DPO fine-tuned Mistral 7B beats Llama 70B on MT Bench

3 points by clmnt 1 year ago | 3 comments
  • amilios 1 year ago
    Can anyone corroborate this anecdotally? I.e. has anyone actually looked at the output of the two models side-by-side for common tasks? There's lots of talks these days about academic benchmarks being pretty "broken" for modern LMs, and not really properly showcasing the differences between models. I wonder if that's the case here or if the model is genuinely better.
    • brucethemoose2 1 year ago
      I dunno. I totally missed this and will check it out.

      - Huggingface is less likely to "cheat" by training on tests than other orgs, I think.

      - Some finetunes are really good at a particular test (like XWin). This isnt necessarily a bad thing, if they are good at a specific niche.

    • brucethemoose2 1 year ago
      > <|system|>, <|user|> and <|model|>.

      Oh hey, thats almost Metharme's format.

      It must originate from an older model, as most new models dont use that syntax.