TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

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

3 pointsby clmntover 1 year ago

2 comments

amiliosover 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.
评论 #37849556 未加载
brucethemoose2over 1 year ago
&gt; &lt;|system|&gt;, &lt;|user|&gt; and &lt;|model|&gt;.<p>Oh hey, thats almost Metharme&#x27;s format.<p>It must originate from an older model, as most new models dont use that syntax.