MedQA_L3_1000steps_1e6rate_SFT
This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.3666
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-06
- train_batch_size: 2
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- training_steps: 1000
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.0895 | 0.0489 | 50 | 0.8521 |
| 0.3865 | 0.0977 | 100 | 0.4119 |
| 0.4156 | 0.1466 | 150 | 0.3943 |
| 0.4302 | 0.1954 | 200 | 0.3870 |
| 0.3788 | 0.2443 | 250 | 0.3808 |
| 0.3964 | 0.2931 | 300 | 0.3773 |
| 0.3753 | 0.3420 | 350 | 0.3749 |
| 0.359 | 0.3908 | 400 | 0.3727 |
| 0.3874 | 0.4397 | 450 | 0.3711 |
| 0.3722 | 0.4885 | 500 | 0.3699 |
| 0.3615 | 0.5374 | 550 | 0.3686 |
| 0.3807 | 0.5862 | 600 | 0.3677 |
| 0.3643 | 0.6351 | 650 | 0.3673 |
| 0.3513 | 0.6839 | 700 | 0.3669 |
| 0.358 | 0.7328 | 750 | 0.3667 |
| 0.3648 | 0.7816 | 800 | 0.3666 |
| 0.3911 | 0.8305 | 850 | 0.3666 |
| 0.3475 | 0.8793 | 900 | 0.3666 |
| 0.3511 | 0.9282 | 950 | 0.3665 |
| 0.3673 | 0.9770 | 1000 | 0.3666 |
Framework versions
- Transformers 4.41.0
- Pytorch 2.0.0+cu117
- Datasets 2.19.1
- Tokenizers 0.19.1