Qwen3-4B-Base-Continued-GRP…/mergekit_config.yml

# TIES merge: Judge as base, inject sparse GRPO knowledge
merge_method: ties
base_model: ./merged_models/llm-judge-merged-fixed
models:
  - model: ./merged_models/grpo-cabs
    parameters:
      density: 0.5
      weight: 0.4
dtype: bfloat16
初始化项目，由ModelHub XC社区提供模型 Model: Lambent/Qwen3-4B-Base-Continued-GRPO-Merge Source: Original Platform 2026-06-02 03:31:25 +08:00			`# TIES merge: Judge as base, inject sparse GRPO knowledge`
			`merge_method: ties`
			`base_model: ./merged_models/llm-judge-merged-fixed`
			`models:`
			`- model: ./merged_models/grpo-cabs`
			`parameters:`
			`density: 0.5`
			`weight: 0.4`
			`dtype: bfloat16`