RT @KyleHessling1: Qwopus 3.5 v3 weights are now available in FP8! I was even delegated some work on this one in coordination with Jackrong, so it could be up and running out of the box for vLLM! That way he could get back to his incredible work on "Gemopus" (Gemma 4-Opus 4.6 fine tune)...which is coming soon! You're free to run all of the various Quant methods on it at half the base model size! Please post them in the comments so we can test them! It should be production-ready for inference in vLLM! I ran several small tests to ensure readiness, but if you encounter any issues, please let us know! It should be pretty rock solid! Honored to have been able to contribute, even small, to Jackrongs great work! Please follow as we will be posting Gemopus here when it gets wrapped up, and the time is nigh! I'll be helping Jackrong with anything I can to make that happen, too! Hope you're all having an excellent morning! Follow Jackrong on huggingface too! huggingface.co/Jackrong/Qwop…

Mehr auf Arint.info

#huggingface #make #us #vLLM #arint_info

https://x.com/KyleHessling1/status/2041556849598832895#m