LegalBert_NoDuplicates_20Partition_5000WordsFrequency
This model is a fine-tuned version of nlpaueb/legal-bert-base-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4410
- Accuracy: 0.9049
- F1 Macro: 0.8387
- F1 Class 0: 0.9350
- F1 Class 1: 0.6
- F1 Class 2: 0.9290
- F1 Class 3: 0.8000
- F1 Class 4: 0.9014
- F1 Class 5: 0.9388
- F1 Class 6: 0.8119
- F1 Class 7: 0.9317
- F1 Class 8: 0.9804
- F1 Class 9: 0.8595
- F1 Class 10: 0.8834
- F1 Class 11: 0.6087
- F1 Class 12: 0.8280
- F1 Class 13: 0.8333
- F1 Class 14: 0.8808
- F1 Class 15: 0.5588
- F1 Class 16: 0.7273
- F1 Class 17: 0.9799
- F1 Class 18: 0.8440
- F1 Class 19: 0.9412
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: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Macro | F1 Class 0 | F1 Class 1 | F1 Class 2 | F1 Class 3 | F1 Class 4 | F1 Class 5 | F1 Class 6 | F1 Class 7 | F1 Class 8 | F1 Class 9 | F1 Class 10 | F1 Class 11 | F1 Class 12 | F1 Class 13 | F1 Class 14 | F1 Class 15 | F1 Class 16 | F1 Class 17 | F1 Class 18 | F1 Class 19 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1.4309 | 0.44 | 250 | 0.9084 | 0.7836 | 0.5020 | 0.8895 | 0.0 | 0.8333 | 0.6111 | 0.5205 | 0.6966 | 0.4928 | 0.7551 | 0.8496 | 0.7397 | 0.8260 | 0.0 | 0.6420 | 0.0 | 0.5325 | 0.0 | 0.0 | 0.9700 | 0.6813 | 0.0 |
0.6988 | 0.88 | 500 | 0.5860 | 0.8602 | 0.6434 | 0.9078 | 0.0 | 0.8871 | 0.6829 | 0.7907 | 0.7835 | 0.7579 | 0.9091 | 0.9293 | 0.8421 | 0.8542 | 0.0 | 0.7524 | 0.3448 | 0.8243 | 0.0 | 0.0 | 0.9736 | 0.8000 | 0.8276 |
0.4866 | 1.33 | 750 | 0.5249 | 0.8765 | 0.7060 | 0.9189 | 0.0 | 0.8866 | 0.8511 | 0.8095 | 0.8269 | 0.8387 | 0.9375 | 0.9412 | 0.7883 | 0.8801 | 0.1538 | 0.7870 | 0.8085 | 0.8203 | 0.0976 | 0.0 | 0.9690 | 0.8348 | 0.9697 |
0.4198 | 1.77 | 1000 | 0.4760 | 0.8796 | 0.7172 | 0.9177 | 0.0 | 0.9137 | 0.8000 | 0.8406 | 0.7957 | 0.7957 | 0.9308 | 0.9524 | 0.7612 | 0.8718 | 0.2857 | 0.8012 | 0.8085 | 0.8344 | 0.2800 | 0.0 | 0.9799 | 0.8333 | 0.9412 |
0.3275 | 2.21 | 1250 | 0.4650 | 0.8867 | 0.7405 | 0.9221 | 0.0 | 0.9201 | 0.8000 | 0.8608 | 0.8421 | 0.8043 | 0.9367 | 0.9259 | 0.8372 | 0.8731 | 0.3478 | 0.8121 | 0.8085 | 0.8407 | 0.5312 | 0.0 | 0.9737 | 0.8333 | 0.9412 |
0.2874 | 2.65 | 1500 | 0.4662 | 0.8916 | 0.7792 | 0.9221 | 0.6 | 0.9160 | 0.8000 | 0.7879 | 0.8224 | 0.7959 | 0.9325 | 0.9804 | 0.8430 | 0.8896 | 0.5 | 0.7862 | 0.8085 | 0.8693 | 0.5634 | 0.0 | 0.9829 | 0.8421 | 0.9412 |
0.2563 | 3.1 | 1750 | 0.4427 | 0.8978 | 0.7627 | 0.9310 | 0.25 | 0.9272 | 0.8000 | 0.88 | 0.8515 | 0.8602 | 0.9383 | 0.9615 | 0.8438 | 0.8896 | 0.3333 | 0.8182 | 0.8085 | 0.8542 | 0.5574 | 0.0 | 0.9784 | 0.8302 | 0.9412 |
0.2206 | 3.54 | 2000 | 0.4378 | 0.8996 | 0.7920 | 0.9298 | 0.6 | 0.9251 | 0.8000 | 0.9067 | 0.8468 | 0.8200 | 0.9317 | 0.9804 | 0.8413 | 0.8913 | 0.5217 | 0.8208 | 0.8333 | 0.8571 | 0.5574 | 0.0 | 0.9829 | 0.8519 | 0.9412 |
0.1966 | 3.98 | 2250 | 0.4262 | 0.9031 | 0.8378 | 0.9361 | 0.6 | 0.9247 | 0.8000 | 0.9067 | 0.9143 | 0.8235 | 0.9317 | 0.9709 | 0.8739 | 0.8820 | 0.5714 | 0.8239 | 0.8571 | 0.8658 | 0.5846 | 0.7273 | 0.9784 | 0.8421 | 0.9412 |
0.1565 | 4.42 | 2500 | 0.4355 | 0.9075 | 0.8394 | 0.9390 | 0.6 | 0.9307 | 0.8000 | 0.9315 | 0.9184 | 0.8235 | 0.9317 | 0.9703 | 0.8889 | 0.8872 | 0.5833 | 0.8317 | 0.8085 | 0.8725 | 0.5758 | 0.7273 | 0.9829 | 0.8440 | 0.9412 |
0.149 | 4.87 | 2750 | 0.4410 | 0.9049 | 0.8387 | 0.9350 | 0.6 | 0.9290 | 0.8000 | 0.9014 | 0.9388 | 0.8119 | 0.9317 | 0.9804 | 0.8595 | 0.8834 | 0.6087 | 0.8280 | 0.8333 | 0.8808 | 0.5588 | 0.7273 | 0.9799 | 0.8440 | 0.9412 |
Framework versions
- Transformers 4.32.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3
- Downloads last month
- 19
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for Kamer/LegalBert_NoDuplicates_20Partition_5000WordsFrequency
Base model
nlpaueb/legal-bert-base-uncased