metadata
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: []
phi-3-mini-QLoRA
This model is a fine-tuned version of microsoft/Phi-3-mini-4k-instruct on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4084
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: 0.0002
- 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: 30
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.415 | 1.1765 | 5 | 1.4148 |
| 1.2791 | 2.3529 | 10 | 1.2542 |
| 1.0303 | 3.5294 | 15 | 0.9828 |
| 0.7989 | 4.7059 | 20 | 0.7193 |
| 0.5792 | 5.8824 | 25 | 0.5793 |
| 0.5074 | 7.0588 | 30 | 0.5133 |
| 0.4558 | 8.2353 | 35 | 0.4714 |
| 0.361 | 9.4118 | 40 | 0.4478 |
| 0.3751 | 10.5882 | 45 | 0.4236 |
| 0.2908 | 11.7647 | 50 | 0.4106 |
| 0.263 | 12.9412 | 55 | 0.3855 |
| 0.2515 | 14.1176 | 60 | 0.3760 |
| 0.2391 | 15.2941 | 65 | 0.3752 |
| 0.1973 | 16.4706 | 70 | 0.3723 |
| 0.1638 | 17.6471 | 75 | 0.3740 |
| 0.1776 | 18.8235 | 80 | 0.3868 |
| 0.2008 | 20.0 | 85 | 0.3798 |
| 0.1569 | 21.1765 | 90 | 0.3848 |
| 0.1284 | 22.3529 | 95 | 0.3901 |
| 0.1171 | 23.5294 | 100 | 0.3969 |
| 0.1364 | 24.7059 | 105 | 0.3950 |
| 0.1401 | 25.8824 | 110 | 0.4070 |
| 0.1195 | 27.0588 | 115 | 0.4091 |
| 0.1219 | 28.2353 | 120 | 0.4084 |
Framework versions
- PEFT 0.13.2
- Transformers 4.45.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.0