This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
DARE TIES merge method using
Qwen/Qwen3-4B-Instruct-2507 as a base.
1base_model: Qwen/Qwen3-4B-Instruct-2507
2dtype: float16
3merge_method: dare_ties
4modules:
5 default:
6 slices:
7 - sources:
8 - layer_range: [0, 36]
9 model: Alelcv27/Qwen3-4B-INST-Math
10 parameters:
11 density: 0.8
12 weight: 0.7
13 - layer_range: [0, 36]
14 model: Alelcv27/Qwen3-4B-INST-Code
15 parameters:
16 density: 0.6
17 weight: 0.5
18 - layer_range: [0, 36]
19 model: Qwen/Qwen3-4B-Instruct-2507
20parameters:
21 normalize: 1.0
