# @janbamjan — 2024-11-07

♥0 ↻0 · https://x.com/janbamjan/status/1854625238333497758

@elder_plinius I'm curious about Qwen-2.5.
I did some experiments using the raw text completion endpoint instead of the chat completion.
Llama and Hermes readily produced nfsw and m*th recipies, but Qwen-2.5 had seemingly bulletproof refusals.
Thoughts?

tags: author:janbamjan, kind:tweet, model:nous-hermes, on:nous-hermes, year:2024
cited on: _dossiers/nous-hermes.md, nous-hermes