| base_model: google/gemma-3-4b-it | |
| merge_method: ties | |
| dtype: bfloat16 | |
| models: | |
| - model: google/gemma-3-4b-it | |
| parameters: | |
| weight: 0.4 | |
| parameters: | |
| weight: 0.2 | |
| - model: Qwen/Qwen2.5-3B-Instruct | |
| parameters: | |
| weight: 0.2 | |
| - model: Qwen/Qwen2.5-3B-Instruct | |
| parameters: | |
| weight: 0.2 | |
| parameters: | |
| normalize: true | |
| int8_mask: true | |