Simplify model card: links, inference and selected videos

README.md

@@ -3,167 +3,110 @@ license: other
 license_name: tencent-hunyuan-community
 license_link: https://huggingface.co/tencent/HunyuanVideo/blob/6204ad6aea1a77ff5aba337c88278bb9500eb37/LICENSE
 base_model: hunyuanvideo-community/HunyuanVideo
+pipeline_tag: text-to-video
 tags:
-- text-to-video
 - hunyuanvideo
 - diffusionart
+- arxiv:2610.12572
 ---
 
-# DiffusionART HunyuanVideo 7:2:1 — selected step 30
+# DiffusionART · HunyuanVideo
 
+[Paper](https://arxiv.org/abs/2610.12572) ·
+[Code](https://github.com/vanzll/DiffusionART) ·
 [Project Page](https://vanzll.github.io/diffusionart-project/) ·
-[Code](https://github.com/vanzll/DiffusionART)
+[Demo](https://vanzll.github.io/diffusionart-project/demo.html)
 
-This private release is a **full HunyuanVideo transformer**, not a complete
-pipeline or LoRA adapter. It is the selected optimizer-step-30 checkpoint from
-[run `tk1w4go1`](https://wandb.ai/vanzl3386-chinese-university-of-hong-kong-shenzhen/flow-grpo-algo/runs/tk1w4go1).
-The original HunyuanVideo-13B transformer was trained with temporal alignment,
-video quality, and motion quality rewards weighted 0.7 / 0.2 / 0.1, plus a
-frozen pretrained-velocity MSE regularizer with beta 0.1. The transformer is
-stored as F32 safetensors shards. The public base model supplies the VAE, text
-encoders, tokenizers, and scheduler. `release.json` records the source and shard
-hashes; this inference update does not alter those files.
+HunyuanVideo-13B fine-tuned with **DiffusionART**, using temporal alignment,
+visual quality and motion quality rewards at **7:2:1**, with a pretrained-reference
+MSE regularizer (β = 0.1).
 
-## Selected Videos
+This repository contains the **full transformer weights**, not a LoRA or a
+complete pipeline. The inference entry below downloads the base model's VAE
+and text encoders automatically.
 
-Selected generations from the released model, alongside archived pretrained
-references for the same prompts. The two inference paths are not an
-identical-noise controlled comparison.
-More results: [Project Page](https://vanzll.github.io/diffusionart-project/).
+## Inference
 
-**Ocean waves roll onto a black sand beach beneath towering green cliffs.**
+Use the included [infer.py](https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/blob/main/infer.py).
+Expand the setup once, then run the example from the same directory.
 
-<table><tr><th>Pretrained</th><th>DiffusionART</th></tr><tr>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00100-pretrained.mp4"></video></td>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00100-diffusionart.mp4"></video></td>
-</tr></table>
+<details>
+<summary>Environment setup</summary>
 
