| --- |
| license: mit |
| base_model: microsoft/deberta-v3-large |
| tags: |
| - generated_from_trainer |
| metrics: |
| - accuracy |
| model-index: |
| - name: Classifier_with_external_sets_05 |
| 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. --> |
|
|
| # Classifier_with_external_sets_05 |
|
|
| This model is a fine-tuned version of [microsoft/deberta-v3-large](https://huggingface.co/microsoft/deberta-v3-large) on the None dataset. |
| It achieves the following results on the evaluation set: |
| - Loss: 0.2840 |
| - Accuracy: 0.9627 |
|
|
| ## 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-06 |
| - 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: 50 |
| - mixed_precision_training: Native AMP |
| |
| ### Training results |
| |
| | Training Loss | Epoch | Step | Validation Loss | Accuracy | |
| |:-------------:|:-------:|:-----:|:---------------:|:--------:| |
| | No log | 0.9983 | 289 | 0.6833 | 0.7547 | |
| | 0.403 | 2.0 | 579 | 0.4286 | 0.7700 | |
| | 0.403 | 2.9983 | 868 | 0.5718 | 0.8196 | |
| | 0.1978 | 4.0 | 1158 | 0.3336 | 0.8813 | |
| | 0.1978 | 4.9983 | 1447 | 0.3455 | 0.8795 | |
| | 0.1523 | 6.0 | 1737 | 0.5141 | 0.8398 | |
| | 0.1371 | 6.9983 | 2026 | 0.2422 | 0.9291 | |
| | 0.1371 | 8.0 | 2316 | 0.1653 | 0.9486 | |
| | 0.1073 | 8.9983 | 2605 | 0.1606 | 0.9480 | |
| | 0.1073 | 10.0 | 2895 | 0.3522 | 0.8991 | |
| | 0.0966 | 10.9983 | 3184 | 0.2096 | 0.9309 | |
| | 0.0966 | 12.0 | 3474 | 0.1263 | 0.9664 | |
| | 0.0887 | 12.9983 | 3763 | 0.2030 | 0.9529 | |
| | 0.0935 | 14.0 | 4053 | 0.1045 | 0.9676 | |
| | 0.0935 | 14.9983 | 4342 | 0.1270 | 0.9664 | |
| | 0.0751 | 16.0 | 4632 | 0.1873 | 0.9596 | |
| | 0.0751 | 16.9983 | 4921 | 0.2181 | 0.9621 | |
| | 0.0644 | 18.0 | 5211 | 0.1207 | 0.9713 | |
| | 0.0589 | 18.9983 | 5500 | 0.3134 | 0.9315 | |
| | 0.0589 | 20.0 | 5790 | 0.2447 | 0.9505 | |
| | 0.0451 | 20.9983 | 6079 | 0.2650 | 0.9474 | |
| | 0.0451 | 22.0 | 6369 | 0.2205 | 0.9596 | |
| | 0.0414 | 22.9983 | 6658 | 0.1899 | 0.9657 | |
| | 0.0414 | 24.0 | 6948 | 0.2518 | 0.9590 | |
| | 0.0415 | 24.9983 | 7237 | 0.2175 | 0.9572 | |
| | 0.0358 | 26.0 | 7527 | 0.3080 | 0.9462 | |
| | 0.0358 | 26.9983 | 7816 | 0.2570 | 0.9474 | |
| | 0.0332 | 28.0 | 8106 | 0.2519 | 0.9554 | |
| | 0.0332 | 28.9983 | 8395 | 0.3117 | 0.9492 | |
| | 0.028 | 30.0 | 8685 | 0.3270 | 0.9517 | |
| | 0.028 | 30.9983 | 8974 | 0.2641 | 0.9602 | |
| | 0.0281 | 32.0 | 9264 | 0.2669 | 0.9615 | |
| | 0.0227 | 32.9983 | 9553 | 0.2558 | 0.9615 | |
| | 0.0227 | 34.0 | 9843 | 0.3255 | 0.9505 | |
| | 0.0218 | 34.9983 | 10132 | 0.3818 | 0.9431 | |
| | 0.0218 | 36.0 | 10422 | 0.2411 | 0.9657 | |
| | 0.0224 | 36.9983 | 10711 | 0.2391 | 0.9645 | |
| | 0.0201 | 38.0 | 11001 | 0.3097 | 0.9602 | |
| | 0.0201 | 38.9983 | 11290 | 0.3057 | 0.9590 | |
| | 0.0168 | 40.0 | 11580 | 0.2537 | 0.9621 | |
| | 0.0168 | 40.9983 | 11869 | 0.2661 | 0.9615 | |
| | 0.0171 | 42.0 | 12159 | 0.3151 | 0.9590 | |
| | 0.0171 | 42.9983 | 12448 | 0.2814 | 0.9621 | |
| | 0.0176 | 44.0 | 12738 | 0.2748 | 0.9633 | |
| | 0.0153 | 44.9983 | 13027 | 0.2950 | 0.9633 | |
| | 0.0153 | 46.0 | 13317 | 0.3171 | 0.9596 | |
| | 0.0133 | 46.9983 | 13606 | 0.2659 | 0.9633 | |
| | 0.0133 | 48.0 | 13896 | 0.3022 | 0.9633 | |
| | 0.0142 | 48.9983 | 14185 | 0.3028 | 0.9609 | |
| | 0.0142 | 49.9136 | 14450 | 0.2840 | 0.9627 | |
| |
| |
| ### Framework versions |
| |
| - Transformers 4.40.0 |
| - Pytorch 2.2.2+cu121 |
| - Datasets 2.19.0 |
| - Tokenizers 0.19.1 |
| |
