add readme

Co-authored-by: RandNMR73 <notomatthew31@gmail.com>
Co-authored-by: JerryZhou54 <zhouw.jerry2017@outlook.com>
Co-authored-by: Davids048 <jundasu@ucsd.edu>
Co-authored-by: Peiyuan Zhang <a1286225768@slurm-h200-204-239.slurm-compute.tenant-slurm.svc.cluster.local>

.github/workflows/docs.yml

@@ -45,6 +45,12 @@ jobs:
       - name: Setup Pages
         uses: actions/configure-pages@v4
 
+      - name: Generate docs examples
+        run: python docs/generate_examples.py
+
+      - name: Check docs links
+        run: python scripts/check_docs_links.py
+
       - name: Build documentation
         run: mkdocs build
 
@@ -63,4 +69,4 @@ jobs:
     steps:
       - name: Deploy to GitHub Pages
         id: deployment
-        uses: actions/deploy-pages@v4
+        uses: actions/deploy-pages@v4

.github/workflows/fastvideo-kernel-publish.yml

@@ -64,7 +64,10 @@ jobs:
             # - torch-version: '2.7.1'
             #   cuda-version: '12.8.0'
             #   torch-cuda-short: 'cu128'
-            - torch-version: '2.9.1'
+            # - torch-version: '2.9.1'
+            #   cuda-version: '12.8.0'
+            #   torch-cuda-short: 'cu128'
+            - torch-version: '2.10.0'
               cuda-version: '12.8.0'
               torch-cuda-short: 'cu128'
 

.gitignore

@@ -18,6 +18,7 @@ venv/
 .venv/
 runs/
 samples/
+Miniconda3-latest-Linux-x86_64.sh
 *validation/
 data/
 outputs/
@@ -33,6 +34,11 @@ env
 *.log
 weights/
 
+# SSIM test outputs
+fastvideo/tests/ssim/generated_videos/
+**/.cache/**
+
+
 # Distribution / packaging
 build/
 dist/
@@ -69,6 +75,11 @@ docs/distillation/examples/
 !docs/assets/images/**/*.png
 !comfyui/assets/**/*.png
 !comfyui/assets/**/*.gif
+!assets/images/**/*.png
+!assets/images/**/*.jpg
+!assets/images/**/*.jpeg
+!assets/images/**/*.gif
+!assets/videos/**/*.mp4
 
 dmd_t2v_output/
 preprocess_output_text/

.pre-commit-config.yaml

@@ -10,7 +10,7 @@ exclude: |
     demo/.*|
     predict\.py|
     scripts/.*|
-    prompts/.*|
+    assets/prompts/.*|
     fastvideo/data_preprocess/.*|
     fastvideo/dataset/.*|
     fastvideo/models/.*|

AGENTS.md

@@ -0,0 +1,42 @@
+# Repository Guidelines
+
+## Project Structure & Module Organization
+- Core Python package: `fastvideo/` (models, pipelines, training, distributed runtime, CLI entrypoints).
+- CUDA/custom kernels: `fastvideo-kernel/` (separate build/test flow).
+- Tests:
+  - `fastvideo/tests/` for package-level tests (dataset, encoders, inference, training, SSIM, workflow).
+  - `tests/local_tests/` for additional local/component checks.
+- Docs and guides: `docs/` (MkDocs source), with contributor docs in `docs/contributing/`.
+- Runnable examples and scripts: `examples/` and `scripts/`.
+- Static assets: `assets/` (including `assets/images/`, `assets/videos/`, and `assets/prompts/`) and `comfyui/assets/`.
+
+## Build, Test, and Development Commands
+- `uv pip install -e .[dev]`: editable install with lint/test extras.
+- `pre-commit install --hook-type pre-commit --hook-type commit-msg`: enable local hooks.
+- `pre-commit run --all-files`: run formatter/lint/type/spelling checks.
+- `pytest tests/`: run top-level test suite.
+- `pytest fastvideo/tests/ -v`: run package tests.
+- `pytest fastvideo/tests/ssim/ -vs`: run SSIM regression tests (GPU-heavy).
+- `cd fastvideo-kernel && ./build.sh`: build kernel extensions.
+
+## Coding Style & Naming Conventions
+- Python 3.10+; 4-space indentation; keep code and imports readable and explicit.
+- Style tools are configured in `pyproject.toml` and `.pre-commit-config.yaml`:
+  - `yapf` (format), `ruff` (lint, auto-fix), `mypy` (typing), `codespell`.
+- Target line length is 80.
+- Naming: `snake_case` for functions/files, `PascalCase` for classes, `UPPER_SNAKE_CASE` for constants.
+
+## Testing Guidelines
+- Use `pytest` and place tests near relevant domains (e.g., `fastvideo/tests/encoders/`).
+- Prefer descriptive names like `test_<feature>_<expected_behavior>.py`.
+- For new pipelines/backends, include at least one regression-oriented test; add SSIM coverage when output quality must be preserved.
+- Document GPU assumptions in tests that require specific hardware.
+
+## Commit & Pull Request Guidelines
+- Follow existing commit style: short subject with optional tag prefix, e.g. `[bugfix]: ...`, `[feat]: ...`, `[misc]: ...`, and include PR reference like `(#1234)` when applicable.
+- Keep commits focused by concern (feature, refactor, fix).
+- PRs should include:
+  - clear problem/solution summary,
+  - test evidence (`pytest`/SSIM outputs or rationale if skipped),
+  - linked issue/PR context,
+  - screenshots or sample outputs for UI/demo/docs changes.

