1base_model: meta-llama/Llama-3.2-1B
2dtype: bfloat16
3merge_method: sce
4modules:
5 default:
6 slices:
7 - sources:
8 - layer_range: [0, 16]
9 model: Alelcv27/llama3-1b-math-dpo
10 - layer_range: [0, 16]
11 model: Alelcv27/llama3-1b-code-dpo
12 - layer_range: [0, 16]
13 model: meta-llama/Llama-3.2-1B
14parameters:
15 select_topk: 0.8