FuseChat: Knowledge Fusion of Chat Models
Paper
•
2408.07990
•
Published
•
14
This is a merge of pre-trained language models created using mergekit.
This model was merged using the SCE merge method using Pedro13543/mega_blend_model as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
# Pivot model
- model: Pedro13543/mega_blend_model
# Target models
- model: Pedro13543/experimental_merge_no2
- model: Pedro13543/good_mix_r1
merge_method: sce
base_model: Pedro13543/mega_blend_model
parameters:
select_topk: 1.0
dtype: bfloat16