|
Download README.md from EliovpAI/MiniMax-H3-FP8-Paiton-RDNA4: direct link, hf CLI and curl.
- Browser
- Download file 5.18 kB
-
https://huggingface.co/EliovpAI/MiniMax-H3-FP8-Paiton-RDNA4/resolve/main/README.md
- Command line
-
hf download hf://EliovpAI/MiniMax-H3-FP8-Paiton-RDNA4/README.md
-
curl -L -o README.md https://huggingface.co/EliovpAI/MiniMax-H3-FP8-Paiton-RDNA4/resolve/main/README.md
5.18 kB
| license: other | |
| license_name: minimax-h3-community-license | |
| license_link: LICENSE | |
| base_model: MiniMaxAI/MiniMax-H3 | |
| library_name: paiton | |
| pipeline_tag: text-to-video | |
| tags: | |
| - video-generation | |
| - audio-video-generation | |
| - reference-to-video | |
| - amd | |
| - rocm | |
| - rdna4 | |
| - gfx1201 | |
| - fp8 | |
| - paiton | |
| # MiniMax H3 Ref2VA in FP8: video with sound on one R9700 | |
| Send a prompt and 1 to 9 reference pictures (or one reference video), and get back a 5.2-second 864×480 clip with | |
| stereo sound in **30 seconds**, on one Radeon AI PRO R9700 (32 GB, 300 W). | |
| These are the weights for the native MiniMax H3 Ref2VA runner in | |
| [paiton-vllm-plugin](https://github.com/Eliovp-BV/paiton-vllm-plugin/tree/main/models/MiniMax-H3) (v1.2). The runner is | |
| compiled HIP code for RDNA4, with no PyTorch, diffusers or ComfyUI inside. You call it over HTTP or from a ComfyUI node. | |
| ## Numbers | |
| The original BF16 release is 144 GB, and its transformer alone (66.3 GB) is twice the memory of the card. This build | |
| needs 50.9 GB on disk for one preset and peaks at 24.7 to 26.8 GiB of VRAM. | |
| Time per request on one R9700 with the 4-step preset (Turbo4), 124 frames at 24 fps, from the prepared inputs to the | |
| finished MP4. v1.1.0 is our earlier ComfyUI build with a W4A8 denoiser | |
| ([EliovpAI/MiniMax-H3-W4A8-Paiton-RDNA4](https://huggingface.co/EliovpAI/MiniMax-H3-W4A8-Paiton-RDNA4)) on the same | |
| card. | |
| | Request | v1.2 (these weights) | v1.1.0 | Faster by | | |
| |---|---:|---:|---:| | |
| | 864×480, 1 picture | **30.4 s** | 50.6 s | 1.7× | | |
| | 864×480, 3 pictures | **33.4 s** | 85.3 s | 2.6× | | |
| | 864×480, 9 pictures | **44.3 s** | 178.7 s | 4.0× | | |
| | 1344×768, 9 pictures | **114.1 s** | 258.4 s | 2.3× | | |
| | 864×480, 1 video with its soundtrack | **203.8 s** | not measured | | | |
| | 864×480, 1 picture, 8 steps (Turbo8) | **51.4 s** | 109.2 s | 2.1× | | |
| Peak VRAM for one picture at 864×480 went from 30.6 GiB (v1.1.0) to 24.8 GiB. The server loads the weights once when | |
| it starts (76 s in our test, including a SHA-256 check of every file) and keeps them in memory between requests. | |
| Quality, compared with the BF16 model in diffusers (MI355X, same prompts, references, seed and starting noise). CLIP-T | |
| scores how well the clip matches its prompt: | |
| | Request | CLIP-T, BF16 | CLIP-T, these weights | | |
| |---|---:|---:| | |
| | 864×480, 1 picture | 29.52 | 29.77 | | |
| | 864×480, 3 pictures | 34.68 | 34.35 | | |
| | 864×480, 9 pictures | 26.31 | 25.70 | | |
| | 1344×768, 9 pictures | 25.80 | 25.71 | | |
| | 864×480, 1 video | 29.87 | 29.86 | | |
| Over 15 clips (two seeds, both presets) the frames have a mean similarity of 0.932 (CLIP-I) and 0.871 (DINOv2) to the | |
| same frames of the BF16 clip, where 1.0 is the same image. With nine references the subjects are all there but can | |
| stand in other places than in the BF16 clip. Method and limits: | |
| [BENCHMARKS.md](https://github.com/Eliovp-BV/paiton-vllm-plugin/blob/main/models/MiniMax-H3/BENCHMARKS.md#quality-against-the-bf16-model). | |
| ## Run it | |
| ```sh | |
| git clone --depth 1 https://github.com/Eliovp-BV/paiton-vllm-plugin.git | |
| cd paiton-vllm-plugin | |
| PAITON_DETACH=1 ./models/MiniMax-H3/native/launch.sh http | |
| docker logs -f paiton-minimax-h3-v12 # wait for "READY http://..."; the first start downloads the weights | |
| curl -o panda.mp4 -F image=@panda.png \ | |
| -F prompt='The red panda from <Picture 1> walks through a sunlit forest. Audio: birds and a stream.' \ | |
| http://127.0.0.1:30000/v1/videos/generations | |
| ``` | |
| The runner downloads these files on first use and checks each one against the SHA-256 in its `checkpoints.lock.json`, | |
| so you don't need to download anything by hand. For ComfyUI, copy the `paiton_h3v_client` node from the plugin into | |
| `custom_nodes`; it sends the job to the server, so ComfyUI itself can run on the CPU. | |
| You need one R9700 with nothing else on it, Docker, 51 GB of disk for one preset, and about 41 GB of free host RAM for | |
| the fastest text-encoder path (with less, the runner reads the text encoder from disk instead). | |
| ## Files | |
| The files only work with this runner: they are packed for its kernels, not for diffusers or ComfyUI. | |
| | Folder | Content | Size | | |
| |---|---|---:| | |
| | `weights/turbo4/` | transformer, 4 steps (Turbo LoRA merged, FP8) | 20.3 GB | | |
| | `weights/turbo8/` | transformer, 8 steps (Turbo LoRA merged, FP8) | 20.5 GB | | |
| | `dec8.bin`, `enc.bin`, `audio.bin` | video VAE decoder (INT8), video VAE encoder (BF16), audio VAE (FP32) | 3.4 GB | | |
| | `model/tokenizer/` | tokenizer | 7 MB | | |
| The text encoder (Qwen3-VL-32B, layers 0 to 49 in FP8, and the vision tower, 27.2 GB) is the same as the one of | |
| MiniMax H3 Image, so it is not stored twice: the runner downloads it from the `te/` folder of | |
| [EliovpAI/MiniMax-H3-Image-FP8-Paiton-RDNA4](https://huggingface.co/EliovpAI/MiniMax-H3-Image-FP8-Paiton-RDNA4/tree/b33b88f27c8919ca05632ccde801087796d603dd/te), pinned by revision and SHA-256. | |
| The changes to each file are in `MODIFICATIONS.md`. | |
| ## License | |
| MiniMax H3 Community License (`LICENSE`, `NOTICE`). The grant excludes the European Union, the United Kingdom, the | |
| Republic of Korea and the United States, and commercial use above the agreement's revenue threshold needs MiniMax's | |
| written authorization. Read the agreement before you download or use these files. | |