README.md

metadata

base_model:
  - zhiyuanhucs/qwen2.5-7b-sequence-200-5
  - zhiyuanhucs/qwen2.5-7b-backward-reasoning-7B-level-2-0310
  - zhiyuanhucs/qwen2.5-7B-formula-1k-base-final-1-step280
library_name: transformers
tags:
  - mergekit
  - merge

nonlinear2

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the DARE TIES merge method using zhiyuanhucs/qwen2.5-7B-formula-1k-base-final-1-step280 as a base.

Models Merged

The following models were included in the merge:

• zhiyuanhucs/qwen2.5-7b-sequence-200-5
• zhiyuanhucs/qwen2.5-7b-backward-reasoning-7B-level-2-0310

Configuration

The following YAML configuration was used to produce this model:

# dare_ties_qwen.yml
models:
  - model: zhiyuanhucs/qwen2.5-7B-formula-1k-base-final-1-step280
    parameters:              # base 模型不需要 density
      weight: 1
  - model: zhiyuanhucs/qwen2.5-7b-sequence-200-5
    parameters:
      density: 0.6           # DARE‑TIES 推荐略高一点
      weight: 0.1
  - model: zhiyuanhucs/qwen2.5-7b-backward-reasoning-7B-level-2-0310
    parameters:
      density: 0.6
      weight: 0.2
merge_method: dare_ties
base_model: zhiyuanhucs/qwen2.5-7B-formula-1k-base-final-1-step280
parameters:
  normalize: false
  int8_mask: true
dtype: float16