Resolving Interference When Merging Models
Paper
• 2306.01708 • Published
• 17
This is a merge of pre-trained language models created using mergekit.
This model was merged using the TIES merge method using bunnycore/Qwen-2.5-3b-Evol-CoT-V3 as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
- model: bunnycore/Qwen-2.5-3b-Evol-CoT-V3
parameters:
density: 0.5
weight: 0.5
- model: qnguyen3/raspberry-3B
parameters:
density: 0.5
weight: 0.5
- model: bunnycore/Qwen-2.5-3b-RP
parameters:
density: 0.5
weight: 0.5
- model: bunnycore/Replete-Qwen-2.5-3b-CoT-RP
parameters:
density: 0.5
weight: 0.5
merge_method: ties
base_model: bunnycore/Qwen-2.5-3b-Evol-CoT-V3
parameters:
normalize: false
int8_mask: true
dtype: float16