This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
TIES merge method using
unsloth/Qwen3-14B as a base.
1models:
2 - model: unsloth/Qwen3-14B
3 parameters:
4 density: [1, 0.7, 0.1] # density gradient
5 weight: 1.0
6 - model: Qwen/Qwen3-14B
7 parameters:
8 density: 0.5
9 weight: [0, 0.3, 0.7, 1] # weight gradient
10 - model: bamec66557/Qwen3-14B-ADD
11 parameters:
12 density: 0.33
13 weight:
14 - filter: mlp
15 value: 0.5
16 - value: 0
17merge_method: ties
18base_model: unsloth/Qwen3-14B
19parameters:
20 normalize: true
21 int8_mask: true
22dtype: bfloat16
23
