| model_name: mistralai/Mistral-7B-v0.1 | |
| output_dir: ./models/mistral-mk-small | |
| epochs: 1 | |
| per_device_train_batch_size: 2 | |
| gradient_accumulation_steps: 4 | |
| max_steps: 500 | |
| learning_rate: 2e-4 | |
| load_in_4bit: true | |
| lora: | |
| r: 16 | |
| alpha: 16 | |
| dropout: 0.05 | |
| model_name: mistralai/Mistral-7B-v0.1 | |
| output_dir: ./models/mistral-mk-small | |
| epochs: 1 | |
| per_device_train_batch_size: 2 | |
| gradient_accumulation_steps: 4 | |
| max_steps: 500 | |
| learning_rate: 2e-4 | |
| load_in_4bit: true | |
| lora: | |
| r: 16 | |
| alpha: 16 | |
| dropout: 0.05 | |