Instructions to use Efficient-Large-Model/LongLive-Plug-Wan2.2-TI2V-5B-few-step with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Efficient-Large-Model/LongLive-Plug-Wan2.2-TI2V-5B-few-step with PEFT:
Task type is invalid.
- Inference
- Notebooks
- Google Colab
- Kaggle
Download inference_config.yaml from Efficient-Large-Model/LongLive-Plug-Wan2.2-TI2V-5B-few-step: direct link, hf CLI and curl.
- Browser
- Download file 586 Bytes
-
https://huggingface.co/Efficient-Large-Model/LongLive-Plug-Wan2.2-TI2V-5B-few-step/resolve/main/inference_config.yaml
- Command line
-
hf download hf://Efficient-Large-Model/LongLive-Plug-Wan2.2-TI2V-5B-few-step/inference_config.yaml
-
curl -L -o inference_config.yaml https://huggingface.co/Efficient-Large-Model/LongLive-Plug-Wan2.2-TI2V-5B-few-step/resolve/main/inference_config.yaml
586 Bytes
| # Minimal LongLive-LoRA overrides for this non-AR generator adapter. | |
| generator_is_causal: false | |
| use_ema: false | |
| merge_lora: false | |
| model_kwargs: | |
| model_name: Wan2.2-TI2V-5B | |
| timestep_shift: 5.0 | |
| num_frame_per_block: 8 | |
| local_attn_size: 32 | |
| checkpoints: | |
| generator_ckpt: null | |
| lora_ckpt: /path/to/generator_lora.pt | |
| inference: | |
| sampling_steps: 4 | |
| guidance_scale: 1.0 | |
| sink_size: 0 | |
| multi_shot_rope_offset: 0 | |
| adapter: | |
| type: lora | |
| rank: 128 | |
| alpha: 128 | |
| dropout: 0.0 | |
| verbose: true | |
| data: | |
| data_path: /path/to/inference_prompts | |
| image_or_video_shape: [1, 32, 48, 44, 80] | |