\r\nApologies if it's been included (great to see where) and if not, it's one of the better medium/large NLP dataset for semi-supervised learning, albeit a bit out of date. \r\n\r\nThanks!! \r\n\r\ncc @sshleifer ",
"On the Amazon Reviews dataset, the original UCSD website has noted these are now updated to include product reviews through 2018 -- actually quite recent compared to many other datasets. Almost certainly the largest NLP dataset out there with labels!\r\nhttps://jmcauley.ucsd.edu/data/amazon/ \r\n\r\nAny chance someone has time to onboard this dataset in a HF way?\r\n\r\ncc @sshleifer ",
"@albertvillanova How up to date is this issue? I see that some of these datasets are now on huggingface but have not been checked off the list"
] | 2020-07-08T12:17:58
| 2024-02-07T20:07:15
| null |
MEMBER
| null | null | null |
We are missing a few datasets for Text Classification which is an important field.
Namely, it would be really nice to add:
- [x] TREC-6 dataset (see here for instance: https://pytorchnlp.readthedocs.io/en/latest/source/torchnlp.datasets.html#torchnlp.datasets.trec_dataset) **[done]**
- #386
- [x] Yelp-5
- #1315
- [x] Movie review (Movie Review (MR) dataset [156]) **[done (same as rotten_tomatoes)]**
- [x] SST (Stanford Sentiment Treebank) **[include in glue]**
- #1934
- [ ] Multi-Perspective Question Answering (MPQA) dataset **[require authentication (indeed manual download)]**
- [x] Amazon. This is a popular corpus of product reviews collected from the Amazon website [159]. It contains labels for both binary classification and multi-class (5-class) classification
- #791
- #1389
- [x] 20 Newsgroups. The 20 Newsgroups dataset **[done]**
- #410
- [x] Sogou News dataset **[done]**
- #450
- [x] Reuters news. The Reuters-21578 dataset [165] **[done]**
- #471
- [x] DBpedia. The DBpedia dataset [170]
- #1116
- [ ] Ohsumed. The Ohsumed collection [171] is a subset of the MEDLINE database
- [ ] EUR-Lex. The EUR-Lex dataset
- [x] WOS. The Web Of Science (WOS) dataset **[done]**
- #424
- [ ] PubMed. PubMed [173]
- [x] TREC-QA: TREC-6 + TREC-50
- See above: TREC-6 dataset
- [x] Quora. The Quora dataset [180]
- #366
All these datasets are cited in https://arxiv.org/abs/2004.03705
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/353/reactions",
"total_count": 6,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 3,
"rocket": 3,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/353/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/315
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/315/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/315/comments
|
https://api.github.com/repos/huggingface/datasets/issues/315/events
|
https://github.com/huggingface/datasets/issues/315
| 645,888,943
|
MDU6SXNzdWU2NDU4ODg5NDM=
| 315
|
[Question] Best way to batch a large dataset?
|
{
"login": "jarednielsen",
"id": 4564897,
"node_id": "MDQ6VXNlcjQ1NjQ4OTc=",
"avatar_url": "https://avatars.githubusercontent.com/u/4564897?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/jarednielsen",
"html_url": "https://github.com/jarednielsen",
"followers_url": "https://api.github.com/users/jarednielsen/followers",
"following_url": "https://api.github.com/users/jarednielsen/following{/other_user}",
"gists_url": "https://api.github.com/users/jarednielsen/gists{/gist_id}",
"starred_url": "https://api.github.com/users/jarednielsen/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/jarednielsen/subscriptions",
"organizations_url": "https://api.github.com/users/jarednielsen/orgs",
"repos_url": "https://api.github.com/users/jarednielsen/repos",
"events_url": "https://api.github.com/users/jarednielsen/events{/privacy}",
"received_events_url": "https://api.github.com/users/jarednielsen/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 2067400324,
"node_id": "MDU6TGFiZWwyMDY3NDAwMzI0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/generic%20discussion",
"name": "generic discussion",
"color": "c5def5",
"default": false,
"description": "Generic discussion on the library"
}
] |
open
| false
| null |
[] | null |
[
"Update: I think I've found a solution.\r\n\r\n```python\r\noutput_types = {\"input_ids\": tf.int64, \"token_type_ids\": tf.int64, \"attention_mask\": tf.int64}\r\ndef train_dataset_gen():\r\n for i in range(len(train_dataset)):\r\n yield train_dataset[i]\r\ntf_dataset = tf.data.Dataset.from_generator(train_dataset_gen, output_types=output_types)\r\n```\r\n\r\nloads WikiText-2 in 20 ms, and WikiText-103 in 20 ms. It appears to be lazily loading via indexing train_dataset.",
"Yes this is the current best solution. We should probably show it in the tutorial notebook.\r\n\r\nNote that this solution unfortunately doesn't allow to train on TPUs (yet). See #193 ",
"This approach still seems quite slow. When using TFRecords with a similar training loop, I get ~3.0-3.5 it/s on multi-node, multi-GPU training. I notice a pretty severe performance regression when scaling, with observed performance numbers. Since the allreduce step takes less than 100ms/it and I've achieved 80% scaling efficiency up to 64 GPUs, it must be the data pipeline.\r\n\r\n| Nodes | GPUs | Iterations/Second |\r\n| --- | --- | --- |\r\n| 1 | 2 | 2.01 |\r\n| 1 | 8 | 0.81 |\r\n| 2 | 16 | 0.37 |\r\n\r\nHere are performance metrics over 10k steps. The iteration speed appears to follow some sort of caching pattern. I would love to use `nlp` in my project, but a slowdown from 3.0 it/s to 0.3 it/s is too great to stomach.\r\n\r\n<img width=\"1361\" alt=\"Screen Shot 2020-07-02 at 8 29 22 AM\" src=\"https://user-images.githubusercontent.com/4564897/86378156-2f8d3900-bc3e-11ea-918b-c395c3df5377.png\">\r\n",
