Jogg-Avatar 14B
Jogg-Avatar 14B is an audio-driven 720p avatar video generation model based on Wan2.1-T2V-14B. It adds audio conditioning and LoRA adapters to the Wan video diffusion model.
Source code and complete inference instructions: chanjing-ai/Jogg-Avatar
Checkpoint
| Directory | Base model | Parameters stored | Weight dtype |
|---|---|---|---|
Jogg-Avatar-14B/ |
Wan2.1-T2V-14B | Audio modules, input projection, and LoRA adapters | BF16 |
The Wan2.1 base model and Wav2Vec audio encoder are not duplicated here. Download them separately:
Repository Layout
.
├── Jogg-Avatar-14B/
│ ├── config.json
│ └── diffusion_pytorch_model.safetensors
├── LICENSE
├── README.md
└── SHA256SUMS
Download
mkdir -p models
hf download cicada-ai/jogg-avatar \
--include "Jogg-Avatar-14B/*" \
--local-dir models
Then follow the inference guide.
中文说明
本仓库只提供基于 Wan2.1-T2V-14B 的 Jogg-Avatar 14B 音频驱动权重。 权重包含音频条件模块、输入投影和 LoRA 参数，不重复分发 Wan2.1 基座模型与 Wav2Vec。完整环境、模型目录和推理命令请参考 开源代码仓库。
License
Released under the Apache License 2.0. The license and usage terms of the Wan2.1 base model also apply.
- Downloads last month
- -
Model tree for cicada-ai/jogg-avatar
Base model
Wan-AI/Wan2.1-T2V-14B