CLAUDE.md

@@ -0,0 +1 @@
+@AGENTS.md

README.md

@@ -1,147 +1,101 @@
+# Sliding Tile Attention (STA) Branch
 
-| **[Documentation](https://hao-ai-lab.github.io/FastVideo)** | **[Quick Start](https://hao-ai-lab.github.io/FastVideo/inference/inference_quick_start/)** | **[Weekly Dev Meeting](https://github.com/hao-ai-lab/FastVideo/discussions/982)** | 🟣💬 **[Slack](https://join.slack.com/t/fastvideo/shared_invite/zt-3f4lao1uq-u~Ipx6Lt4J27AlD2y~IdLQ)** |
+This branch is a stash/testing branch for people who want to try
+Sliding Tile Attention (STA). The top-level README is intentionally
+STA-only.
 
-**FastVideo is a unified post-training and inference framework for accelerated video generation.**
+## What is STA
 
-## NEWS
+Sliding Tile Attention is an optimized attention backend for
+window-based video generation.
 
-- `2025/11/19`: Release [CausalWan2.2 I2V A14B Preview](https://huggingface.co/FastVideo/CausalWan2.2-I2V-A14B-Preview-Diffusers) models, [Blog](https://hao-ai-lab.github.io/blogs/fastvideo_causalwan_preview/) and [Inference Code!](https://github.com/hao-ai-lab/FastVideo/blob/main/examples/inference/basic/basic_self_forcing_causal_wan2_2_i2v.py)
-- `2025/08/04`: Release [FastWan](https://hao-ai-lab.github.io/FastVideo/distillation/dmd) models and [Sparse-Distillation](https://hao-ai-lab.github.io/blogs/fastvideo_post_training/).
+- Blog: https://hao-ai-lab.github.io/blogs/sta/
+- Paper: https://arxiv.org/abs/2502.04507
+- In-repo STA docs: `docs/attention/sta/index.md`
 
-### More News
+## Setup
 
-- `2025/06/14`: Release finetuning and inference code for [VSA](https://arxiv.org/pdf/2505.13389)
-- `2025/04/24`: [FastVideo V1](https://hao-ai-lab.github.io/blogs/fastvideo/) is released!
-- `2025/02/18`: Release the inference code for [Sliding Tile Attention](https://hao-ai-lab.github.io/blogs/sta/).
-
-## Key Features
-
-FastVideo has the following features:
-
-- End-to-end post-training support for bidirectional and autoregressive models:
-  - Support full finetuning and LoRA finetuning for state-of-the-art open video DiTs
-  - Data preprocessing pipeline for video, image, and text data
-  - Distribution Matching Distillation (DMD2) stepwise distillation.
-  - Sparse attention with [Video Sparse Attention](https://arxiv.org/pdf/2505.13389)
-  - [Sparse distillation](https://hao-ai-lab.github.io/blogs/fastvideo_post_training/) to achieve >50x denoising speedup
-  - Scalable training with FSDP2, sequence parallelism, and selective activation checkpointing.
-  - Causal distillation through Self-Forcing
-  - See this [page](https://hao-ai-lab.github.io/FastVideo/training/overview/) for full list of supported models and recipes.
-- State-of-the-art performance optimizations for inference
-  - Sequence Parallelism for distributed inference
-  - Multiple state-of-the-art attention backends
-  - User-friendly CLI and Python API
-  - See this [page](https://hao-ai-lab.github.io/FastVideo/inference/optimizations/) for full list of supported optimizations.
-- Diverse hardware and OS support
-  - Support H100, A100, 4090
-  - Support Linux, Windows, MacOS
-  - See this [page](https://hao-ai-lab.github.io/FastVideo/inference/hardware_support/) for full list of supported hardware and OS.
-
-## Getting Started
-
-We recommend using an environment manager such as `Conda` to create a clean environment:
+Install FastVideo from source:
 
 ```bash
-# Create and activate a new conda environment
-conda create -n fastvideo python=3.12
-conda activate fastvideo
-
-# Install FastVideo
-pip install fastvideo
+uv pip install -e .
 ```
 
