This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
TIES merge method using
Qwen/Qwen2.5-32B as a base.
1models:
2 - model: unsloth/DeepSeek-R1-Distill-Qwen-32B
3 parameters:
4 weight: 0.65
5 density: 0.65
6 - model: Qwen/QwQ-32B-Preview
7 parameters:
8 weight: 0.35
9 density: 0.35
10merge_method: ties
11base_model: Qwen/Qwen2.5-32B
12parameters:
13 normalize: true
14 int8_mask: true
15tokenizer_source: "unsloth/DeepSeek-R1-Distill-Qwen-32B"
16dtype: float16
