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 Qwen/Qwen2.5-0.5B-Instruct as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
slices:
- sources:
- model: Qwen/Qwen2.5-0.5B-Instruct
layer_range: [0, 11]
- model: Qwen/Qwen2.5-0.5B-Instruct
layer_range: [12, 23]
merge_method: ties
base_model: Qwen/Qwen2.5-0.5B-Instruct
parameters:
t:
- filter: self_attn
value: [0.1, 0.3, 0.5, 0.7, 0.9]
- filter: mlp
value: [0.1, 0.3, 0.5, 0.7, 0.9]
- value: 0.5
dtype: bfloat16