peft-lora-starcoder1B-v2-personal-copilot-A100-40GB-yfw-from-yjx
This model is a fine-tuned version of bigcode/starcoderbase-1b on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.9260
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: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 30
- training_steps: 2000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.4846 | 0.05 | 100 | 0.4749 |
0.4216 | 0.1 | 200 | 0.4329 |
0.39 | 0.15 | 300 | 0.4452 |
0.3364 | 0.2 | 400 | 0.5184 |
0.2917 | 0.25 | 500 | 0.5963 |
0.2736 | 0.3 | 600 | 0.6457 |
0.2636 | 0.35 | 700 | 0.6698 |
0.2512 | 0.4 | 800 | 0.7002 |
0.2384 | 0.45 | 900 | 0.7437 |
0.2253 | 0.5 | 1000 | 0.7726 |
0.2132 | 0.55 | 1100 | 0.8081 |
0.2041 | 0.6 | 1200 | 0.8356 |
0.197 | 0.65 | 1300 | 0.8593 |
0.192 | 0.7 | 1400 | 0.8862 |
0.187 | 0.75 | 1500 | 0.8862 |
0.1829 | 0.8 | 1600 | 0.9074 |
0.1817 | 0.85 | 1700 | 0.9207 |
0.179 | 0.9 | 1800 | 0.9225 |
0.1787 | 0.95 | 1900 | 0.9243 |
0.1779 | 1.0 | 2000 | 0.9260 |
Framework versions
- PEFT 0.7.1
- Transformers 4.37.0
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.1
- Downloads last month
- 8
Model tree for scnuyjx/peft-lora-starcoder1B-v2-personal-copilot-A100-40GB-yfw-from-yjx
Base model
bigcode/starcoderbase-1b