|
|
--- |
|
|
license: mit |
|
|
tags: |
|
|
- generated_from_trainer |
|
|
metrics: |
|
|
- accuracy |
|
|
- precision |
|
|
- recall |
|
|
base_model: law-ai/InLegalBERT |
|
|
model-index: |
|
|
- name: InLegalBERT-lora |
|
|
results: [] |
|
|
--- |
|
|
|
|
|
<!-- 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. --> |
|
|
|
|
|
# InLegalBERT-lora |
|
|
|
|
|
This model is a fine-tuned version of [law-ai/InLegalBERT](https://huggingface.co/law-ai/InLegalBERT) on an unknown dataset. |
|
|
It achieves the following results on the evaluation set: |
|
|
- Loss: 0.6344 |
|
|
- Accuracy: 0.8203 |
|
|
- Precision: 0.8092 |
|
|
- Recall: 0.8203 |
|
|
- Precision Macro: 0.6487 |
|
|
- Recall Macro: 0.6625 |
|
|
- Macro Fpr: 0.0160 |
|
|
- Weighted Fpr: 0.0154 |
|
|
- Weighted Specificity: 0.9771 |
|
|
- Macro Specificity: 0.9865 |
|
|
- Weighted Sensitivity: 0.8203 |
|
|
- Macro Sensitivity: 0.6625 |
|
|
- F1 Micro: 0.8203 |
|
|
- F1 Macro: 0.6461 |
|
|
- F1 Weighted: 0.8125 |
|
|
|
|
|
## 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: 5e-05 |
|
|
- train_batch_size: 8 |
|
|
- eval_batch_size: 8 |
|
|
- seed: 42 |
|
|
- gradient_accumulation_steps: 4 |
|
|
- total_train_batch_size: 32 |
|
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
|
- lr_scheduler_type: linear |
|
|
- num_epochs: 15 |
|
|
|
|
|
### Training results |
|
|
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | Precision Macro | Recall Macro | Macro Fpr | Weighted Fpr | Weighted Specificity | Macro Specificity | Weighted Sensitivity | Macro Sensitivity | F1 Micro | F1 Macro | F1 Weighted | |
|
|
|:-------------:|:-----:|:----:|:---------------:|:--------:|:---------:|:------:|:---------------:|:------------:|:---------:|:------------:|:--------------------:|:-----------------:|:--------------------:|:-----------------:|:--------:|:--------:|:-----------:| |
|
|
| No log | 1.0 | 160 | 1.2013 | 0.6553 | 0.6007 | 0.6553 | 0.3279 | 0.3903 | 0.0365 | 0.0362 | 0.9556 | 0.9741 | 0.6553 | 0.3903 | 0.6553 | 0.3420 | 0.6147 | |
|
|
| No log | 2.0 | 321 | 0.8279 | 0.7382 | 0.7211 | 0.7382 | 0.4092 | 0.4658 | 0.0248 | 0.0247 | 0.9713 | 0.9806 | 0.7382 | 0.4658 | 0.7382 | 0.4250 | 0.7237 | |
|
|
| No log | 3.0 | 482 | 0.7130 | 0.7545 | 0.7255 | 0.7545 | 0.4800 | 0.4770 | 0.0233 | 0.0227 | 0.9701 | 0.9816 | 0.7545 | 0.4770 | 0.7545 | 0.4431 | 0.7305 | |
|
|
| 1.1985 | 4.0 | 643 | 0.6922 | 0.7823 | 0.7594 | 0.7823 | 0.5188 | 0.5283 | 0.0200 | 0.0195 | 0.9740 | 0.9838 | 0.7823 | 0.5283 | 0.7823 | 0.5048 | 0.7660 | |
|
|
| 1.1985 | 5.0 | 803 | 0.6710 | 0.7940 | 0.7734 | 0.7940 | 0.5450 | 0.5571 | 0.0190 | 0.0182 | 0.9739 | 0.9845 | 0.7940 | 0.5571 | 0.7940 | 0.5257 | 0.7718 | |
|
|
| 1.1985 | 6.0 | 964 | 0.6455 | 0.7971 | 0.7757 | 0.7971 | 0.5353 | 0.5622 | 0.0184 | 0.0179 | 0.9754 | 0.9848 | 0.7971 | 0.5622 | 0.7971 | 0.5316 | 0.7790 | |
|
|
| 0.5721 | 7.0 | 1125 | 0.6395 | 0.8002 | 0.7801 | 0.8002 | 0.5443 | 0.5784 | 0.0181 | 0.0175 | 0.9762 | 0.9851 | 0.8002 | 0.5784 | 0.8002 | 0.5486 | 0.7845 | |
|
|
| 0.5721 | 8.0 | 1286 | 0.6317 | 0.8025 | 0.7833 | 0.8025 | 0.5439 | 0.5773 | 0.0178 | 0.0173 | 0.9765 | 0.9853 | 0.8025 | 0.5773 | 0.8025 | 0.5475 | 0.7874 | |
|
|
| 0.5721 | 9.0 | 1446 | 0.6137 | 0.8009 | 0.7828 | 0.8009 | 0.5593 | 0.5842 | 0.0179 | 0.0174 | 0.9765 | 0.9852 | 0.8009 | 0.5842 | 0.8009 | 0.5609 | 0.7875 | |
|
|
| 0.4166 | 10.0 | 1607 | 0.6249 | 0.8156 | 0.8055 | 0.8156 | 0.6398 | 0.6430 | 0.0165 | 0.0159 | 0.9772 | 0.9862 | 0.8156 | 0.6430 | 0.8156 | 0.6305 | 0.8067 | |
|
|
| 0.4166 | 11.0 | 1768 | 0.6426 | 0.8125 | 0.8014 | 0.8125 | 0.6397 | 0.6520 | 0.0169 | 0.0162 | 0.9762 | 0.9859 | 0.8125 | 0.6520 | 0.8125 | 0.6372 | 0.8042 | |
|
|
| 0.4166 | 12.0 | 1929 | 0.6305 | 0.8164 | 0.8050 | 0.8164 | 0.6358 | 0.6526 | 0.0164 | 0.0158 | 0.9770 | 0.9862 | 0.8164 | 0.6526 | 0.8164 | 0.6372 | 0.8083 | |
|
|
| 0.3406 | 13.0 | 2089 | 0.6276 | 0.8203 | 0.8102 | 0.8203 | 0.6418 | 0.6467 | 0.0160 | 0.0154 | 0.9774 | 0.9865 | 0.8203 | 0.6467 | 0.8203 | 0.6353 | 0.8129 | |
|
|
| 0.3406 | 14.0 | 2250 | 0.6428 | 0.8187 | 0.8079 | 0.8187 | 0.6467 | 0.6618 | 0.0162 | 0.0156 | 0.9771 | 0.9864 | 0.8187 | 0.6618 | 0.8187 | 0.6446 | 0.8107 | |
|
|
| 0.3406 | 14.93 | 2400 | 0.6344 | 0.8203 | 0.8092 | 0.8203 | 0.6487 | 0.6625 | 0.0160 | 0.0154 | 0.9771 | 0.9865 | 0.8203 | 0.6625 | 0.8203 | 0.6461 | 0.8125 | |
|
|
|
|
|
|
|
|
### Framework versions |
|
|
|
|
|
- Transformers 4.35.2 |
|
|
- Pytorch 2.1.0+cu121 |
|
|
- Datasets 2.18.0 |
|
|
- Tokenizers 0.15.1 |
|
|
|