-Please see our [docs](https://hao-ai-lab.github.io/FastVideo/getting_started/installation/) for more detailed installation instructions.
-
-## Sparse Distillation
-
-For our sparse distillation techniques, please see our [distillation docs](https://hao-ai-lab.github.io/FastVideo/distillation/dmd/) and check out our [blog](https://hao-ai-lab.github.io/blogs/fastvideo_post_training/).
-
-See below for recipes and datasets:
-
-| Model                                                                                 | Sparse Distillation                                                                                             | Dataset                                                                                                  |
-| ------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------- |
-| [FastWan2.1-T2V-1.3B](https://huggingface.co/FastVideo/FastWan2.1-T2V-1.3B-Diffusers) | [Recipe](https://github.com/hao-ai-lab/FastVideo/tree/main/examples/distill/Wan2.1-T2V/Wan-Syn-Data-480P)       | [FastVideo Synthetic Wan2.1 480P](https://huggingface.co/datasets/FastVideo/Wan-Syn_77x448x832_600k)     |
-| [FastWan2.2-TI2V-5B](https://huggingface.co/FastVideo/FastWan2.2-TI2V-5B-Diffusers)   | [Recipe](https://github.com/hao-ai-lab/FastVideo/tree/main/examples/distill/Wan2.2-TI2V-5B-Diffusers/Data-free) | [FastVideo Synthetic Wan2.2 720P](https://huggingface.co/datasets/FastVideo/Wan2.2-Syn-121x704x1280_32k) |
-
-## Inference
-
-### Generating Your First Video
-
-Here's a minimal example to generate a video using the default settings. Make sure VSA kernels are [installed](https://hao-ai-lab.github.io/FastVideo/video_sparse_attention/installation/). Create a file called `example.py` with the following code:
-
-```python
-import os
-from fastvideo import VideoGenerator
-
-def main():
-    os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "VIDEO_SPARSE_ATTN"
-
-    # Create a video generator with a pre-trained model
-    generator = VideoGenerator.from_pretrained(
-        "FastVideo/FastWan2.1-T2V-1.3B-Diffusers",
-        num_gpus=1,  # Adjust based on your hardware
-    )
-
-    # Define a prompt for your video
-    prompt = "A curious raccoon peers through a vibrant field of yellow sunflowers, its eyes wide with interest."
-
-    # Generate the video
-    video = generator.generate_video(
-        prompt,
-        return_frames=True,  # Also return frames from this call (defaults to False)
-        output_path="my_videos/",  # Controls where videos are saved
-        save_video=True
-    )
-
-if __name__ == '__main__':
-    main()
-```
-
-Run the script with:
+Build the STA kernel package:
 
 ```bash
-python example.py
+cd fastvideo-kernel
+./build.sh
+cd ..
 ```
 
-For a more detailed guide, please see our [inference quick start](https://hao-ai-lab.github.io/FastVideo/inference/inference_quick_start/).
+## Run STA Inference
 
-## More Guides
+STA backend:
 
-- [Design Overview](https://hao-ai-lab.github.io/FastVideo/design/overview/)
-- [Distillation Guide](https://hao-ai-lab.github.io/FastVideo/distillation/dmd/)
-- [Contribution Guide](https://hao-ai-lab.github.io/FastVideo/contributing/overview/)
+```bash
+export FASTVIDEO_ATTENTION_BACKEND=SLIDING_TILE_ATTN
+```
 
-## Awesome work using FastVideo or our research projects
+Ready-to-run examples:
 
