| base_model: meta-llama/Llama-3.1-8B | |
| dtype: float16 | |
| merge_method: dare_linear | |
| parameters: | |
| normalize: 1.0 | |
| slices: | |
| - sources: | |
| - layer_range: [0, 32] | |
| model: Alelcv27/llama3-8b-code-dpo | |
| parameters: | |
| weight: 0.5 | |
| - layer_range: [0, 32] | |
| model: Alelcv27/llama3-8b-math-dpo | |
| parameters: | |
| weight: 0.8 | |
| - layer_range: [0, 32] | |
| model: meta-llama/Llama-3.1-8B |