chat_600STEPS_1e8rate_SFT
This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.6169
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: 1e-08
- train_batch_size: 4
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- training_steps: 600
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.6205 | 0.1 | 50 | 1.6173 |
| 1.5976 | 0.2 | 100 | 1.6169 |
| 1.6086 | 0.29 | 150 | 1.6172 |
| 1.6093 | 0.39 | 200 | 1.6169 |
| 1.618 | 0.49 | 250 | 1.6169 |
| 1.6087 | 0.59 | 300 | 1.6168 |
| 1.6166 | 0.68 | 350 | 1.6168 |
| 1.6135 | 0.78 | 400 | 1.6170 |
| 1.6054 | 0.88 | 450 | 1.6169 |
| 1.6162 | 0.98 | 500 | 1.6169 |
| 1.6052 | 1.07 | 550 | 1.6169 |
| 1.6057 | 1.17 | 600 | 1.6169 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.0.0+cu117
- Datasets 2.17.0
- Tokenizers 0.15.2