-- [SGLang](https://github.com/sgl-project/sglang/tree/main/python/sglang/multimodal_gen): SGLang's diffusion inference functionality is based on a fork of FastVideo on Sept. 24, 2025.
-- [DanceGRPO](https://github.com/XueZeyue/DanceGRPO): A unified framework to adapt Group Relative Policy Optimization (GRPO) to visual generation paradigms. Code based on FastVideo.
-- [SRPO](https://github.com/Tencent-Hunyuan/SRPO): A method to directly align the full diffusion trajectory with fine-grained human preference. Code based on FastVideo.
-- [DCM](https://github.com/Vchitect/DCM): Dual-expert consistency model for efficient and high-quality video generation. Code based on FastVideo.
-- [Hunyuan Video 1.5](https://github.com/Tencent-Hunyuan/HunyuanVideo-1.5): A leading lightweight video generation model, where they proposed SSTA based on Sliding Tile Attention.
-- [Kandinsky-5.0](https://github.com/kandinskylab/kandinsky-5): A family of diffusion models for video & image generation, where their NABLA attention includes a Sliding Tile Attention branch.
-- [LongCat Video](https://github.com/meituan-longcat/LongCat-Video): A foundational video generation model with 13.6B parameters with block-sparse attention similar to Video Sparse Attention.
+- HunyuanVideo: `scripts/inference/v1_inference_hunyuan_STA.sh`
+- Wan2.1-T2V-14B: `scripts/inference/v1_inference_wan_STA.sh`
 
-## 🤝 Contributing
+Run:
 
-We welcome all contributions. Please check out our guide [here](https://hao-ai-lab.github.io/FastVideo/contributing/overview/).
-See details in [development roadmap](https://github.com/hao-ai-lab/FastVideo/issues/899).
+```bash
+bash scripts/inference/v1_inference_hunyuan_STA.sh
+# or
+bash scripts/inference/v1_inference_wan_STA.sh
+```
 
-## Acknowledgement
+Both scripts already set STA-related env vars:
 
-We learned the design and reused code from the following projects: [Wan-Video](https://github.com/Wan-Video), [ThunderKittens](https://github.com/HazyResearch/ThunderKittens), [DMD2](https://github.com/tianweiy/DMD2), [diffusers](https://github.com/huggingface/diffusers), [xDiT](https://github.com/xdit-project/xDiT), [vLLM](https://github.com/vllm-project/vllm), [SGLang](https://github.com/sgl-project/sglang). We thank [MBZUAI](https://ifm.mbzuai.ac.ae/), [Anyscale](https://www.anyscale.com/), and [GMI Cloud](https://www.gmicloud.ai/) for their support throughout this project.
+- `FASTVIDEO_ATTENTION_BACKEND=SLIDING_TILE_ATTN`
+- `FASTVIDEO_ATTENTION_CONFIG` to an STA mask strategy JSON
+
+## STA Mask Strategy Files
+
+- HunyuanVideo config: `assets/mask_strategy_hunyuan.json`
+- Wan config: `assets/mask_strategy_wan.json`
+
+## STA Mask Search (Wan2.1-T2V-14B)
+
+Run mask search + tuning from repo root:
+
+```bash
+bash examples/inference/sta_mask_search/inference_wan_sta.sh
+```
+
+What this script does:
+
+- Runs `STA_searching` first.
+- Runs `STA_tuning` next (`skip_time_steps=12` by default).
+- Uses prompt shards from `assets/prompt_0.txt` to `assets/prompt_7.txt`.
+
+Important notes:
+
+- Default script is set to 8 GPUs (`num_gpu=8`). If needed, edit
+  `examples/inference/sta_mask_search/inference_wan_sta.sh`.
+- STA searching/tuning currently supports `69x768x1280` (Wan setting).
+
+Generated files:
+
+- Search results: `output/mask_search_result_pos_1280x768/`
+- Tuned strategy: `output/mask_search_strategy_1280x768/mask_strategy_s12.json`
+
+Use the tuned mask for STA inference:
+
+```bash
+export FASTVIDEO_ATTENTION_BACKEND=SLIDING_TILE_ATTN
+export FASTVIDEO_ATTENTION_CONFIG=output/mask_search_strategy_1280x768/mask_strategy_s12.json
+python examples/inference/sta_mask_search/wan_example.py --STA_mode STA_inference --num_gpus 1
+```
 
