| --- |
| license: other |
| license_name: minimax-h3-community |
| license_link: LICENSE |
| base_model: MiniMaxAI/MiniMax-H3 |
| library_name: minimax-h3 |
| tags: |
| - video-generation |
| - text-to-audio-video |
| - minimax-h3 |
| - lora |
| - distillation |
| - dmd2 |
| - few-step |
| - fastvideo |
| - fasth3 |
| - preview |
| --- |
| |
| <p align="center"> |
| <a href="https://github.com/hao-ai-lab/FastVideo"><img src="https://raw.githubusercontent.com/hao-ai-lab/FastVideo/main/assets/logos/logo.svg" width="320" alt="FastVideo"></a> |
| </p> |
|
|
| # FastVideo-FastH3-4-step-Preview-v1-LoRA |
|
|
| The compact LoRA releases for FastH3 Preview v1 from |
| [FastVideo](https://github.com/hao-ai-lab/FastVideo). Each adapter reconstructs |
| one four-forward FastH3 transformer from the MiniMax H3 base model. |
|
|
| [Blog](https://haoailab.com/blogs/fasth3-preview/) 路 |
| [Recommended full checkpoint](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree) 路 |
| [FastH3 collection](https://huggingface.co/collections/FastVideo/fastvideo-fasth3) |
|
|
| > The three VSA adapters require FastVideo's VSA-H3 backend and kernel. Use |
| > FastVideo's launchers rather than a generic PEFT loader; these adapters also |
| > contain exact delta and VSA gate tensors. |
|
|
| ## Run the recommended adapter |
|
|
| Install [uv](https://docs.astral.sh/uv/getting-started/installation/), then use |
| the CUDA 13 / Blackwell path below. It selects FastVideo's published CUDA |
| kernel wheel instead of compiling the kernel locally. See the |
| [installation guide](https://hao-ai-lab.github.io/FastVideo/getting_started/installation/) |
| for other platforms. |
|
|
| ```bash |
| git clone https://github.com/hao-ai-lab/FastVideo.git |
| cd FastVideo |
| uv venv --python 3.12 --seed |
| source .venv/bin/activate |
| UV_TORCH_BACKEND=cu130 uv pip install \ |
| --no-sources-package fastvideo-kernel \ |
| -e ".[fasth3]" |
| ``` |
|
|
| ```bash |
| bash examples/inference/basic/run_fasth3_lora_preview_vsa_datafree.sh \ |
| --prompt "your prompt" \ |
| --no-warmup \ |
| --repeats 1 |
| ``` |
|
|
| Set `FASTH3_LORA_STRENGTH` to change the adapter strength from its default of |
| `1.0`. The tested defaults use four B200 GPUs. On other multi-GPU CUDA |
| systems, follow the installation guide and add |
| `--no-replicated-dit --vsa-kernel triton --no-fa4` to a VSA launcher. The GPU |
| count must divide H3's 56 attention heads. |
|
|
| ## Variants |
|
|
| | Adapter | Full checkpoint | FastVideo launcher | |
| |---|---|---| |
| | [VSA / Data-Free](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA/tree/main/vsa-datafree) | [VSA / Data-Free](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree) | `run_fasth3_lora_preview_vsa_datafree.sh` | |
| | [VSA / Synthetic, step 1300](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA/tree/main/vsa-synthetic-step1300) | [VSA / Synthetic, step 1300](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300) | `run_fasth3_lora_preview_vsa_synthetic_step1300.sh` | |
| | [VSA / Synthetic, step 1900](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA/tree/main/vsa-synthetic-step1900) | [VSA / Synthetic, step 1900](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1900) | `run_fasth3_lora_preview_vsa_synthetic_step1900.sh` | |
| | [Dense / Data-Free](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-LoRA/tree/main/dense-datafree) | [Dense / Data-Free](https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-Dense-DataFree) | `run_fasth3_lora_preview_dense_datafree.sh` | |
|
|
| The VSA launchers select VSA-H3 automatically; the dense launcher disables it. |
| This preview is for text-to-audio-video generation and inherits the |
| [MiniMax H3 Community License](LICENSE). |
|
|
| ## Acknowledgements |
|
|
| We thank [Nuva Lab](https://nuvalab.ai/) for bringing production grounding to FastH3 through its experience with real-world creative video-agent workloads. Its production-aligned post-training insights help bridge open-source research to practical data-assisted distillation for commercial video workflows, with Omni Ref as the next focus. |
|
|
| We thank the [NVIDIA FastGen](https://github.com/NVlabs/FastGen) team for the [DMD2](https://arxiv.org/abs/2405.14867) framework and H3 reference experiment that helped us align the score clock, modality shifts, and backward simulation. |
|
|
| We also thank [MiniMax](https://huggingface.co/MiniMaxAI/MiniMax-H3) for releasing H3-Base, and the [vLLM project](https://vllm.ai/), [NVIDIA](https://www.nvidia.com/en-us/), and [MBZUAI](https://mbzuai.ac.ae/) for their continued sponsorship and support of [FastVideo](https://github.com/hao-ai-lab/FastVideo). |
|
|
