CoolFace
Modelpublic

alamios/Qwenstral-Small-3.1-0.5B

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
8likes33downloads
Model Card

Qwenstral-Small-3.1-0.5B

Qwen/Qwen2.5-0.5B, but with the vocab of mistralai/Mistral-Small-3.1-24B-Instruct-2503 / mistralai/Mistral-Small-24B-Instruct-2501 transplanted using transplant-vocab.

It can be used as draft model for Mistral-Small directly, but there is a more performant variant finetuned on Mistral's outputs: