| --- |
| base_model: microsoft/Phi-3-mini-4k-instruct |
| library_name: peft |
| license: mit |
| tags: |
| - trl |
| - sft |
| - generated_from_trainer |
| model-index: |
| - name: phi-3-mini-QLoRA |
| 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. --> |
|
|
| # phi-3-mini-QLoRA |
|
|
| This model is a fine-tuned version of [microsoft/Phi-3-mini-4k-instruct](https://huggingface.co/microsoft/Phi-3-mini-4k-instruct) on an unknown dataset. |
| It achieves the following results on the evaluation set: |
| - Loss: 0.5739 |
|
|
| ## 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: 0.0001 |
| - 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 |
| - lr_scheduler_warmup_ratio: 0.1 |
| - num_epochs: 1 |
|
|
| ### Training results |
|
|
| | Training Loss | Epoch | Step | Validation Loss | |
| |:-------------:|:------:|:----:|:---------------:| |
| | 1.0193 | 0.0181 | 10 | 1.0582 | |
| | 1.0408 | 0.0362 | 20 | 1.0129 | |
| | 0.9957 | 0.0543 | 30 | 0.9095 | |
| | 0.8047 | 0.0724 | 40 | 0.7711 | |
| | 0.699 | 0.0905 | 50 | 0.6689 | |
| | 0.6143 | 0.1085 | 60 | 0.6385 | |
| | 0.6472 | 0.1266 | 70 | 0.6175 | |
| | 0.6077 | 0.1447 | 80 | 0.6073 | |
| | 0.6028 | 0.1628 | 90 | 0.6012 | |
| | 0.5929 | 0.1809 | 100 | 0.5978 | |
| | 0.6117 | 0.1990 | 110 | 0.5948 | |
| | 0.5904 | 0.2171 | 120 | 0.5925 | |
| | 0.5852 | 0.2352 | 130 | 0.5909 | |
| | 0.5662 | 0.2533 | 140 | 0.5895 | |
| | 0.6183 | 0.2714 | 150 | 0.5880 | |
| | 0.5872 | 0.2895 | 160 | 0.5873 | |
| | 0.5807 | 0.3076 | 170 | 0.5863 | |
| | 0.6169 | 0.3256 | 180 | 0.5853 | |
| | 0.5705 | 0.3437 | 190 | 0.5841 | |
| | 0.6143 | 0.3618 | 200 | 0.5835 | |
| | 0.5705 | 0.3799 | 210 | 0.5828 | |
| | 0.5683 | 0.3980 | 220 | 0.5821 | |
| | 0.6077 | 0.4161 | 230 | 0.5818 | |
| | 0.586 | 0.4342 | 240 | 0.5811 | |
| | 0.5724 | 0.4523 | 250 | 0.5804 | |
| | 0.5941 | 0.4704 | 260 | 0.5799 | |
| | 0.5989 | 0.4885 | 270 | 0.5798 | |
| | 0.5582 | 0.5066 | 280 | 0.5793 | |
| | 0.5798 | 0.5246 | 290 | 0.5792 | |
| | 0.5545 | 0.5427 | 300 | 0.5785 | |
| | 0.597 | 0.5608 | 310 | 0.5783 | |
| | 0.6093 | 0.5789 | 320 | 0.5779 | |
| | 0.5736 | 0.5970 | 330 | 0.5778 | |
| | 0.5698 | 0.6151 | 340 | 0.5772 | |
| | 0.5659 | 0.6332 | 350 | 0.5769 | |
| | 0.5877 | 0.6513 | 360 | 0.5764 | |
| | 0.5837 | 0.6694 | 370 | 0.5763 | |
| | 0.5858 | 0.6875 | 380 | 0.5761 | |
| | 0.5877 | 0.7056 | 390 | 0.5760 | |
| | 0.5802 | 0.7237 | 400 | 0.5756 | |
| | 0.6009 | 0.7417 | 410 | 0.5754 | |
| | 0.5713 | 0.7598 | 420 | 0.5751 | |
| | 0.5509 | 0.7779 | 430 | 0.5751 | |
| | 0.5646 | 0.7960 | 440 | 0.5750 | |
| | 0.5458 | 0.8141 | 450 | 0.5748 | |
| | 0.5694 | 0.8322 | 460 | 0.5746 | |
| | 0.576 | 0.8503 | 470 | 0.5744 | |
| | 0.5864 | 0.8684 | 480 | 0.5742 | |
| | 0.5645 | 0.8865 | 490 | 0.5741 | |
| | 0.5531 | 0.9046 | 500 | 0.5742 | |
| | 0.6176 | 0.9227 | 510 | 0.5742 | |
| | 0.5987 | 0.9408 | 520 | 0.5742 | |
| | 0.5703 | 0.9588 | 530 | 0.5740 | |
| | 0.6023 | 0.9769 | 540 | 0.5740 | |
| | 0.5637 | 0.9950 | 550 | 0.5739 | |
|
|
|
|
| ### Framework versions |
|
|
| - PEFT 0.13.0 |
| - Transformers 4.45.1 |
| - Pytorch 2.4.1+cu121 |
| - Datasets 3.0.1 |
| - Tokenizers 0.20.0 |
