| --- |
| license: mit |
| base_model: microsoft/deberta-v3-large |
| tags: |
| - generated_from_trainer |
| metrics: |
| - accuracy |
| model-index: |
| - name: Classifier_30k |
| 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_30k |
| |
| 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: 0.1296 |
| - Accuracy: 0.9876 |
| |
| ## 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 | |
| |:-------------:|:-------:|:-----:|:---------------:|:--------:| |
| | 0.3588 | 0.9994 | 831 | 0.3084 | 0.9091 | |
| | 0.1252 | 2.0 | 1663 | 0.2260 | 0.9453 | |
| | 0.1123 | 2.9994 | 2494 | 0.1241 | 0.9604 | |
| | 0.0896 | 4.0 | 3326 | 0.1372 | 0.9655 | |
| | 0.0749 | 4.9994 | 4157 | 0.1541 | 0.9708 | |
| | 0.0743 | 6.0 | 4989 | 0.1127 | 0.9715 | |
| | 0.0596 | 6.9994 | 5820 | 0.1782 | 0.9672 | |
| | 0.0494 | 8.0 | 6652 | 0.1352 | 0.9749 | |
| | 0.0443 | 8.9994 | 7483 | 0.1232 | 0.9681 | |
| | 0.0405 | 10.0 | 8315 | 0.0756 | 0.9838 | |
| | 0.0383 | 10.9994 | 9146 | 0.2025 | 0.9600 | |
| | 0.0361 | 12.0 | 9978 | 0.1130 | 0.9796 | |
| | 0.0288 | 12.9994 | 10809 | 0.0906 | 0.9855 | |
| | 0.0249 | 14.0 | 11641 | 0.1122 | 0.9827 | |
| | 0.0222 | 14.9994 | 12472 | 0.0713 | 0.9862 | |
| | 0.0239 | 16.0 | 13304 | 0.0552 | 0.9876 | |
| | 0.0234 | 16.9994 | 14135 | 0.0728 | 0.9864 | |
| | 0.0258 | 18.0 | 14967 | 0.0558 | 0.9891 | |
| | 0.0208 | 18.9994 | 15798 | 0.0715 | 0.9879 | |
| | 0.0199 | 20.0 | 16630 | 0.0753 | 0.9885 | |
| | 0.0143 | 20.9994 | 17461 | 0.0812 | 0.9872 | |
| | 0.0255 | 22.0 | 18293 | 0.1661 | 0.9744 | |
| | 0.0156 | 22.9994 | 19124 | 0.0751 | 0.9883 | |
| | 0.013 | 24.0 | 19956 | 0.0718 | 0.9862 | |
| | 0.0126 | 24.9994 | 20787 | 0.0829 | 0.9853 | |
| | 0.0123 | 26.0 | 21619 | 0.0848 | 0.9857 | |
| | 0.0109 | 26.9994 | 22450 | 0.0913 | 0.9864 | |
| | 0.0095 | 28.0 | 23282 | 0.1607 | 0.9774 | |
| | 0.0096 | 28.9994 | 24113 | 0.0958 | 0.9853 | |
| | 0.0074 | 30.0 | 24945 | 0.1264 | 0.9857 | |
| | 0.0091 | 30.9994 | 25776 | 0.1030 | 0.9881 | |
| | 0.0096 | 32.0 | 26608 | 0.0954 | 0.9879 | |
| | 0.0074 | 32.9994 | 27439 | 0.1103 | 0.9885 | |
| | 0.0067 | 34.0 | 28271 | 0.1803 | 0.9791 | |
| | 0.0044 | 34.9994 | 29102 | 0.1597 | 0.9817 | |
| | 0.0045 | 36.0 | 29934 | 0.0878 | 0.9894 | |
| | 0.0034 | 36.9994 | 30765 | 0.1680 | 0.9806 | |
| | 0.0066 | 38.0 | 31597 | 0.1114 | 0.9870 | |
| | 0.0041 | 38.9994 | 32428 | 0.0910 | 0.9896 | |
| | 0.0043 | 40.0 | 33260 | 0.1435 | 0.9840 | |
| | 0.0037 | 40.9994 | 34091 | 0.1233 | 0.9881 | |
| | 0.0046 | 42.0 | 34923 | 0.1347 | 0.9864 | |
| | 0.0029 | 42.9994 | 35754 | 0.1134 | 0.9883 | |
| | 0.0017 | 44.0 | 36586 | 0.1125 | 0.9879 | |
| | 0.0025 | 44.9994 | 37417 | 0.1400 | 0.9859 | |
| | 0.0023 | 46.0 | 38249 | 0.1228 | 0.9879 | |
| | 0.0017 | 46.9994 | 39080 | 0.1445 | 0.9862 | |
| | 0.0011 | 48.0 | 39912 | 0.1375 | 0.9876 | |
| | 0.0013 | 48.9994 | 40743 | 0.1323 | 0.9876 | |
| | 0.0021 | 49.9699 | 41550 | 0.1296 | 0.9876 | |
|
|
|
|
| ### Framework versions |
|
|
| - Transformers 4.40.0 |
| - Pytorch 2.2.2+cu121 |
| - Datasets 2.19.0 |
| - Tokenizers 0.19.1 |
|
|