"An interesting alternative to investigate here would be to use the tf.io library which has some support for Arrow to TF conversion: https://www.tensorflow.org/io/api_docs/python/tfio/arrow/ArrowDataset\r\n\r\nThere are quite a few types supported, including lists so if the unsupported columns are dropped then we could maybe have a zero-copy mapping from Arrow to TensorFlow, including tokenized inputs and 1D tensors like the ones we mostly use in NLP: https://github.com/tensorflow/io/blob/322b3170c43ecac5c6af9e39dbd18fd747913e5a/tensorflow_io/arrow/python/ops/arrow_dataset_ops.py#L44-L72\r\n\r\nHere is an introduction on Arrow to TF using tf.io: https://medium.com/tensorflow/tensorflow-with-apache-arrow-datasets-cdbcfe80a59f",
"Interesting. There's no support for strings, but it does enable int and floats so that would work for tokenized inputs. \r\n\r\nArrowStreamDataset requires loading from a \"record batch iterator\", which can be instantiated from in-memory arrays as described here: https://arrow.apache.org/docs/python/ipc.html. \r\n\r\nBut the nlp.Dataset stores its data as a `pyarrow.lib.Table`, and the underlying features are `pyarrow.lib.ChunkedArray`. I can't find any documentation about lazily creating a record batch iterator from a ChunkedArray or a Table. Have you had any success?\r\n\r\nI can't find [any uses](https://grep.app/search?q=ArrowDataset&filter[lang][0]=Python) of tfio.arrow.ArrowDataset on GitHub.",
"You can use `to_batches` maybe?\r\nhttps://arrow.apache.org/docs/python/generated/pyarrow.Table.html#pyarrow.Table.to_batches",
"Also note that since #322 it is now possible to do\r\n```python\r\nids = [1, 10, 42, 100]\r\nbatch = dataset[ids]\r\n```\r\nFrom my experience it is quite fast but it can take lots of memory for large batches (haven't played that much with it).\r\nLet me know if you think there could be a better way to implement it. (current code is [here](https://github.com/huggingface/nlp/blob/78628649962671b4aaa31a6b24e7275533416845/src/nlp/arrow_dataset.py#L463))",
"Thanks @lhoestq! That format is much better to work with.\r\n\r\nI put together a benchmarking script. This doesn't measure the CPU-to-GPU efficiency, nor how it scales with multi-GPU multi-node training where many processes are making the same demands on the same dataset. But it does show some interesting results:\r\n\r\n```python\r\nimport nlp\r\nimport numpy as np\r\nimport tensorflow as tf\r\nimport time\r\n\r\ndset = nlp.load_dataset(\"wikitext\", \"wikitext-2-raw-v1\", split=\"train\")\r\ndset = dset.filter(lambda ex: len(ex[\"text\"]) > 0)\r\nbsz = 1024\r\nn_batches = 100\r\n\r\ndef single_item_gen():\r\n for i in range(len(dset)):\r\n yield dset[i]\r\n\r\ndef sequential_batch_gen():\r\n for i in range(0, len(dset), bsz):\r\n yield dset[i:i+bsz]\r\n\r\ndef random_batch_gen():\r\n for i in range(len(dset)):\r\n indices = list(np.random.randint(len(dset), size=(bsz,)))\r\n yield dset[indices]\r\n\r\noutput_types = {\"text\": tf.string}\r\nsingle_item = tf.data.Dataset.from_generator(single_item_gen, output_types=output_types).batch(bsz)\r\ninterleaved = tf.data.Dataset.range(10).interleave(\r\n lambda idx: tf.data.Dataset.from_generator(single_item_gen, output_types=output_types),\r\n cycle_length=10,\r\n)\r\nsequential_batch = tf.data.Dataset.from_generator(sequential_batch_gen, output_types=output_types)\r\nrandom_batch = tf.data.Dataset.from_generator(random_batch_gen, output_types=output_types)\r\n\r\ndef iterate(tf_dset):\r\n start = time.perf_counter()\r\n for i, batch in enumerate(tf_dset.take(n_batches)):\r\n pass\r\n elapsed = time.perf_counter() - start\r\n print(f\"{tf_dset} took {elapsed:.3f} secs\")\r\n\r\niterate(single_item)\r\niterate(interleaved)\r\niterate(sequential_batch)\r\niterate(random_batch)\r\n```\r\n\r\nResults:\r\n```\r\n<BatchDataset shapes: {text: <unknown>}, types: {text: tf.string}> took 23.005 secs\r\n<InterleaveDataset shapes: {text: <unknown>}, types: {text: tf.string}> took 0.135 secs\r\n<FlatMapDataset shapes: {text: <unknown>}, types: {text: tf.string}> took 0.074 secs\r\n<FlatMapDataset shapes: {text: <unknown>}, types: {text: tf.string}> took 0.550 secs\r\n```\r\n\r\n- Batching a generator which fetches a single item is terrible.\r\n- Interleaving performs well on a single process, but doesn't scale well to multi-GPU training. I believe the bottleneck here is in Arrow dataset locking or something similar. The numbers from the table above are with interleaving.\r\n- The sequential access dominates the random access (7x faster). Is there any way to bring random access times closer to sequential access? Maybe re-indexing the dataset after shuffling each pass over the data.",
"Hey @jarednielsen \r\n\r\nThanks for this very interesting analysis!! IMHO to read text data one should use `tf.data.TextLineDataset`. It would be interesting to compare what you have done with simply load with a `TextLineDataset` and see if there is a difference.\r\n\r\nA good example can be found here https://www.tensorflow.org/tutorials/load_data/text",
"Thanks! I'm not actually loading in raw text data, that was just the synthetic data I created for this benchmark. A more realistic use case would be a dataset of tokenized examples, which would be a dict of lists of integers. TensorFlow's TextLineDataset greedily loads the dataset into the graph itself, which can lead to out-of-memory errors - one of the main reason I'm so drawn to the `nlp` library is its zero-copy no-RAM approach to dataset loading and mapping. \r\n\r\nIt's quite helpful for running a preprocessing pipeline - a sample ELECTRA pipeline I've built is here: https://github.com/jarednielsen/deep-learning-models/blob/nlp/models/nlp/common/preprocess.py.",
"Sorry, I think I badly expressed myself, my bad. What I suggested is to compare with the usual loading textual data in pure TF with `TextLineDataset` with `nlp`. I know it is not recommended with very large datasets to use it, but I was curious to see how it behaves compared to a processing with `nlp` on smaller datasets.\r\n\r\nBTW your script looks very interesting, thanks for sharing!!"
