deberta-v3-large-model-edit-classifier
This model is a fine-tuned version of microsoft/deberta-v3-large on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.9102
- Krippendorff: 0.7991
- Accuracy: 0.8571
- Macro F1: 0.8377
- Macro Precision: 0.8426
- Macro Recall: 0.8333
- Micro F1: 0.8571
- Micro Precision: 0.8571
- Micro Recall: 0.8571
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: 6e-06
- 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
- lr_scheduler_warmup_steps: 1000
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Krippendorff | Accuracy | Macro F1 | Macro Precision | Macro Recall | Micro F1 | Micro Precision | Micro Recall |
---|---|---|---|---|---|---|---|---|---|---|---|
No log | 1.0 | 83 | 1.0596 | -0.2872 | 0.5076 | 0.2245 | 0.1692 | 0.3333 | 0.5076 | 0.5076 | 0.5076 |
No log | 2.0 | 166 | 0.9677 | -0.2872 | 0.5076 | 0.2245 | 0.1692 | 0.3333 | 0.5076 | 0.5076 | 0.5076 |
No log | 3.0 | 249 | 0.8841 | -0.2872 | 0.5076 | 0.2245 | 0.1692 | 0.3333 | 0.5076 | 0.5076 | 0.5076 |
No log | 4.0 | 332 | 0.8515 | -0.2872 | 0.5076 | 0.2245 | 0.1692 | 0.3333 | 0.5076 | 0.5076 | 0.5076 |
No log | 5.0 | 415 | 0.8048 | 0.2020 | 0.6140 | 0.5405 | 0.6953 | 0.5634 | 0.6140 | 0.6140 | 0.6140 |
No log | 6.0 | 498 | 0.6971 | 0.5211 | 0.7112 | 0.6850 | 0.7512 | 0.7224 | 0.7112 | 0.7112 | 0.7112 |
0.9047 | 7.0 | 581 | 0.6050 | 0.6242 | 0.7599 | 0.7236 | 0.7615 | 0.7092 | 0.7599 | 0.7599 | 0.7599 |
0.9047 | 8.0 | 664 | 0.4781 | 0.7468 | 0.8116 | 0.7820 | 0.7974 | 0.7747 | 0.8116 | 0.8116 | 0.8116 |
0.9047 | 9.0 | 747 | 0.4446 | 0.7413 | 0.8267 | 0.7986 | 0.8219 | 0.8000 | 0.8267 | 0.8267 | 0.8267 |
0.9047 | 10.0 | 830 | 0.4541 | 0.7510 | 0.8480 | 0.8261 | 0.8455 | 0.8134 | 0.8480 | 0.8480 | 0.8480 |
0.9047 | 11.0 | 913 | 0.4691 | 0.7527 | 0.8419 | 0.8189 | 0.8353 | 0.8069 | 0.8419 | 0.8419 | 0.8419 |
0.9047 | 12.0 | 996 | 0.6103 | 0.7107 | 0.8237 | 0.8011 | 0.8170 | 0.8011 | 0.8237 | 0.8237 | 0.8237 |
0.4334 | 13.0 | 1079 | 0.5507 | 0.7751 | 0.8511 | 0.8360 | 0.8312 | 0.8423 | 0.8511 | 0.8511 | 0.8511 |
0.4334 | 14.0 | 1162 | 0.6642 | 0.7548 | 0.8359 | 0.8168 | 0.8174 | 0.8163 | 0.8359 | 0.8359 | 0.8359 |
0.4334 | 15.0 | 1245 | 0.6858 | 0.8054 | 0.8632 | 0.8443 | 0.8454 | 0.8433 | 0.8632 | 0.8632 | 0.8632 |
0.4334 | 16.0 | 1328 | 0.8332 | 0.7152 | 0.8176 | 0.8007 | 0.7995 | 0.8025 | 0.8176 | 0.8176 | 0.8176 |
0.4334 | 17.0 | 1411 | 0.8105 | 0.7753 | 0.8480 | 0.8308 | 0.8314 | 0.8303 | 0.8480 | 0.8480 | 0.8480 |
0.4334 | 18.0 | 1494 | 0.8472 | 0.8009 | 0.8602 | 0.8416 | 0.8454 | 0.8382 | 0.8602 | 0.8602 | 0.8602 |
0.1213 | 19.0 | 1577 | 0.9164 | 0.7734 | 0.8450 | 0.8273 | 0.8263 | 0.8283 | 0.8450 | 0.8450 | 0.8450 |
0.1213 | 20.0 | 1660 | 0.9102 | 0.7991 | 0.8571 | 0.8377 | 0.8426 | 0.8333 | 0.8571 | 0.8571 | 0.8571 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.13.1
- Datasets 2.10.1
- Tokenizers 0.12.1
- Downloads last month
- 14