Safetensors
qwen2

merge strategy

#1
by sorrymakerrr - opened

Thank you for your open source. Can you share the strategy for model merge? For example, the configuration of mergekit.

Thank you for your open source. Can you share the strategy for model merge? For example, the configuration of mergekit.

models:
  - model: sft_model
  - model: dpo_model1
    parameters:
      density: 0.5
      weight: 0.6
  - model: dpo_model2
    parameters:
      density: 0.5
      weight: 0.5
merge_method: ties
base_model: sft_model
parameters:
  normalize: true
  int8_mask: true
dtype: bfloat16

Sign up or log in to comment