Instructions to use BornSaint/minimax-music3-latent-refiner-v0.10 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use BornSaint/minimax-music3-latent-refiner-v0.10 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("BornSaint/minimax-music3-latent-refiner-v0.10", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
ComfyUI node
Install from the directory containing ComfyUI:
git lfs install
git clone https://huggingface.co/terminusresearch/minimax-music3-latent-refiner-v0.10
ComfyUI/venv/bin/python -m pip install -e minimax-music3-latent-refiner-v0.10
ln -s "$PWD/minimax-music3-latent-refiner-v0.10/comfyui_node" \
ComfyUI/custom_nodes/minimax_music3_latent_refiner
Restart ComfyUI. Load comfyui_workflow_example.json from the repository root.
MiniMax Music3 Latent Refine uses the encoder bundled with the refiner loader when audio_vae_encoder is not connected. A connected AUDIO_VAE_ENCODER always takes precedence. A stock MiniMax Music3 VAE may be connected to audio_vae_decoder; otherwise the bundled full DAV decodes the result.