MergeMonster / README.md
Gryphe's picture
Update README.md
2dbc41c
|
raw
history blame
No virus
829 Bytes
metadata
language:
  - en

MergeMonster

An unsupervised merging algorithm for Transformers-based language models, using a list of phrases (both good and bad) and a fully automated strategy that strives to decrease (or increase) the probability of these phrases occuring in the final merge.

Code available here

A place to store various promising quantizations of things produced by the Merge Monster.

Current variations

  • MergeMonster-7b - First model: Mistral 0.1, with the objective to reduce overused GPT style roleplaying phrases.
  • MergeMonster-13b - First model: MythoMax, with the objective to reduce overused GPT style roleplaying phrases.
  • MergeMonster-Decensored-7b - First model: Mistral 0.1, with the objective to reduce refusals on a global level.