Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch
Paper • 2311.03099 • Published • 30
This is a merge of pre-trained language models created using mergekit.
This model was merged using the DARE TIES merge method using merged_model as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
- model: "merged_model"
- model: Qwen/Qwen3-30B-A3B-Thinking-2507
parameters:
density: 0.35
weight: 0.35
- model: unsloth/Qwen3-Coder-30B-A3B-Instruct
parameters:
density: 0.25
weight: 0.25
merge_method: dare_ties
base_model: "merged_model"
parameters:
int8_mask: true
dtype: bfloat16