Commit
•
865f2b8
1
Parent(s):
e8db62c
update model card README.md
Browse files
README.md
CHANGED
|
@@ -16,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
This model is a fine-tuned version of [nlpaueb/legal-bert-base-uncased](https://huggingface.co/nlpaueb/legal-bert-base-uncased) on the lex_glue dataset.
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
-
- Loss: 0.
|
| 20 |
-
- Macro-f1: 0.
|
| 21 |
-
- Micro-f1: 0.
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
@@ -37,23 +37,41 @@ More information needed
|
|
| 37 |
### Training hyperparameters
|
| 38 |
|
| 39 |
The following hyperparameters were used during training:
|
| 40 |
-
- learning_rate:
|
| 41 |
- train_batch_size: 8
|
| 42 |
- eval_batch_size: 8
|
| 43 |
- seed: 42
|
| 44 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 45 |
- lr_scheduler_type: linear
|
| 46 |
-
- num_epochs:
|
| 47 |
- mixed_precision_training: Native AMP
|
| 48 |
|
| 49 |
### Training results
|
| 50 |
|
| 51 |
-
| Training Loss | Epoch | Step
|
| 52 |
-
|
| 53 |
-
| 0.
|
| 54 |
-
| 0.
|
| 55 |
-
| 0.
|
| 56 |
-
| 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
|
| 58 |
|
| 59 |
### Framework versions
|
|
|
|
| 16 |
|
| 17 |
This model is a fine-tuned version of [nlpaueb/legal-bert-base-uncased](https://huggingface.co/nlpaueb/legal-bert-base-uncased) on the lex_glue dataset.
|
| 18 |
It achieves the following results on the evaluation set:
|
| 19 |
+
- Loss: 0.1998
|
| 20 |
+
- Macro-f1: 0.5295
|
| 21 |
+
- Micro-f1: 0.6157
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
|
|
|
| 37 |
### Training hyperparameters
|
| 38 |
|
| 39 |
The following hyperparameters were used during training:
|
| 40 |
+
- learning_rate: 3e-05
|
| 41 |
- train_batch_size: 8
|
| 42 |
- eval_batch_size: 8
|
| 43 |
- seed: 42
|
| 44 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 45 |
- lr_scheduler_type: linear
|
| 46 |
+
- num_epochs: 10
|
| 47 |
- mixed_precision_training: Native AMP
|
| 48 |
|
| 49 |
### Training results
|
| 50 |
|
| 51 |
+
| Training Loss | Epoch | Step | Validation Loss | Macro-f1 | Micro-f1 |
|
| 52 |
+
|:-------------:|:-----:|:-----:|:---------------:|:--------:|:--------:|
|
| 53 |
+
| 0.2142 | 0.44 | 500 | 0.2887 | 0.2391 | 0.4263 |
|
| 54 |
+
| 0.172 | 0.89 | 1000 | 0.2672 | 0.2908 | 0.4628 |
|
| 55 |
+
| 0.1737 | 1.33 | 1500 | 0.2612 | 0.3657 | 0.5102 |
|
| 56 |
+
| 0.1581 | 1.78 | 2000 | 0.2412 | 0.3958 | 0.5468 |
|
| 57 |
+
| 0.1509 | 2.22 | 2500 | 0.2264 | 0.3950 | 0.5552 |
|
| 58 |
+
| 0.1606 | 2.67 | 3000 | 0.2342 | 0.4006 | 0.5511 |
|
| 59 |
+
| 0.1491 | 3.11 | 3500 | 0.2176 | 0.4558 | 0.5622 |
|
| 60 |
+
| 0.1392 | 3.56 | 4000 | 0.2454 | 0.4128 | 0.5596 |
|
| 61 |
+
| 0.15 | 4.0 | 4500 | 0.2113 | 0.4684 | 0.5874 |
|
| 62 |
+
| 0.1461 | 4.44 | 5000 | 0.2179 | 0.4631 | 0.5815 |
|
| 63 |
+
| 0.1457 | 4.89 | 5500 | 0.2151 | 0.4805 | 0.5949 |
|
| 64 |
+
| 0.1443 | 5.33 | 6000 | 0.2155 | 0.5123 | 0.5917 |
|
| 65 |
+
| 0.1279 | 5.78 | 6500 | 0.2131 | 0.4915 | 0.5998 |
|
| 66 |
+
| 0.1377 | 6.22 | 7000 | 0.2244 | 0.4705 | 0.5944 |
|
| 67 |
+
| 0.1242 | 6.67 | 7500 | 0.2150 | 0.5089 | 0.5918 |
|
| 68 |
+
| 0.1222 | 7.11 | 8000 | 0.2045 | 0.4801 | 0.5981 |
|
| 69 |
+
| 0.1372 | 7.56 | 8500 | 0.2074 | 0.5317 | 0.5962 |
|
| 70 |
+
| 0.1289 | 8.0 | 9000 | 0.2035 | 0.5323 | 0.6126 |
|
| 71 |
+
| 0.1295 | 8.44 | 9500 | 0.2058 | 0.5213 | 0.6073 |
|
| 72 |
+
| 0.123 | 8.89 | 10000 | 0.2027 | 0.5486 | 0.6135 |
|
| 73 |
+
| 0.1335 | 9.33 | 10500 | 0.1984 | 0.5442 | 0.6249 |
|
| 74 |
+
| 0.1258 | 9.78 | 11000 | 0.1998 | 0.5295 | 0.6157 |
|
| 75 |
|
| 76 |
|
| 77 |
### Framework versions
|