Instructions to use mlx-community/Bernini-v2-int4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/Bernini-v2-int4 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir Bernini-v2-int4 mlx-community/Bernini-v2-int4
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
| { | |
| "source": "ByteDance/Bernini-Diffusers-v2", | |
| "source_commit": "399cf6a18a4c523b367b2b1ac25a2a61009e7df3", | |
| "dtype": "bfloat16", | |
| "converted": "2026-08-18", | |
| "components": [ | |
| "experts", | |
| "planner", | |
| "t5+vae copied from mlx-community/Bernini-R-bf16 (bit-identical, AB-R-0097)" | |
| ], | |
| "note": "experts RETRAINED vs Bernini-R (AB-R-0097) -> full reconvert via premap_diffusers_to_wan + mlx-video sanitize, CPU-stream, chunked evals; in-checkpoint fp32 T5 shards 26-31 never downloaded; mllm saved in HF Qwen2.5-VL layout (mllm. prefix stripped); connector/mask_tokens keys verbatim upstream in planner_glue.safetensors" | |
| } |