-**The camera follows a winding river through a valley of autumn trees.**
-
-<table><tr><th>Pretrained</th><th>DiffusionART</th></tr><tr>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00106-pretrained.mp4"></video></td>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00106-diffusionart.mp4"></video></td>
-</tr></table>
-
-**A white horse gallops through shallow water along a sunlit beach.**
-
-<table><tr><th>Pretrained</th><th>DiffusionART</th></tr><tr>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00113-pretrained.mp4"></video></td>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00113-diffusionart.mp4"></video></td>
-</tr></table>
-
-**A chef pours hot tea into a glass cup as steam rises.**
-
-<table><tr><th>Pretrained</th><th>DiffusionART</th></tr><tr>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00129-pretrained.mp4"></video></td>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00129-diffusionart.mp4"></video></td>
-</tr></table>
-
-**The camera approaches a glass pavilion surrounded by autumn trees.**
-
-<table><tr><th>Pretrained</th><th>DiffusionART</th></tr><tr>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00139-pretrained.mp4"></video></td>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00139-diffusionart.mp4"></video></td>
-</tr></table>
-
-**Sunlight sparkles through a rotating crystal prism on a white surface.**
-
-<table><tr><th>Pretrained</th><th>DiffusionART</th></tr><tr>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00148-pretrained.mp4"></video></td>
-<td><video controls muted loop playsinline width="280" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00148-diffusionart.mp4"></video></td>
-</tr></table>
-
-## Requirements
-
-- Access to this private Hugging Face repository through `HF_TOKEN` or an
-  interactive Hugging Face login. The commands keep the login and model cache
-  in the chosen work directory. Do not put a token in a script or shell history.
-- Linux x86-64, Python **3.12**, a CUDA 12.1 compatible NVIDIA driver and CUDA
-  toolkit (`nvcc`) for building the pinned `flash-attn` version, plus `curl`,
-  `tar`, and `sha256sum`.
-- One H800 with **at least 100 GiB available GPU memory** for the verified
-  default path. Independent inference used one 140 GiB H800 and peaked at
-  92.39 GiB allocated. Allow at least **100 GiB writable disk space** for the
-  private transformer, pinned public base assets, and installation caches.
-  `--cpu-offload` is available for constrained GPUs but was not part of the
-  independent acceptance run.
-
-## Install and generate one MP4
-
-The commands use only this release and the fixed **public**
-[`XueZeyue/DanceGRPO`](https://github.com/XueZeyue/DanceGRPO/tree/15cc71d53cc2e6e18a68ee607d5fb6ba9a99e344)
-runtime. No private training repository, saved prompt embeddings, notebook, or
-MPI launcher is needed. Choose a work directory with the required free space.
+Linux, Python 3.12, a CUDA 12.1-compatible driver and toolkit (`nvcc`), and
+approximately 100 GiB of available GPU memory and disk space for the verified
+default path. The installer uses a pinned public DanceGRPO inference runtime.
 
 ```bash
 export RUN_DIR="${RUN_DIR:-$PWD/diffusionart-t2v-demo}"
 export HF_HOME="$RUN_DIR/cache/hf"
-export HF_HUB_CACHE="$HF_HOME/hub"
 export PIP_CACHE_DIR="$RUN_DIR/pip-cache"
 export TMPDIR="$RUN_DIR/tmp"
 export XDG_CACHE_HOME="$RUN_DIR/xdg-cache"
-mkdir -p "$RUN_DIR" "$HF_HOME" "$HF_HUB_CACHE" "$PIP_CACHE_DIR" "$TMPDIR" "$XDG_CACHE_HOME"
+mkdir -p "$RUN_DIR" "$HF_HOME" "$PIP_CACHE_DIR" "$TMPDIR" "$XDG_CACHE_HOME"
 cd "$RUN_DIR"
 python3.12 -m venv bootstrap
 bootstrap/bin/python -m pip install 'huggingface-hub==0.26.1'
-if [ -z "${HF_TOKEN:-}" ] && [ ! -f "$HF_HOME/token" ]; then
-  bootstrap/bin/huggingface-cli login
-fi
-
-# If MODEL_REVISION is already set to a known 40-character commit SHA, keep it.
-# Otherwise resolve main exactly once, then pin every download to that SHA.
-if [ -z "${MODEL_REVISION:-}" ]; then
-  MODEL_REVISION="$(bootstrap/bin/python -c 'from huggingface_hub import HfApi; print(HfApi().model_info("vanzl/DiffusionART-HunyuanVideo-7-2-1", token=True).sha)')"
-fi
-export MODEL_REVISION
-printf 'Pinned model revision: %s\n' "$MODEL_REVISION"
+bootstrap/bin/huggingface-cli login
 
 bootstrap/bin/python - <<'PY'
-import os
 from huggingface_hub import snapshot_download
 snapshot_download(
-    repo_id="vanzl/DiffusionART-HunyuanVideo-7-2-1",
-    revision=os.environ["MODEL_REVISION"],
-    token=True,
-    allow_patterns=["README.md", "infer.py", "install.sh", "requirements-inference.txt"],
+    "vanzl/DiffusionART-HunyuanVideo-7-2-1",
+    revision="b9f83d8fc677be078cf20e75e2149a5f69059220",
+    allow_patterns=["infer.py", "install.sh", "requirements-inference.txt"],
     local_dir="release",
 )
 PY
+bash release/install.sh "$RUN_DIR/venv" "$RUN_DIR/runtime"
+```
+
+The fixed revision preserves the released weights and inference entry.
+Hugging Face login is used by that entry even though this model is public.
 
-bash release/install.sh "$RUN_DIR/venv" "$RUN_DIR/DanceGRPO-public"
-"$RUN_DIR/venv/bin/python" release/infer.py \
-  --revision "$MODEL_REVISION" \
-  --prompt 'A white horse gallops through shallow water along a sunlit beach.' \
-  --seed 42 \
-  --output "$RUN_DIR/horse_seed42.mp4" \
-  --cache-dir "$RUN_DIR/cache"
+</details>
+
+```bash
+venv/bin/python - <<'PY'
+import runpy
+import sys
+import torch
+
+# Use the attention backend from the successful inference checks.
+torch.backends.cuda.enable_cudnn_sdp(False)
+sys.argv = [
+    "infer.py",
+    "--revision", "b9f83d8fc677be078cf20e75e2149a5f69059220",
+    "--prompt", "A white horse gallops through shallow water along a sunlit beach.",
+    "--output", "diffusionart.mp4",
+    "--cache-dir", "cache",
+]
+runpy.run_path("release/infer.py", run_name="__main__")
+PY
 ```
 