] | 2020-06-25T22:30:20
| 2020-10-27T15:38:17
| null |
CONTRIBUTOR
| null | null | null |
I'm training on large datasets such as Wikipedia and BookCorpus. Following the instructions in [the tutorial notebook](https://colab.research.google.com/github/huggingface/nlp/blob/master/notebooks/Overview.ipynb), I see the following recommended for TensorFlow:
```python
train_tf_dataset = train_tf_dataset.filter(remove_none_values, load_from_cache_file=False)
columns = ['input_ids', 'token_type_ids', 'attention_mask', 'start_positions', 'end_positions']
train_tf_dataset.set_format(type='tensorflow', columns=columns)
features = {x: train_tf_dataset[x].to_tensor(default_value=0, shape=[None, tokenizer.max_len]) for x in columns[:3]}
labels = {"output_1": train_tf_dataset["start_positions"].to_tensor(default_value=0, shape=[None, 1])}
labels["output_2"] = train_tf_dataset["end_positions"].to_tensor(default_value=0, shape=[None, 1])
### Question about this last line ###
tfdataset = tf.data.Dataset.from_tensor_slices((features, labels)).batch(8)
```
This code works for something like WikiText-2. However, scaling up to WikiText-103, the last line takes 5-10 minutes to run. I assume it is because tf.data.Dataset.from_tensor_slices() is pulling everything into memory, not lazily loading. This approach won't scale up to datasets 25x larger such as Wikipedia.
So I tried manual batching using `dataset.select()`:
```python
idxs = np.random.randint(len(dataset), size=bsz)
batch = dataset.select(idxs).map(lambda example: {"input_ids": tokenizer(example["text"])})
tf_batch = tf.constant(batch["ids"], dtype=tf.int64)
```
This appears to create a new Apache Arrow dataset with every batch I grab, and then tries to cache it. The runtime of `dataset.select([0, 1])` appears to be much worse than `dataset[:2]`. So using `select()` doesn't seem to be performant enough for a training loop.
Is there a performant scalable way to lazily load batches of nlp Datasets?
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/315/reactions",
"total_count": 1,
"+1": 0,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 1
}
|
https://api.github.com/repos/huggingface/datasets/issues/315/timeline
| null | null | false
|
https://api.github.com/repos/huggingface/datasets/issues/217
|
https://api.github.com/repos/huggingface/datasets
|
https://api.github.com/repos/huggingface/datasets/issues/217/labels{/name}
|
https://api.github.com/repos/huggingface/datasets/issues/217/comments
|
https://api.github.com/repos/huggingface/datasets/issues/217/events
|
https://github.com/huggingface/datasets/issues/217
| 627,128,403
|
MDU6SXNzdWU2MjcxMjg0MDM=
| 217
|
Multi-task dataset mixing
|
{
"login": "ghomasHudson",
"id": 13795113,
"node_id": "MDQ6VXNlcjEzNzk1MTEz",
"avatar_url": "https://avatars.githubusercontent.com/u/13795113?v=4",
"gravatar_id": "",
"url": "https://api.github.com/users/ghomasHudson",
"html_url": "https://github.com/ghomasHudson",
"followers_url": "https://api.github.com/users/ghomasHudson/followers",
"following_url": "https://api.github.com/users/ghomasHudson/following{/other_user}",
"gists_url": "https://api.github.com/users/ghomasHudson/gists{/gist_id}",
"starred_url": "https://api.github.com/users/ghomasHudson/starred{/owner}{/repo}",
"subscriptions_url": "https://api.github.com/users/ghomasHudson/subscriptions",
"organizations_url": "https://api.github.com/users/ghomasHudson/orgs",
"repos_url": "https://api.github.com/users/ghomasHudson/repos",
"events_url": "https://api.github.com/users/ghomasHudson/events{/privacy}",
"received_events_url": "https://api.github.com/users/ghomasHudson/received_events",
"type": "User",
"site_admin": false
}
|
[
{
"id": 1935892871,
"node_id": "MDU6TGFiZWwxOTM1ODkyODcx",
"url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement",
"name": "enhancement",
"color": "a2eeef",
"default": true,
"description": "New feature or request"
},
{
"id": 2067400324,
"node_id": "MDU6TGFiZWwyMDY3NDAwMzI0",
"url": "https://api.github.com/repos/huggingface/datasets/labels/generic%20discussion",
"name": "generic discussion",
"color": "c5def5",
"default": false,
"description": "Generic discussion on the library"
}
] |
open
| false
| null |
[] | null |
[
"I like this feature! I think the first question we should decide on is how to convert all datasets into the same format. In T5, the authors decided to format every dataset into a text-to-text format. If the dataset had \"multiple\" inputs like MNLI, the inputs were concatenated. So in MNLI the input:\r\n\r\n> - **Hypothesis**: The St. Louis Cardinals have always won.\r\n> \r\n> - **Premise**: yeah well losing is i mean i’m i’m originally from Saint Louis and Saint Louis Cardinals when they were there were uh a mostly a losing team but \r\n\r\nwas flattened to a single input:\r\n\r\n> mnli hypothesis: The St. Louis Cardinals have always won. premise:\r\n> yeah well losing is i mean i’m i’m originally from Saint Louis and Saint Louis Cardinals\r\n> when they were there were uh a mostly a losing team but.\r\n\r\nThis flattening is actually a very simple operation in `nlp` already. You would just need to do the following:\r\n\r\n```python \r\ndef flatten_inputs(example):\r\n return {\"input\": \"mnli hypothesis: \" + example['hypothesis'] + \" premise: \" + example['premise']}\r\n\r\nt5_ready_mnli_ds = mnli_ds.map(flatten_inputs, remove_columns=[<all columns except output>])\r\n```\r\n\r\nSo I guess converting the datasets into the same format can be left to the user for now. \r\nThen the question is how we can merge the datasets. I would probably be in favor of a simple \r\n\r\n```python \r\ndataset.add()\r\n```\r\n\r\nfunction that checks if the dataset is of the same format and if yes merges the two datasets. Finally, how should the sampling be implemented? **Examples-proportional mixing** corresponds to just merging the datasets and shuffling. For the other two sampling approaches we would need some higher-level features, maybe even a `dataset.sample()` function for merged datasets. \r\n\r\nWhat are your thoughts on this @thomwolf @lhoestq @ghomasHudson @enzoampil ?",
