asahi417's picture
model update
bc66b6a
|
raw
history blame
6.85 kB
---
datasets:
- tner/tweetner7
metrics:
- f1
- precision
- recall
model-index:
- name: tner/bert-base-tweetner7-2020-2021-concat
results:
- task:
name: Token Classification
type: token-classification
dataset:
name: tner/tweetner7/test_2021
type: tner/tweetner7/test_2021
args: tner/tweetner7/test_2021
metrics:
- name: F1
type: f1
value: 0.6230258640421148
- name: Precision
type: precision
value: 0.6166742183960127
- name: Recall
type: recall
value: 0.6295097132284921
- name: F1 (macro)
type: f1_macro
value: 0.5758556427048315
- name: Precision (macro)
type: precision_macro
value: 0.5715554663683273
- name: Recall (macro)
type: recall_macro
value: 0.5821234872899773
- name: F1 (entity span)
type: f1_entity_span
value: 0.7661839619941617
- name: Precision (entity span)
type: precision_entity_span
value: 0.7584995466908432
- name: Recall (entity span)
type: recall_entity_span
value: 0.7740256736440384
- task:
name: Token Classification
type: token-classification
dataset:
name: tner/tweetner7/test_2020
type: tner/tweetner7/test_2020
args: tner/tweetner7/test_2020
metrics:
- name: F1
type: f1
value: 0.6210070384407147
- name: Precision
type: precision
value: 0.6491228070175439
- name: Recall
type: recall
value: 0.5952257394914374
- name: F1 (macro)
type: f1_macro
value: 0.577436139660066
- name: Precision (macro)
type: precision_macro
value: 0.6119340101835135
- name: Recall (macro)
type: recall_macro
value: 0.549500601374034
- name: F1 (entity span)
type: f1_entity_span
value: 0.7298321602598808
- name: Precision (entity span)
type: precision_entity_span
value: 0.7628749292586304
- name: Recall (entity span)
type: recall_entity_span
value: 0.6995329527763363
pipeline_tag: token-classification
widget:
- text: "Get the all-analog Classic Vinyl Edition of `Takin' Off` Album from {{@Herbie Hancock@}} via {{USERNAME}} link below: {{URL}}"
example_title: "NER Example 1"
---
# tner/bert-base-tweetner7-2020-2021-concat
This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/bert-base-cased) on the
[tner/tweetner7](https://huggingface.co/datasets/tner/tweetner7) dataset (`train_all` split).
Model fine-tuning is done via [T-NER](https://github.com/asahi417/tner)'s hyper-parameter search (see the repository
for more detail). It achieves the following results on the test set of 2021:
- F1 (micro): 0.6230258640421148
- Precision (micro): 0.6166742183960127
- Recall (micro): 0.6295097132284921
- F1 (macro): 0.5758556427048315
- Precision (macro): 0.5715554663683273
- Recall (macro): 0.5821234872899773
The per-entity breakdown of the F1 score on the test set are below:
- corporation: 0.5141176470588235
- creative_work: 0.3886075949367089
- event: 0.4580617122990004
- group: 0.5660613650594865
- location: 0.6264564770390679
- person: 0.8196536144578314
- product: 0.6580310880829014
For F1 scores, the confidence interval is obtained by bootstrap as below:
- F1 (micro):
- 90%: [0.6139925448708724, 0.632549139769655]
- 95%: [0.612303125388328, 0.6336744975616968]
- F1 (macro):
- 90%: [0.6139925448708724, 0.632549139769655]
- 95%: [0.612303125388328, 0.6336744975616968]
Full evaluation can be found at [metric file of NER](https://huggingface.co/tner/bert-base-tweetner7-2020-2021-concat/raw/main/eval/metric.json)
and [metric file of entity span](https://huggingface.co/tner/bert-base-tweetner7-2020-2021-concat/raw/main/eval/metric_span.json).
### Usage
This model can be used through the [tner library](https://github.com/asahi417/tner). Install the library via pip
```shell
pip install tner
```
and activate model as below.
```python
from tner import TransformersNER
model = TransformersNER("tner/bert-base-tweetner7-2020-2021-concat")
model.predict(["Jacob Collier is a Grammy awarded English artist from London"])
```
It can be used via transformers library but it is not recommended as CRF layer is not supported at the moment.
### Training hyperparameters
The following hyperparameters were used during training:
- dataset: ['tner/tweetner7']
- dataset_split: train_all
- dataset_name: None
- local_dataset: None
- model: bert-base-cased
- crf: True
- max_length: 128
- epoch: 30
- batch_size: 32
- lr: 0.0001
- random_seed: 0
- gradient_accumulation_steps: 1
- weight_decay: 1e-07
- lr_warmup_step_ratio: 0.3
- max_grad_norm: 1
The full configuration can be found at [fine-tuning parameter file](https://huggingface.co/tner/bert-base-tweetner7-2020-2021-concat/raw/main/trainer_config.json).
### Reference
If you use any resource from T-NER, please consider to cite our [paper](https://aclanthology.org/2021.eacl-demos.7/).
```
@inproceedings{ushio-camacho-collados-2021-ner,
title = "{T}-{NER}: An All-Round Python Library for Transformer-based Named Entity Recognition",
author = "Ushio, Asahi and
Camacho-Collados, Jose",
booktitle = "Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations",
month = apr,
year = "2021",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.eacl-demos.7",
doi = "10.18653/v1/2021.eacl-demos.7",
pages = "53--62",
abstract = "Language model (LM) pretraining has led to consistent improvements in many NLP downstream tasks, including named entity recognition (NER). In this paper, we present T-NER (Transformer-based Named Entity Recognition), a Python library for NER LM finetuning. In addition to its practical utility, T-NER facilitates the study and investigation of the cross-domain and cross-lingual generalization ability of LMs finetuned on NER. Our library also provides a web app where users can get model predictions interactively for arbitrary text, which facilitates qualitative model evaluation for non-expert programmers. We show the potential of the library by compiling nine public NER datasets into a unified format and evaluating the cross-domain and cross- lingual performance across the datasets. The results from our initial experiments show that in-domain performance is generally competitive across datasets. However, cross-domain generalization is challenging even with a large pretrained LM, which has nevertheless capacity to learn domain-specific features if fine- tuned on a combined dataset. To facilitate future research, we also release all our LM checkpoints via the Hugging Face model hub.",
}
```