| | --- |
| | license: mit |
| | base_model: microsoft/phi-2 |
| | tags: |
| | - generated_from_trainer |
| | model-index: |
| | - name: V0414H2 |
| | 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. --> |
| |
|
| | # V0414H2 |
| |
|
| | This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on an unknown dataset. |
| | It achieves the following results on the evaluation set: |
| | - Loss: 0.0460 |
| |
|
| | ## 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: 0.003 |
| | - train_batch_size: 8 |
| | - eval_batch_size: 8 |
| | - seed: 42 |
| | - gradient_accumulation_steps: 8 |
| | - total_train_batch_size: 64 |
| | - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
| | - lr_scheduler_type: cosine_with_restarts |
| | - lr_scheduler_warmup_steps: 100 |
| | - num_epochs: 3 |
| | - mixed_precision_training: Native AMP |
| |
|
| | ### Training results |
| |
|
| | | Training Loss | Epoch | Step | Validation Loss | |
| | |:-------------:|:-----:|:----:|:---------------:| |
| | | 2.149 | 0.05 | 10 | 1.1737 | |
| | | 0.4718 | 0.09 | 20 | 0.1474 | |
| | | 0.1443 | 0.14 | 30 | 0.1201 | |
| | | 0.1135 | 0.18 | 40 | 0.0995 | |
| | | 0.0994 | 0.23 | 50 | 0.0855 | |
| | | 0.1011 | 0.27 | 60 | 0.0916 | |
| | | 0.0941 | 0.32 | 70 | 0.0861 | |
| | | 0.097 | 0.36 | 80 | 0.0793 | |
| | | 0.0799 | 0.41 | 90 | 0.0790 | |
| | | 0.0896 | 0.45 | 100 | 0.0803 | |
| | | 0.0904 | 0.5 | 110 | 0.0796 | |
| | | 0.0918 | 0.54 | 120 | 0.0736 | |
| | | 0.081 | 0.59 | 130 | 0.0717 | |
| | | 0.0785 | 0.63 | 140 | 0.0730 | |
| | | 0.0765 | 0.68 | 150 | 0.0761 | |
| | | 0.0823 | 0.73 | 160 | 0.0772 | |
| | | 0.0809 | 0.77 | 170 | 0.0706 | |
| | | 0.0836 | 0.82 | 180 | 0.0858 | |
| | | 0.0991 | 0.86 | 190 | 0.0790 | |
| | | 0.0788 | 0.91 | 200 | 0.0770 | |
| | | 0.0939 | 0.95 | 210 | 0.0734 | |
| | | 0.0891 | 1.0 | 220 | 0.0741 | |
| | | 0.0704 | 1.04 | 230 | 0.0831 | |
| | | 0.0833 | 1.09 | 240 | 0.0766 | |
| | | 0.0777 | 1.13 | 250 | 0.0752 | |
| | | 0.0723 | 1.18 | 260 | 0.0763 | |
| | | 0.0803 | 1.22 | 270 | 0.0738 | |
| | | 0.0694 | 1.27 | 280 | 0.0800 | |
| | | 0.0894 | 1.31 | 290 | 0.0728 | |
| | | 0.0891 | 1.36 | 300 | 0.0888 | |
| | | 0.088 | 1.41 | 310 | 0.0712 | |
| | | 0.0759 | 1.45 | 320 | 0.0709 | |
| | | 0.0876 | 1.5 | 330 | 0.0695 | |
| | | 0.0741 | 1.54 | 340 | 0.0770 | |
| | | 0.0805 | 1.59 | 350 | 0.0683 | |
| | | 0.0713 | 1.63 | 360 | 0.0739 | |
| | | 0.0884 | 1.68 | 370 | 0.0816 | |
| | | 0.083 | 1.72 | 380 | 0.0620 | |
| | | 0.0665 | 1.77 | 390 | 0.0613 | |
| | | 0.0718 | 1.81 | 400 | 0.0674 | |
| | | 0.0655 | 1.86 | 410 | 0.0551 | |
| | | 0.0517 | 1.9 | 420 | 0.0510 | |
| | | 0.0532 | 1.95 | 430 | 0.0491 | |
| | | 0.0537 | 1.99 | 440 | 0.0470 | |
| | | 0.0483 | 2.04 | 450 | 0.0474 | |
| | | 0.0337 | 2.08 | 460 | 0.0513 | |
| | | 0.0361 | 2.13 | 470 | 0.0566 | |
| | | 0.0464 | 2.18 | 480 | 0.0654 | |
| | | 0.0509 | 2.22 | 490 | 0.0524 | |
| | | 0.0425 | 2.27 | 500 | 0.0502 | |
| | | 0.0421 | 2.31 | 510 | 0.0500 | |
| | | 0.0351 | 2.36 | 520 | 0.0512 | |
| | | 0.039 | 2.4 | 530 | 0.0486 | |
| | | 0.0477 | 2.45 | 540 | 0.0484 | |
| | | 0.0425 | 2.49 | 550 | 0.0459 | |
| | | 0.0461 | 2.54 | 560 | 0.0455 | |
| | | 0.0413 | 2.58 | 570 | 0.0460 | |
| | | 0.0546 | 2.63 | 580 | 0.0455 | |
| | | 0.0344 | 2.67 | 590 | 0.0455 | |
| | | 0.0362 | 2.72 | 600 | 0.0458 | |
| | | 0.0373 | 2.76 | 610 | 0.0459 | |
| | | 0.0379 | 2.81 | 620 | 0.0461 | |
| | | 0.0459 | 2.86 | 630 | 0.0461 | |
| | | 0.0386 | 2.9 | 640 | 0.0460 | |
| | | 0.0346 | 2.95 | 650 | 0.0460 | |
| | | 0.0277 | 2.99 | 660 | 0.0460 | |
| |
|
| |
|
| | ### Framework versions |
| |
|
| | - Transformers 4.36.0.dev0 |
| | - Pytorch 2.1.2+cu121 |
| | - Datasets 2.14.6 |
| | - Tokenizers 0.14.1 |
| |
|
