Instructions to use Perflow-Shuai/MiniMax-H3-3Step-LoRA with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Perflow-Shuai/MiniMax-H3-3Step-LoRA with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("Perflow-Shuai/MiniMax-H3-3Step-LoRA", 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
Download inference_config.json from Perflow-Shuai/MiniMax-H3-3Step-LoRA: direct link, hf CLI and curl.
- Browser
- Download file 937 Bytes
-
https://huggingface.co/Perflow-Shuai/MiniMax-H3-3Step-LoRA/resolve/main/inference_config.json
- Command line
-
hf download hf://Perflow-Shuai/MiniMax-H3-3Step-LoRA/inference_config.json
-
curl -L -o inference_config.json https://huggingface.co/Perflow-Shuai/MiniMax-H3-3Step-LoRA/resolve/main/inference_config.json
937 Bytes
| { | |
| "nfe_per_chunk": 3, | |
| "chunks": 8, | |
| "whole_clip_forwards": 24, | |
| "width": 1344, | |
| "height": 768, | |
| "frames": 124, | |
| "fps": 24, | |
| "audio_sample_rate": 32000, | |
| "audio_channels": 2, | |
| "cfg_scale": 1.0, | |
| "video_latent_edges": [ | |
| 0, | |
| 5, | |
| 10, | |
| 15, | |
| 20, | |
| 25, | |
| 30, | |
| 35, | |
| 37 | |
| ], | |
| "audio_latent_edges_per_channel": [ | |
| 0, | |
| 29, | |
| 57, | |
| 85, | |
| 114, | |
| 142, | |
| 170, | |
| 199, | |
| 207 | |
| ], | |
| "base_sigmas": [ | |
| 1, | |
| 0.6666666666666666, | |
| 0.3333333333333333, | |
| 0 | |
| ], | |
| "video_shift": 12, | |
| "audio_shift": 3, | |
| "video_sigmas_fp32": [ | |
| 1, | |
| 0.9599999785423279, | |
| 0.8571428656578064, | |
| 0 | |
| ], | |
| "audio_sigmas_fp32": [ | |
| 1, | |
| 0.8571428656578064, | |
| 0.6000000238418579, | |
| 0 | |
| ], | |
| "context_clean_clock": 0.999, | |
| "attention": "text + first chunk + previous two chunks + current chunk", | |
| "update": "clean = current + sigma * velocity; next = (1-next_sigma)*clean + next_sigma*noise" | |
| } | |