 ## Citation
 
-If you find FastVideo useful, please consider citing our research work:
-
 ```bibtex
-@article{zhang2025vsa,
-  title={Vsa: Faster video diffusion with trainable sparse attention},
-  author={Zhang, Peiyuan and Chen, Yongqi and Huang, Haofeng and Lin, Will and Liu, Zhengzhong and Stoica, Ion and Xing, Eric and Zhang, Hao},
-  journal={arXiv preprint arXiv:2505.13389},
-  year={2025}
-}
-
 @article{zhang2025fast,
   title={Fast video generation with sliding tile attention},
   author={Zhang, Peiyuan and Chen, Yongqi and Su, Runlong and Ding, Hangliang and Stoica, Ion and Liu, Zhengzhong and Zhang, Hao},

assets/prompt_0.txt

@@ -0,0 +1 @@
+Will Smith casually eats noodles, his relaxed demeanor contrasting with the energetic background of a bustling street food market. The scene captures a mix of humor and authenticity. Mid-shot framing, vibrant lighting.

assets/prompt_1.txt

@@ -0,0 +1 @@
+A lone hiker stands atop a towering cliff, silhouetted against the vast horizon. The rugged landscape stretches endlessly beneath, its earthy tones blending into the soft blues of the sky. The scene captures the spirit of exploration and human resilience. High angle, dynamic framing, with soft natural lighting emphasizing the grandeur of nature.

assets/prompt_2.txt

@@ -0,0 +1 @@
+A hand with delicate fingers picks up a bright yellow lemon from a wooden bowl filled with lemons and sprigs of mint against a peach-colored background. The hand gently tosses the lemon up and catches it, showcasing its smooth texture. A beige string bag sits beside the bowl, adding a rustic touch to the scene. Additional lemons, one halved, are scattered around the base of the bowl. The even lighting enhances the vibrant colors and creates a fresh, inviting atmosphere.

assets/prompt_3.txt

@@ -0,0 +1 @@
+A curious raccoon peers through a vibrant field of yellow sunflowers, its eyes wide with interest. The playful yet serene atmosphere is complemented by soft natural light filtering through the petals. Mid-shot, warm and cheerful tones.

assets/prompt_4.txt

@@ -0,0 +1 @@
+A superintelligent humanoid robot waking up. The robot has a sleek metallic body with futuristic design features. Its glowing red eyes are the focal point, emanating a sharp, intense light as it powers on. The scene is set in a dimly lit, high-tech laboratory filled with glowing control panels, robotic arms, and holographic screens. The setting emphasizes advanced technology and an atmosphere of mystery. The ambiance is eerie and dramatic, highlighting the moment of awakening and the robots immense intelligence. Photorealistic style with a cinematic, dark sci-fi aesthetic. Aspect ratio: 16:9 --v 6.1

assets/prompt_5.txt

@@ -0,0 +1 @@
+fox in the forest close-up quickly turned its head to the left

assets/prompt_6.txt

@@ -0,0 +1 @@
+Man walking his dog in the woods on a hot sunny day

assets/prompt_7.txt

@@ -0,0 +1 @@
+A majestic lion strides across the golden savanna, its powerful frame glistening under the warm afternoon sun. The tall grass ripples gently in the breeze, enhancing the lion's commanding presence. The tone is vibrant, embodying the raw energy of the wild. Low angle, steady tracking shot, cinematic.

assets/prompts/mixkit_i2v.jsonl

@@ -1,110 +1,110 @@
 [
     {
         "prompt": "Young man skating with a skateboard on the ramps with graffiti of a park with trees, on a sunny day.", 
-        "image_path": "images/mixkit-boy-skating-with-a-skateboard-in-a-park-with-ramps-34389.png"
+        "image_path": "assets/images/mixkit-boy-skating-with-a-skateboard-in-a-park-with-ramps-34389.png"
     },
     {
         "prompt": "In the midst of the joyous New Year's Eve celebration, the cheerful group of friends, their spirits lifted by the festivities, decides to immortalize the moment with a vibrant snapshot", 
-        "image_path": "images/mixkit-a-cheerful-group-of-friends-celebrate-new-years-eve-and-51525.png"
+        "image_path": "assets/images/mixkit-a-cheerful-group-of-friends-celebrate-new-years-eve-and-51525.png"
     },
     {
         "prompt": "A man and a woman playing in a field with grass, during a bright afternoon, while cars pass by in the distance.", 
-        "image_path": "images/mixkit-a-cute-couple-playing-on-the-grass-4688.png"
+        "image_path": "assets/images/mixkit-a-cute-couple-playing-on-the-grass-4688.png"
     },
     {
         "prompt": "Aerial view of a rocky mountain in the forest at a sunny day drone flight footage", 
-        "image_path": "images/mixkit-aerial-view-of-a-rocky-mountain-in-the-forest-50589.png"
+        "image_path": "assets/images/mixkit-aerial-view-of-a-rocky-mountain-in-the-forest-50589.png"