"I agree that we should leave the flattening of the dataset to the user for now. Especially because although the T5 framing seems obvious, there are slight variations on how the T5 authors do it in comparison to other approaches such as gpt-3 and decaNLP.\r\n\r\nIn terms of sampling, Examples-proportional mixing does seem the simplest to implement so would probably be a good starting point.\r\n\r\nTemperature-scaled mixing would probably most useful, offering flexibility as it can simulate the other 2 methods by setting the temperature parameter. There is a [relevant part of the T5 repo](https://github.com/google-research/text-to-text-transfer-transformer/blob/03c94165a7d52e4f7230e5944a0541d8c5710788/t5/data/utils.py#L889-L1118) which should help with implementation.\r\n\r\nAccording to the T5 authors, equal-mixing performs worst. Among the other two methods, tuning the K value (the artificial dataset size limit) has a large impact.\r\n",
"I agree with going with temperature-scaled mixing for its flexibility!\r\n\r\nFor the function that combines the datasets, I also find `dataset.add()` okay while also considering that users may want it to be easy to combine a list of say 10 data sources in one go.\r\n\r\n`dataset.sample()` should also be good. By the looks of it, we're planning to have as main parameters: `temperature`, and `K`.\r\n\r\nOn converting the datasets to the same format, I agree that we can leave these to the users for now. But, I do imagine it'd be an awesome feature for the future to have this automatically handled, based on a chosen *approach* to formatting :smile: \r\n\r\nE.g. T5, GPT-3, decaNLP, original raw formatting, or a contributed way of formatting in text-to-text. ",
"This is an interesting discussion indeed and it would be nice to make multi-task easier.\r\n\r\nProbably the best would be to have a new type of dataset especially designed for that in order to easily combine and sample from the multiple datasets.\r\n\r\nThis way we could probably handle the combination of datasets with differing schemas as well (unlike T5).",
"@thomwolf Are you suggesting making a wrapper class which can take existing datasets as arguments and do all the required sampling/combining, to present the same interface as a normal dataset?\r\n\r\nThat doesn't seem too complicated to implement.\r\n",
"I guess we're looking at the end user writing something like:\r\n``` python\r\nds = nlp.load_dataset('multitask-t5',datasets=[\"squad\",\"cnn_dm\",...], k=1000, t=2.0)\r\n```\r\nUsing the t5 method of combining here (or this could be a function passed in as an arg) \r\n\r\nPassing kwargs to each 'sub-dataset' might become tricky.",
"From thinking upon @thomwolf 's suggestion, I've started experimenting:\r\n```python\r\nclass MultitaskDataset(DatasetBuilder):\r\n def __init__(self, *args, **kwargs):\r\n super(MultitaskDataset, self).__init__(*args, **kwargs)\r\n self._datasets = kwargs.get(\"datasets\")\r\n\r\n def _info(self):\r\n return nlp.DatasetInfo(\r\n description=_DESCRIPTION,\r\n features=nlp.Features({\r\n \"source\": nlp.Value(\"string\"),\r\n \"target\": nlp.Sequence(nlp.Value(\"string\"))\r\n })\r\n )\r\n\r\n def _get_common_splits(self):\r\n '''Finds the common splits present in all self._datasets'''\r\n min_set = None\r\n for dataset in self._datasets:\r\n if min_set != None:\r\n min_set.intersection(set(dataset.keys()))\r\n else:\r\n min_set = set(dataset.keys())\r\n return min_set\r\n\r\n....\r\n\r\n# Maybe this?:\r\nsquad = nlp.load_dataset(\"squad\")\r\ncnn_dm = nlp.load_dataset(\"cnn_dailymail\",\"3.0.0\")\r\nmultitask_dataset = nlp.load_dataset(\r\n 'multitask_dataset',\r\n datasets=[squad,cnn_dailymail], \r\n k=1000, \r\n t=2.0\r\n)\r\n\r\n```\r\n\r\nDoes anyone know what methods of `MultitaskDataset` I would need to implement? Maybe `as_dataset` and `download_and_prepare`? Most of these should be just calling the methods of the sub-datasets. \r\n\r\nI'm assuming DatasetBuilder is better than the more specific `GeneratorBasedBuilder`, `BeamBasedBuilder`, etc....\r\n\r\nOne of the other problems is that the dataset size is unknown till you construct it (as you can pick the sub-datasets). Am hoping not to need to make changes to `nlp.load_dataset` just for this class.\r\n\r\nI'd appreciate it if anyone more familiar with nlp's internal workings could tell me if I'm on the right track!",
"I think I would probably go for a `MultiDataset` wrapper around a list of `Dataset`.\r\n\r\nI'm not sure we need to give it `k` and `t` parameters at creation, it can maybe be something along the lines of:\r\n```python\r\nsquad = nlp.load_dataset(\"squad\")\r\ncnn_dm = nlp.load_dataset(\"cnn_dailymail\",\"3.0.0\")\r\n\r\nmultitask_dataset = nlp.MultiDataset(squad, cnn_dm)\r\n\r\nbatch = multitask_dataset.sample(10, temperature=2.0, k=1000)\r\n```\r\n\r\nThe first proof-of-concept for multi-task datasets could definitely require that the provided datasets have the same name/type for columns (if needed you easily rename/cast a column prior to instantiating the `MultiDataset`).\r\n\r\nIt's good to think about it for some time though and don't overfit too much on the T5 examples (in particular for the ways/kwargs for sampling among datasets).",
"The problem with changing `k` and `t` per sampling is that you'd have to somehow remember which examples you'd already returned while re-weighting the remaining examples based on the new `k` and `t`values. It seems possible but complicated (I can't really see a reason why you'd want to change the weighting of datasets after you constructed the multidataset).\r\n\r\nWouldn't it be convenient if it implemented the dataset interface? Then if someone has code using a single nlp dataset, they can replace it with a multitask combination of more datasets without having to change other code. We would at least need to be able to pass it into a `DataLoader`.\r\n\r\n",
"A very janky (but working) implementation of `multitask_dataset.sample()` could be something like this:\r\n```python\r\nimport nlp\r\nimport torch\r\n\r\nclass MultiDataset():\r\n def __init__(self, *args, temperature=2.0, k=1000, maximum=None, scale=1):\r\n self.datasets = args\r\n self._dataloaders = {}\r\n for split in self._get_common_splits():\r\n split_datasets = [ds[split] for ds in self.datasets]\r\n mixing_rates = self._calc_mixing_rates(split_datasets,temperature, k, maximum, scale)\r\n weights = []\r\n for i in range(len(self.datasets)):\r\n weights += [mixing_rates[i]]*len(self.datasets[i][split])\r\n self._dataloaders[split] = torch.utils.data.DataLoader(torch.utils.data.ConcatDataset(split_datasets),\r\n sampler=torch.utils.data.sampler.WeightedRandomSampler(\r\n num_samples=len(weights),\r\n weights = weights,\r\n replacement=True),\r\n shuffle=False)\r\n\r\n def _get_common_splits(self):\r\n '''Finds the common splits present in all self.datasets'''\r\n min_set = None\r\n for dataset in self.datasets:\r\n if min_set != None:\r\n min_set.intersection(set(dataset.keys()))\r\n else:\r\n min_set = set(dataset.keys())\r\n return min_set\r\n\r\n\r\n def _calc_mixing_rates(self,datasets, temperature=2.0, k=1000, maximum=None, scale=1):\r\n '''Work out the weighting of each dataset based on t and k'''\r\n mixing_rates = []\r\n for dataset in datasets:\r\n rate = len(dataset)\r\n rate *= scale\r\n if maximum:\r\n rate = min(rate, maximum)\r\n if temperature != 1.0:\r\n rate = rate ** (1.0/temperature)\r\n mixing_rates.append(rate)\r\n return mixing_rates\r\n\r\n def sample(self,n,split):\r\n batch = []\r\n for example in self._dataloaders[split]:\r\n batch.append(example)\r\n n -= 1\r\n if n == 0:\r\n return batch\r\n\r\n\r\ndef flatten(dataset,flatten_fn):\r\n for k in dataset.keys():\r\n if isinstance(dataset[k],nlp.Dataset):\r\n dataset[k] = dataset[k].map(flatten_fn,remove_columns=dataset[k].column_names)\r\n\r\n# Squad\r\ndef flatten_squad(example):\r\n return {\"source\": \"squad context: \" + example['context'] + \" question: \" + example['question'],\"target\":example[\"answers\"][\"text\"]}\r\nsquad = nlp.load_dataset(\"squad\")\r\nflatten(squad,flatten_squad)\r\n\r\n# CNN_DM\r\ndef flatten_cnn_dm(example):\r\n return {\"source\": \"cnn_dm: \" + example['article'],\"target\":[example[\"highlights\"]]}\r\ncnn_dm = nlp.load_dataset(\"cnn_dailymail\", \"3.0.0\")\r\nflatten(cnn_dm,flatten_cnn_dm)\r\n\r\nmultitask_dataset = MultiDataset(squad, cnn_dm)\r\nbatch = multitask_dataset.sample(100,\"train\")\r\n```\r\n\r\nThere's definitely a more sensible way than embedding `DataLoader`s inside. ",
"There is an interesting related investigation by @zphang here https://colab.research.google.com/github/zphang/zphang.github.io/blob/master/files/notebooks/Multi_task_Training_with_Transformers_NLP.ipynb",
"Good spot! Here are my thoughts:\r\n\r\n- Aside: Adding `MultitaskModel` to transformers might be a thing to raise - even though having task-specific heads has become unfashionable in recent times in favour of text-to-text type models.\r\n- Adding the task name as an extra field also seems useful for these kind of models which have task-specific heads\r\n- There is some validation of our approach that the user should be expected to `map` datasets into a common form.\r\n- The size-proportional sampling (also called \"Examples-proportional mixing\") used here doesn't perform too badly in the T5 paper (it's comparable to temperature-scaled mixing in many cases but less flexible. This is only reasonable with a `K` maximum size parameter to prevent very large datasets dominating). This might be good for a first prototype using:\r\n ```python\r\n def __iter__(self):\r\n \"\"\"\r\n For each batch, sample a task, and yield a batch from the respective\r\n task Dataloader.\r\n\r\n We use size-proportional sampling, but you could easily modify this\r\n to sample from some-other distribution.\r\n \"\"\"\r\n task_choice_list = []\r\n for i, task_name in enumerate(self.task_name_list):\r\n task_choice_list += [i] * self.num_batches_dict[task_name]\r\n task_choice_list = np.array(task_choice_list)\r\n np.random.shuffle(task_choice_list)\r\n\r\n dataloader_iter_dict = {\r\n task_name: iter(dataloader) \r\n for task_name, dataloader in self.dataloader_dict.items()\r\n }\r\n for task_choice in task_choice_list:\r\n task_name = self.task_name_list[task_choice]\r\n yield next(dataloader_iter_dict[task_name]) \r\n ```\r\n We'd just need to pull samples from the raw datasets and not from `DataLoader`s for each task. We can assume the user has done `dataset.shuffle()` if they want to.\r\n\r\n Other sampling methods can later be implemented by changing how the `task_choice_list` is generated. This should allow more flexibility and not tie us to specific methods for sampling among datasets.\r\n",
