peft-deepseek-code-lora
This model is a fine-tuned version of deepseek-ai/deepseek-coder-1.3b-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.7771
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: 0.0005
- train_batch_size: 12
- eval_batch_size: 12
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 45
- training_steps: 3000
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.869 | 0.0333 | 100 | 0.8371 |
| 0.8608 | 0.0667 | 200 | 0.7918 |
| 0.7746 | 0.1 | 300 | 0.7638 |
| 0.7381 | 0.1333 | 400 | 0.7487 |
| 0.7078 | 0.1667 | 500 | 0.7371 |
| 0.7066 | 0.2 | 600 | 0.7261 |
| 0.6709 | 0.2333 | 700 | 0.7235 |
| 0.6487 | 0.2667 | 800 | 0.7191 |
| 0.6103 | 0.3 | 900 | 0.7196 |
| 0.6109 | 0.3333 | 1000 | 0.7197 |
| 0.5804 | 0.3667 | 1100 | 0.7112 |
| 0.563 | 0.4 | 1200 | 0.7162 |
| 0.5406 | 0.4333 | 1300 | 0.7157 |
| 0.5286 | 0.4667 | 1400 | 0.7256 |
| 0.4839 | 0.5 | 1500 | 0.7208 |
| 0.5268 | 0.5333 | 1600 | 0.7258 |
| 0.4565 | 0.5667 | 1700 | 0.7280 |
| 0.4366 | 0.6 | 1800 | 0.7298 |
| 0.4729 | 0.6333 | 1900 | 0.7393 |
| 0.4451 | 0.6667 | 2000 | 0.7463 |
| 0.4008 | 0.7 | 2100 | 0.7533 |
| 0.3915 | 0.7333 | 2200 | 0.7609 |
| 0.3769 | 0.7667 | 2300 | 0.7601 |
| 0.3776 | 0.8 | 2400 | 0.7671 |
| 0.3896 | 0.8333 | 2500 | 0.7694 |
| 0.3798 | 0.8667 | 2600 | 0.7727 |
| 0.3683 | 0.9 | 2700 | 0.7756 |
| 0.36 | 0.9333 | 2800 | 0.7774 |
| 0.3713 | 0.9667 | 2900 | 0.7769 |
| 0.352 | 1.0 | 3000 | 0.7771 |
Framework versions
- PEFT 0.11.1
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.14.6
- Tokenizers 0.19.1
- Downloads last month
- 49
Model tree for liyingjian/peft-deepseek-code-lora
Base model
deepseek-ai/deepseek-coder-1.3b-base