|
|
--- |
|
|
license: apache-2.0 |
|
|
tags: |
|
|
- generated_from_trainer |
|
|
datasets: |
|
|
- wikiann |
|
|
metrics: |
|
|
- precision |
|
|
- recall |
|
|
- f1 |
|
|
- accuracy |
|
|
model-index: |
|
|
- name: bert-base-multilingual-cased-finetuned-ner |
|
|
results: |
|
|
- task: |
|
|
name: Token Classification |
|
|
type: token-classification |
|
|
dataset: |
|
|
name: wikiann |
|
|
type: wikiann |
|
|
config: en |
|
|
split: train |
|
|
args: en |
|
|
metrics: |
|
|
- name: Precision |
|
|
type: precision |
|
|
value: 0.8326536254925002 |
|
|
- name: Recall |
|
|
type: recall |
|
|
value: 0.8515481408171921 |
|
|
- name: F1 |
|
|
type: f1 |
|
|
value: 0.8419948974242476 |
|
|
- name: Accuracy |
|
|
type: accuracy |
|
|
value: 0.934650342703884 |
|
|
--- |
|
|
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
|
|
# bert-base-multilingual-cased-finetuned-ner |
|
|
|
|
|
This model is a fine-tuned version of [bert-base-multilingual-cased](https://huggingface.co/bert-base-multilingual-cased) on the wikiann dataset. |
|
|
It achieves the following results on the evaluation set: |
|
|
- Loss: 0.2299 |
|
|
- Precision: 0.8327 |
|
|
- Recall: 0.8515 |
|
|
- F1: 0.8420 |
|
|
- Accuracy: 0.9347 |
|
|
|
|
|
## Model description |
|
|
|
|
|
More information needed |
|
|
|
|
|
## Intended uses & limitations |
|
|
|
|
|
More information needed |
|
|
|
|
|
## Training and evaluation data |
|
|
|
|
|
More information needed |
|
|
|
|
|
## Training procedure |
|
|
|
|
|
### Training hyperparameters |
|
|
|
|
|
The following hyperparameters were used during training: |
|
|
- learning_rate: 2e-05 |
|
|
- train_batch_size: 32 |
|
|
- eval_batch_size: 32 |
|
|
- seed: 42 |
|
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
|
- lr_scheduler_type: linear |
|
|
- num_epochs: 3 |
|
|
|
|
|
### Training results |
|
|
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy | |
|
|
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:| |
|
|
| 0.6617 | 0.16 | 100 | 0.3490 | 0.7259 | 0.7730 | 0.7487 | 0.8983 | |
|
|
| 0.341 | 0.32 | 200 | 0.2942 | 0.7665 | 0.8052 | 0.7854 | 0.9121 | |
|
|
| 0.3052 | 0.48 | 300 | 0.2821 | 0.7694 | 0.8021 | 0.7854 | 0.9152 | |
|
|
| 0.2938 | 0.64 | 400 | 0.2700 | 0.7897 | 0.8122 | 0.8008 | 0.9206 | |
|
|
| 0.2685 | 0.8 | 500 | 0.2482 | 0.7901 | 0.8253 | 0.8073 | 0.9242 | |
|
|
| 0.2622 | 0.96 | 600 | 0.2478 | 0.7989 | 0.8298 | 0.8141 | 0.9250 | |
|
|
| 0.2154 | 1.12 | 700 | 0.2456 | 0.8126 | 0.8365 | 0.8244 | 0.9273 | |
|
|
| 0.2046 | 1.28 | 800 | 0.2429 | 0.8079 | 0.8335 | 0.8205 | 0.9270 | |
|
|
| 0.2114 | 1.44 | 900 | 0.2377 | 0.8125 | 0.8415 | 0.8268 | 0.9300 | |
|
|
| 0.2111 | 1.6 | 1000 | 0.2381 | 0.8231 | 0.8397 | 0.8313 | 0.9309 | |
|
|
| 0.1934 | 1.76 | 1100 | 0.2349 | 0.8179 | 0.8485 | 0.8329 | 0.9308 | |
|
|
| 0.1972 | 1.92 | 1200 | 0.2293 | 0.8287 | 0.8446 | 0.8366 | 0.9332 | |
|
|
| 0.1858 | 2.08 | 1300 | 0.2366 | 0.8280 | 0.8463 | 0.8371 | 0.9327 | |
|
|
| 0.1506 | 2.24 | 1400 | 0.2392 | 0.8255 | 0.8505 | 0.8378 | 0.9332 | |
|
|
| 0.1508 | 2.4 | 1500 | 0.2346 | 0.8266 | 0.8465 | 0.8364 | 0.9334 | |
|
|
| 0.1674 | 2.56 | 1600 | 0.2329 | 0.8249 | 0.8487 | 0.8366 | 0.9329 | |
|
|
| 0.1584 | 2.72 | 1700 | 0.2309 | 0.8316 | 0.8508 | 0.8411 | 0.9341 | |
|
|
| 0.154 | 2.88 | 1800 | 0.2299 | 0.8327 | 0.8515 | 0.8420 | 0.9347 | |
|
|
|
|
|
|
|
|
### Framework versions |
|
|
|
|
|
- Transformers 4.21.0 |
|
|
- Pytorch 1.12.0+cu102 |
|
|
- Datasets 2.4.0 |
|
|
- Tokenizers 0.12.1 |
|
|
|