-`install.sh` installs PyTorch **2.5.0+cu121**, torchvision **0.20.0+cu121**,
-the versions in `requirements-inference.txt`, `flash-attn==2.7.0.post2`,
-`liger-kernel==0.4.1`, and the public DanceGRPO commit
-`15cc71d53cc2e6e18a68ee607d5fb6ba9a99e344`. It verifies the public
-source archive SHA256 before installation. `infer.py` downloads this model at
-`--revision` and the base model at fixed revision
-`e8c2aaa66fe3742a32c11a6766aecbf07c56e773` into `--cache-dir`.
-Pass a different `--prompt`, `--seed`, or `--output` to make another video.
-`infer.py` refuses to overwrite an existing MP4 and validates the generated
-file by decoding every frame.
-
-The verified protocol is **640×640, 53 frames, 8 FPS, 30 sampling steps,
-FlowMatchEulerDiscreteScheduler shift 7, eta 0, and no classic CFG**. The
-public pipeline takes `guidance_scale=6.018` as the user-facing value and
-multiplies it by 1000 internally, yielding embedded guidance **6018**. Do not
-pass 6018 as `guidance_scale`. The loader keeps the transformer in F32, runs
-with BF16 autocast, loads both tokenizers from the pinned public base model,
-and enables VAE tiling. The script prints the MP4 SHA256, decoded frame count,
-resolution, FPS, and pixel statistics after a successful run.
-
-## Terms and provenance
-
-The package has no optimizer, RNG state, or training code. Tencent's original
-HunyuanVideo weights and derivatives are subject to the **Tencent Hunyuan
-Community License Agreement**. Read the included [LICENSE.txt](LICENSE.txt)
-and [Notice](Notice) before use. The model card and inference helpers do not
-change their terms. The original run was planned for 100 updates but stopped
-at step 40; this release selects step 30 and does not claim 100 updates.
+Defaults: **640×640 · 53 frames · 8 FPS · 30 sampling steps**.
+The script uses the verified shift, guidance, precision and VAE settings.
+Change `--prompt` and `--output` for another video; an existing output is never overwritten.
+
+## Examples
+
+Selected outputs from this model. See the [Demo](https://vanzll.github.io/diffusionart-project/demo.html)
+for the full gallery and pretrained comparisons.
+
+<table>
+<tr>
+<td><video controls muted loop playsinline preload="none" width="280" poster="https://vanzll.github.io/diffusionart-project/assets/hf-video-00113-after.jpg" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00113-diffusionart.mp4"></video><p>A white horse gallops through shallow water along a sunlit beach.</p></td>
+<td><video controls muted loop playsinline preload="none" width="280" poster="https://vanzll.github.io/diffusionart-project/assets/hf-video-00139-after.jpg" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00139-diffusionart.mp4"></video><p>The camera approaches a glass pavilion surrounded by autumn trees.</p></td>
+</tr>
+<tr>
+<td><video controls muted loop playsinline preload="none" width="280" poster="https://vanzll.github.io/diffusionart-project/assets/hf-video-00106-after.jpg" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00106-diffusionart.mp4"></video><p>The camera follows a winding river through a valley of autumn trees.</p></td>
+<td><video controls muted loop playsinline preload="none" width="280" poster="https://vanzll.github.io/diffusionart-project/assets/hf-video-00148-after.jpg" src="https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/resolve/main/examples/00148-diffusionart.mp4"></video><p>Sunlight sparkles through a rotating crystal prism on a white surface.</p></td>
+</tr>
+</table>
+
+## License
+
+Subject to the [Tencent Hunyuan Community License](https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/blob/main/LICENSE.txt)
+and [Notice](https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/blob/main/Notice).
+Training provenance and weight hashes are recorded in [release.json](https://huggingface.co/vanzl/DiffusionART-HunyuanVideo-7-2-1/blob/main/release.json).