Views
No views yet
bc1qsvfduzj9fjs9fugpc52yver3f2g8fp7xjxecdvOpenYourMind/Qwopus3.5-122B-A10B-Kimi-K2.6-destill-healed-abliterated for Apple Silicon (LM Studio / mlx-vlm). See the parent repo for the full pipeline: refusal ablation → constrained-LoRA Opus reasoning SFT → unconstrained chosen-completion SFT → Kimi K2.6 reasoning DPO (≈3,000 distilled samples + synthetic data, improving reasoning verbosity on ~12% of requests and removing looping on 2–6% of long-tail conversations).mlx-lm / mlx-vlm, including LM Studio's bundled engine) consumes the qwen3.5 MTP head — they drop mtp.* on load — so it was omitted to keep the model loading cleanly. If/when an MLX runtime adds qwen3.5 MTP support, an MTP-enabled build can be produced from the full weights. (Note: MTP gave no measurable gain in our testing and would need retraining to be useful — see the parent card. For MTP today, use the GGUF build with llama.cpp.)| File | Description | Size |
|---|---|---|
model-*-of-00014.safetensors | 4-bit language weights + BF16 vision tower | ~65 GB total |
config.json | Qwen3_5MoeForConditionalGeneration + quantization (4-bit, g64) | — |
preprocessor_config.json, video_preprocessor_config.json, processor_config.json | Qwen3-VL image/video processor | — |
tokenizer*, chat_template.jinja, generation_config.json | Standard | — |
1pip install mlx-vlm
2python -m mlx_vlm.generate \
3 --model OpenYourMind/Qwopus3.5-122B-A10B-Kimi-K2.6-destill-healed-abliterated-MLX-4bit \
4 --image path/to/image.jpg \
5 --prompt "Describe this image." --max-tokens 256OpenYourMind) and load it with the MLX runtime.