adding tensorboard and weights
Browse files
config.json
ADDED
|
@@ -0,0 +1,72 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_function": "gelu_new",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"GPTNeoForMultipleChoice"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0,
|
| 7 |
+
"attention_layers": [
|
| 8 |
+
"global",
|
| 9 |
+
"local",
|
| 10 |
+
"global",
|
| 11 |
+
"local",
|
| 12 |
+
"global",
|
| 13 |
+
"local",
|
| 14 |
+
"global",
|
| 15 |
+
"local",
|
| 16 |
+
"global",
|
| 17 |
+
"local",
|
| 18 |
+
"global",
|
| 19 |
+
"local",
|
| 20 |
+
"global",
|
| 21 |
+
"local",
|
| 22 |
+
"global",
|
| 23 |
+
"local",
|
| 24 |
+
"global",
|
| 25 |
+
"local",
|
| 26 |
+
"global",
|
| 27 |
+
"local",
|
| 28 |
+
"global",
|
| 29 |
+
"local",
|
| 30 |
+
"global",
|
| 31 |
+
"local"
|
| 32 |
+
],
|
| 33 |
+
"attention_types": [
|
| 34 |
+
[
|
| 35 |
+
[
|
| 36 |
+
"global",
|
| 37 |
+
"local"
|
| 38 |
+
],
|
| 39 |
+
12
|
| 40 |
+
]
|
| 41 |
+
],
|
| 42 |
+
"bos_token_id": 50256,
|
| 43 |
+
"embed_dropout": 0,
|
| 44 |
+
"eos_token_id": 50256,
|
| 45 |
+
"gradient_checkpointing": false,
|
| 46 |
+
"hidden_size": 2048,
|
| 47 |
+
"initializer_range": 0.02,
|
| 48 |
+
"intermediate_size": null,
|
| 49 |
+
"layer_norm_epsilon": 1e-05,
|
| 50 |
+
"max_position_embeddings": 2048,
|
| 51 |
+
"model_type": "gpt_neo",
|
| 52 |
+
"num_heads": 16,
|
| 53 |
+
"num_layers": 24,
|
| 54 |
+
"resid_dropout": 0,
|
| 55 |
+
"summary_activation": null,
|
| 56 |
+
"summary_first_dropout": 0.1,
|
| 57 |
+
"summary_proj_to_labels": true,
|
| 58 |
+
"summary_type": "cls_index",
|
| 59 |
+
"summary_use_proj": true,
|
| 60 |
+
"task_specific_params": {
|
| 61 |
+
"text-generation": {
|
| 62 |
+
"do_sample": true,
|
| 63 |
+
"max_length": 50,
|
| 64 |
+
"temperature": 0.9
|
| 65 |
+
}
|
| 66 |
+
},
|
| 67 |
+
"tokenizer_class": "GPT2Tokenizer",
|
| 68 |
+
"transformers_version": "4.10.0.dev0",
|
| 69 |
+
"use_cache": true,
|
| 70 |
+
"vocab_size": 50257,
|
| 71 |
+
"window_size": 256
|
| 72 |
+
}
|
flax_model.msgpack
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ea0172ae20da21487c58a2b55a8aaf3e67f8ab7e585b5868528210f98a3fe7ed
|
| 3 |
+
size 5262519475
|
results_tensorboard/events.out.tfevents.1627652258.t1v-n-cb40e504-w-0.6491.3.v2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:577723f7409e1a7280adbdaa1fb529a47a4d00fb2a2cf9a22d21da1705ae528f
|
| 3 |
+
size 58744
|
src/gptneo_cose.py
CHANGED
|
@@ -45,8 +45,8 @@ def preprocess(example):
|
|
| 45 |
train_dataset=dataset['train'].map(preprocess)
|
| 46 |
validation_dataset=dataset['validation'].map(preprocess)
|
| 47 |
|
| 48 |
-
len_train_dataset=
|
| 49 |
-
len_validation_dataset=
|
| 50 |
|
| 51 |
train_dataset=train_dataset.select(range(len_train_dataset))
|
| 52 |
validation_dataset=validation_dataset.select(range(len_validation_dataset))
|
|
|
|
| 45 |
train_dataset=dataset['train'].map(preprocess)
|
| 46 |
validation_dataset=dataset['validation'].map(preprocess)
|
| 47 |
|
| 48 |
+
len_train_dataset=9741
|
| 49 |
+
len_validation_dataset=1221
|
| 50 |
|
| 51 |
train_dataset=train_dataset.select(range(len_train_dataset))
|
| 52 |
validation_dataset=validation_dataset.select(range(len_validation_dataset))
|