metadata

license: apache-2.0
language:
  - en
pipeline_tag: text-generation
inference: false
tags:
  - dare
  - super mario merge
  - pytorch
  - mixtral
  - merge

mixtral megamerge 8x7b v1

The following models were merged with DARE using https://github.com/martyn/safetensors-merge-supermario

Mergelist

cognitivecomputations/dolphin-2.6-mixtral-8x7b
mistralai/Mixtral-8x7B-v0.1
mistralai/Mixtral-8x7B-Instruct-v0.1

Merge command

python hf_merge.py mergelist.txt mixtral-merge-1 -p 0.1 -lambda 1.95

Notes

seems to generalize instruct styles
MoE gates are not modified