How to use splats/Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled-oQ8e with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled-oQ8e splats/Qwen3.6-35B-A3B-Claude-4.7-Opus-Reasoning-Distilled-oQ8e
I use M1 macbook.. so, I need fp16(dtype) oQ8e model to make prefill speed faster!https://github.com/jundot/omlx/issues/604
Just uploaded many FP16 Distilled versions of Qwen3.6-35B
thank you!!🙏
· Sign up or log in to comment
