Instructions to use Efficient-Large-Model/LongLive-Plug-Wan2.1-T2V-14B-cfg with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Efficient-Large-Model/LongLive-Plug-Wan2.1-T2V-14B-cfg with PEFT:
Task type is invalid.
- Inference
- Notebooks
- Google Colab
- Kaggle
File size: 875 Bytes
32b8aa3 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 | ---
license: apache-2.0
base_model: Wan-AI/Wan2.1-T2V-14B
base_model_relation: adapter
library_name: peft
pipeline_tag: text-to-video
model_name: LongLive-Plug-Wan2.1-T2V-14B-cfg
tags:
- lora
- distillation
- cfg-distillation
---
# LongLive-Plug-Wan2.1-T2V-14B-cfg
A LoRA adapter for [Wan2.1-T2V-14B](https://huggingface.co/Wan-AI/Wan2.1-T2V-14B) that distills classifier-free guidance into conditional-only video generation, reducing the need for a separate unconditional inference branch.
This adapter does not provide few-step acceleration on its own.
This is an adapter and requires the corresponding base model.
## Usage
Use with the matching [few-step LoRA](https://huggingface.co/Efficient-Large-Model/LongLive-Plug-Wan2.1-T2V-14B-few-step).
Recommended LoRA weights: **few-step : CFG = 1 : 0.5**.
These are adapter weights, not the inference CFG scale.
|