"Another thought: Multitasking over benchmarks (represented as Meta-datasets in nlp) is probably a common use case. Would be nice to pass an entire benchmark to our `MultiDataset` wrapper rather than having to pass individual components.",
"Here's a fully working implementation based on the `__iter__` function of @zphang.\r\n\r\n- I've generated the task choice list in the constructor as it allows us to index into the MultiDataset just like a normal dataset. I'm changing `task_choice_list` into a list of `(dataset_idx, example_idx)` so each entry references a unique dataset example. The shuffling has to be done before this as we don't want to shuffle within each task (we assume this is done by the user if this is what they intend).\r\n- I'm slightly concerned this list could become very large if many large datasets were used. Can't see a way round it at the moment though.\r\n- I've used `task.info.builder_name` as the dataset name. Not sure if this is correct.\r\n- I'd love to add some of the other `Dataset` methods (map, slicing by column, etc...). Would be great to implement the whole interface so a single dataset can be simply replaced by this.\r\n- This does everything on the individual example-level. If some application required batches all from a single task in turn we can't really do that.\r\n\r\n```python\r\nimport nlp\r\nimport numpy as np\r\n\r\nclass MultiDataset:\r\n def __init__(self,tasks):\r\n self.tasks = tasks\r\n\r\n # Create random order of tasks\r\n # Using size-proportional sampling\r\n task_choice_list = []\r\n for i, task in enumerate(self.tasks):\r\n task_choice_list += [i] * len(task)\r\n task_choice_list = np.array(task_choice_list)\r\n np.random.shuffle(task_choice_list)\r\n\r\n # Add index into each dataset\r\n # - We don't want to shuffle within each task\r\n counters = {}\r\n self.task_choice_list = []\r\n for i in range(len(task_choice_list)):\r\n idx = counters.get(task_choice_list[i],0)\r\n self.task_choice_list.append((task_choice_list[i],idx))\r\n counters[task_choice_list[i]] = idx + 1\r\n\r\n\r\n def __len__(self):\r\n return np.sum([len(t) for t in self.tasks])\r\n\r\n def __repr__(self):\r\n task_str = \", \".join([str(t) for t in self.tasks])\r\n return f\"MultiDataset(tasks: {task_str})\"\r\n\r\n def __getitem__(self,key):\r\n if isinstance(key, int):\r\n task_idx, example_idx = self.task_choice_list[key]\r\n task = self.tasks[task_idx]\r\n example = task[example_idx]\r\n example[\"task_name\"] = task.info.builder_name\r\n return example\r\n elif isinstance(key, slice):\r\n raise NotImplementedError()\r\n\r\n def __iter__(self):\r\n for i in range(len(self)):\r\n yield self[i]\r\n\r\n\r\ndef load_multitask(*datasets):\r\n '''Create multitask datasets per split'''\r\n\r\n def _get_common_splits(datasets):\r\n '''Finds the common splits present in all self.datasets'''\r\n min_set = None\r\n for dataset in datasets:\r\n if min_set != None:\r\n min_set.intersection(set(dataset.keys()))\r\n else:\r\n min_set = set(dataset.keys())\r\n return min_set\r\n\r\n common_splits = _get_common_splits(datasets)\r\n out = {}\r\n for split in common_splits:\r\n out[split] = MultiDataset([d[split] for d in datasets])\r\n return out\r\n\r\n\r\n##########################################\r\n# Dataset Flattening\r\n\r\ndef flatten(dataset,flatten_fn):\r\n for k in dataset.keys():\r\n if isinstance(dataset[k],nlp.Dataset):\r\n dataset[k] = dataset[k].map(flatten_fn,remove_columns=dataset[k].column_names)\r\n\r\n# Squad\r\ndef flatten_squad(example):\r\n return {\"source\": \"squad context: \" + example['context'] + \" question: \" + example['question'],\r\n \"target\":example[\"answers\"][\"text\"]}\r\nsquad = nlp.load_dataset(\"squad\")\r\nflatten(squad,flatten_squad)\r\n\r\n# CNN_DM\r\ndef flatten_cnn_dm(example):\r\n return {\"source\": \"cnn_dm: \" + example['article'],\"target\":[example[\"highlights\"]]}\r\ncnn_dm = nlp.load_dataset(\"cnn_dailymail\", \"3.0.0\")\r\nflatten(cnn_dm,flatten_cnn_dm)\r\n\r\n#############################################\r\n\r\nmtds = load_multitask(squad,cnn_dm)\r\n\r\nfor example in mtds[\"train\"]:\r\n print(example[\"task_name\"],example[\"target\"])\r\n```\r\nLet me know if you have any thoughts. I've started using this in some of my projects and it seems to work. If people are happy with the general approach for a first version, I can make a pull request.",
