| slices: | |
| - sources: | |
| - model: Qwen/Qwen2.5-72B-Instruct | |
| layer_range: | |
| - 0 | |
| - 80 | |
| - model: MaziyarPanahi/calme-2.1-qwen2.5-72b | |
| layer_range: | |
| - 0 | |
| - 80 | |
| merge_method: slerp | |
| base_model: Qwen/Qwen2.5-72B-Instruct | |
| parameters: | |
| t: | |
| - filter: self_attn | |
| value: 0.5 | |
| - filter: mlp | |
| value: 0.5 | |
| - value: 0.5 | |
| dtype: bfloat16 | |
| tensor_parallel_devices: auto | |
| allow_different_heads: true |