This model is surprisingly usable, despite being an abomination.
Mergekit GUI, although it is my own duplicate, uploading to Stupidity-AI instead.
No clue.
Me.
USA.
Check the "last updated" field, stupid.
This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
Task Arithmetic merge method using
Qwen/Qwen2.5-0.5B-Instruct as a base.
1base_model: Qwen/Qwen2.5-0.5B-Instruct
2dtype: bfloat16
3merge_method: task_arithmetic
4parameters:
5 normalize: true
6slices:
7- sources:
8 - layer_range: [0, 24]
9 model: Qwen/Qwen2.5-0.5B-Instruct
10 - layer_range: [0, 24]
11 model: Qwen/Qwen2.5-0.5B-Instruct
12 parameters:
13 weight: [1.0, -1.0, 1.0, -1.0, 1.0]
14 - layer_range: [0, 24]
15 model: Qwen/Qwen2-0.5B-Instruct
16 parameters:
17 weight: [-1.0, 1.0, -1.0, 1.0, -1.0]
18