"Hey! Happy to jump into the discussion here. I'm still getting familiar with bits of this code, but the reasons I sampled over data loaders rather than datasets is 1) ensuring that each sampled batch corresponds to only 1 task (in case of different inputs formats/downstream models) and 2) potentially having different batch sizes per task (e.g. some tasks have very long/short inputs). How are you currently dealing with these in your PR?",
"The short answer is - I'm not! Everything is currently on a per-example basis. It would be fairly simple to add a `batch_size` argument which would ensure that every `batch_size` examples come from the same task. That should suit most use-cases (unless you wanted to ensure batches all came from the same task and apply something like `SortishSampler` on each task first)\r\n\r\nYour notebook was really inspiring by the way - thanks!",
"@zphang is having different batch sizes per task actually helpful? Would be interesting to know as it's not something I've come across as a technique used by any MTL papers.",
"mt-dnn's [batcher.py](https://github.com/namisan/mt-dnn/blob/master/mt_dnn/batcher.py) might be worth looking at.",
"> @zphang is having different batch sizes per task actually helpful? Would be interesting to know as it's not something I've come across as a technique used by any MTL papers.\r\n\r\nI think having different batch sizes per task is particularly helpful in some scenarios where each task has different amount of data. For example, the problem I'm currently facing is one task has tens of thousands of samples while one task has a couple hundreds. I think in this case different batch size could help. But if using the same batch size is a lot simpler to implement, I guess it makes sense to go with that.",
"I think that instead of proportional to size sampling you should specify weights or probabilities for drawing a batch from each dataset. We should also ensure that the smaller datasets are repeated so that the encoder layer doesn't overtrain on the largest dataset.",
"Are there any references for people doing different batch sizes per task in the literature? I've only seen constant batch sizes with differing numbers of batches for each task which seems sufficient to prevent the impact of large datasets (Read 3.5.3 of the [T5 paper](https://arxiv.org/pdf/1910.10683.pdf) for example).\r\n\r\n",
"Hi,\r\nregarding building T5 dataset , I think we can use datasets https://github.com/huggingface/datasets and then need something similar to tf.data.experimental.sample_from_datasets, do you know if similar functionality exist in pytorch? Which can sample multiple datasets with the given rates. thanks. ",
"Is this feature part of a `datasets` release yet? ",
"> Here's a fully working implementation based on the `__iter__` function of @zphang.\r\n> \r\n> * I've generated the task choice list in the constructor as it allows us to index into the MultiDataset just like a normal dataset. I'm changing `task_choice_list` into a list of `(dataset_idx, example_idx)` so each entry references a unique dataset example. The shuffling has to be done before this as we don't want to shuffle within each task (we assume this is done by the user if this is what they intend).\r\n> * I'm slightly concerned this list could become very large if many large datasets were used. Can't see a way round it at the moment though.\r\n> * I've used `task.info.builder_name` as the dataset name. Not sure if this is correct.\r\n> * I'd love to add some of the other `Dataset` methods (map, slicing by column, etc...). Would be great to implement the whole interface so a single dataset can be simply replaced by this.\r\n> * This does everything on the individual example-level. If some application required batches all from a single task in turn we can't really do that.\r\n> \r\n> ```python\r\n> import nlp\r\n> import numpy as np\r\n> \r\n> class MultiDataset:\r\n> def __init__(self,tasks):\r\n> self.tasks = tasks\r\n> \r\n> # Create random order of tasks\r\n> # Using size-proportional sampling\r\n> task_choice_list = []\r\n> for i, task in enumerate(self.tasks):\r\n> task_choice_list += [i] * len(task)\r\n> task_choice_list = np.array(task_choice_list)\r\n> np.random.shuffle(task_choice_list)\r\n> \r\n> # Add index into each dataset\r\n> # - We don't want to shuffle within each task\r\n> counters = {}\r\n> self.task_choice_list = []\r\n> for i in range(len(task_choice_list)):\r\n> idx = counters.get(task_choice_list[i],0)\r\n> self.task_choice_list.append((task_choice_list[i],idx))\r\n> counters[task_choice_list[i]] = idx + 1\r\n> \r\n> \r\n> def __len__(self):\r\n> return np.sum([len(t) for t in self.tasks])\r\n> \r\n> def __repr__(self):\r\n> task_str = \", \".join([str(t) for t in self.tasks])\r\n> return f\"MultiDataset(tasks: {task_str})\"\r\n> \r\n> def __getitem__(self,key):\r\n> if isinstance(key, int):\r\n> task_idx, example_idx = self.task_choice_list[key]\r\n> task = self.tasks[task_idx]\r\n> example = task[example_idx]\r\n> example[\"task_name\"] = task.info.builder_name\r\n> return example\r\n> elif isinstance(key, slice):\r\n> raise NotImplementedError()\r\n> \r\n> def __iter__(self):\r\n> for i in range(len(self)):\r\n> yield self[i]\r\n> \r\n> \r\n> def load_multitask(*datasets):\r\n> '''Create multitask datasets per split'''\r\n> \r\n> def _get_common_splits(datasets):\r\n> '''Finds the common splits present in all self.datasets'''\r\n> min_set = None\r\n> for dataset in datasets:\r\n> if min_set != None:\r\n> min_set.intersection(set(dataset.keys()))\r\n> else:\r\n> min_set = set(dataset.keys())\r\n> return min_set\r\n> \r\n> common_splits = _get_common_splits(datasets)\r\n> out = {}\r\n> for split in common_splits:\r\n> out[split] = MultiDataset([d[split] for d in datasets])\r\n> return out\r\n> \r\n> \r\n> ##########################################\r\n> # Dataset Flattening\r\n> \r\n> def flatten(dataset,flatten_fn):\r\n> for k in dataset.keys():\r\n> if isinstance(dataset[k],nlp.Dataset):\r\n> dataset[k] = dataset[k].map(flatten_fn,remove_columns=dataset[k].column_names)\r\n> \r\n> # Squad\r\n> def flatten_squad(example):\r\n> return {\"source\": \"squad context: \" + example['context'] + \" question: \" + example['question'],\r\n> \"target\":example[\"answers\"][\"text\"]}\r\n> squad = nlp.load_dataset(\"squad\")\r\n> flatten(squad,flatten_squad)\r\n> \r\n> # CNN_DM\r\n> def flatten_cnn_dm(example):\r\n> return {\"source\": \"cnn_dm: \" + example['article'],\"target\":[example[\"highlights\"]]}\r\n> cnn_dm = nlp.load_dataset(\"cnn_dailymail\", \"3.0.0\")\r\n> flatten(cnn_dm,flatten_cnn_dm)\r\n> \r\n> #############################################\r\n> \r\n> mtds = load_multitask(squad,cnn_dm)\r\n> \r\n> for example in mtds[\"train\"]:\r\n> print(example[\"task_name\"],example[\"target\"])\r\n> ```\r\n> \r\n> Let me know if you have any thoughts. I've started using this in some of my projects and it seems to work. If people are happy with the general approach for a first version, I can make a pull request.\r\n\r\nNot sure if this is what I'm looking for, but I implemented a version of Examples-Proportional mixing supporting only the basic feature [here](https://stackoverflow.com/a/74070116/10732321), seems to work in my project. ",
"You can use `interleave_datasets` to mix several datasets together. By default it alternates between all the datasets, but you can also provide sampling probabilities if you want to oversample from one of the datasets\r\n\r\n```python\r\nfrom datasets import load_dataset, interleave_datasets\r\n\r\nsquad = load_dataset(\"squad\", split=\"train\")\r\ncnn_dm = load_dataset(\"cnn_dailymail\", \"3.0.0\", split=\"train\")\r\nds = interleave_datasets([squad, cnn_dm])\r\n\r\nprint(ds[0])\r\n# {'id': '5733be284776f41900661182',\r\n# 'title': 'University_of_Notre_Dame',\r\n# 'context': 'Architecturally, the school has a Catholic character...',\r\n# 'question': 'To whom did the Virgin Mary allegedly appear in 1858 in Lourdes France?',\r\n# 'answers': {'text': ['Saint Bernadette Soubirous'], 'answer_start': [515]},\r\n# 'article': None,\r\n# 'highlights': None}\r\nprint(ds[1])\r\n# {'id': '42c027e4ff9730fbb3de84c1af0d2c506e41c3e4',\r\n# 'title': None,\r\n# 'context': None,\r\n# 'question': None,\r\n# 'answers': None,\r\n# 'article': 'LONDON, England (Reuters) -- Harry Potter star Daniel Radcliffe...',\r\n# 'highlights': \"Harry Potter star Daniel Radcliffe...\"}\r\n```\r\n\r\nsee docs at https://huggingface.co/docs/datasets/v2.6.1/en/package_reference/main_classes#datasets.interleave_datasets",
"I also have this implementation of multi-task sampler here which I used it to tune T5: https://github.com/rabeehk/hyperformer/blob/main/hyperformer/data/multitask_sampler.py "
] | 2020-05-29T09:22:26
| 2022-10-22T00:45:50
| null |
CONTRIBUTOR
| null | null | null |
It seems like many of the best performing models on the GLUE benchmark make some use of multitask learning (simultaneous training on multiple tasks).
The [T5 paper](https://arxiv.org/pdf/1910.10683.pdf) highlights multiple ways of mixing the tasks together during finetuning:
- **Examples-proportional mixing** - sample from tasks proportionally to their dataset size
- **Equal mixing** - sample uniformly from each task
- **Temperature-scaled mixing** - The generalized approach used by multilingual BERT which uses a temperature T, where the mixing rate of each task is raised to the power 1/T and renormalized. When T=1 this is equivalent to equal mixing, and becomes closer to equal mixing with increasing T.
Following this discussion https://github.com/huggingface/transformers/issues/4340 in [transformers](https://github.com/huggingface/transformers), @enzoampil suggested that the `nlp` library might be a better place for this functionality.
Some method for combining datasets could be implemented ,e.g.
```
dataset = nlp.load_multitask(['squad','imdb','cnn_dm'], temperature=2.0, ...)
```
We would need a few additions:
- Method of identifying the tasks - how can we support adding a string to each task as an identifier: e.g. 'summarisation: '?
- Method of combining the metrics - a standard approach is to use the specific metric for each task and add them together for a combined score.
It would be great to support common use cases such as pretraining on the GLUE benchmark before fine-tuning on each GLUE task in turn.
I'm willing to write bits/most of this I just need some guidance on the interface and other library details so I can integrate it properly.
|
{
"url": "https://api.github.com/repos/huggingface/datasets/issues/217/reactions",
"total_count": 12,
"+1": 12,
"-1": 0,
"laugh": 0,
"hooray": 0,
"confused": 0,
"heart": 0,
"rocket": 0,
"eyes": 0
}
|
https://api.github.com/repos/huggingface/datasets/issues/217/timeline
| null | null | false
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.