     },
     {
         "prompt": "A little girl wearing a pink security helmet and denim overall discovers the art of cycling amidst the serene park, as the camera captures her graceful progress.", 
-        "image_path": "images/mixkit-a-little-girl-cruises-through-the-forest-path-on-her-50088.png"
+        "image_path": "assets/images/mixkit-a-little-girl-cruises-through-the-forest-path-on-her-50088.png"
     },
     {
         "prompt": "Aerial shot of a beach shore with sea waves. Big rocks on the sand at an alone beach.", 
-        "image_path": "images/mixkit-aerial-shot-of-a-beach-with-sea-waves-1087.png"
+        "image_path": "assets/images/mixkit-aerial-shot-of-a-beach-with-sea-waves-1087.png"
     },
     {
         "prompt": "Young woman cleaning her house decorated with plants and decorations, while dancing happily to music in her headphones.", 
-        "image_path": "images/mixkit-woman-cleaning-her-house-dancing-happy-43379.png"
+        "image_path": "assets/images/mixkit-woman-cleaning-her-house-dancing-happy-43379.png"
     },
     {
         "prompt": "Aerial tour in a meadow surrounded by hills on the horizon, while some birds fly low over a lake.", 
-        "image_path": "images/mixkit-birds-flying-low-over-a-lake-in-a-meadow-41417.png"
+        "image_path": "assets/images/mixkit-birds-flying-low-over-a-lake-in-a-meadow-41417.png"
     },
     {
         "prompt": "In the video, a lone rider guides a majestic horse across an expansive, open field as the sun sets in the background. The rider, dressed in a classic blue shirt and wide-brimmed hat, sits confidently in the saddle, silhouetted against the warm glow of the evening sky. The horse moves gracefully, its mane and tail flowing with each step, creating a sense of harmony between horse and rider. Surrounding the pair, towering trees form a natural border, their leaves gently rustling in the breeze. The shadows lengthen on the ground, accentuating the serene and timeless feel of the scene. The distant hills and wooden fences frame the horizon, adding depth to the tranquil landscape. A few horses graze peacefully in the background, blending into the pastoral setting. The overall ambiance evokes a sense of calmness and quietude, capturing a perfect moment in the golden light of dusk.", 
-        "image_path": "images/mixkit-a-rancher-riding-a-horse-at-sunset-1143.png"
+        "image_path": "assets/images/mixkit-a-rancher-riding-a-horse-at-sunset-1143.png"
     },
     {
         "prompt": "In the video, a martial artist dressed in a traditional white uniform with a black belt demonstrates a series of precise movements against a stark black background. The individual gracefully transitions between stances, embodying a sense of focused discipline and control. Each motion is executed with a deliberate pace, showcasing the fluidity of martial arts techniques. The soft lighting creates subtle highlights on the uniform, adding depth to the figure as it moves. The practitioner begins with an open-hand pose, feet firmly grounded, gradually shifting to a powerful forward punch. The fluidity of the sequence displays a mastery of balance and poise. Every trajectory of the limbs is precise and deliberate, capturing the elegance and strength of martial arts. The serene, isolated setting enhances the intensity and concentration of the practitioner. This visual presentation is an elegant interplay of motion and stillness, displaying the art form's discipline and grace.", 
-        "image_path": "images/mixkit-a-young-man-practicing-his-karate-moves-49635.png"
+        "image_path": "assets/images/mixkit-a-young-man-practicing-his-karate-moves-49635.png"
     },
     {
         "prompt": "In a serene and softly lit yoga studio, three individuals engage in a yoga session, each performing an upward-facing stretch. The central figure is a woman with shoulder-length brown hair, dressed in a light cropped top and green leggings, her posture reflecting grace and concentration. To her right, another participant, a woman in a purple outfit, mirrors the pose with equal poise. On her left, a person with a bun focuses intently, supported slightly by yoga blocks beneath their hands. The warm-colored wooden floor contrasts soothingly with the soft pastel mural on the back wall, featuring an abstract design and partial visage of a serene face. Natural light floods the space from a large window on the right, where lush greens peek through, adding an element of tranquility. In the corner of the room, a collection of meditation instruments, including a gong and a Buddha statue, subtly frame the peaceful setting. The mood is calm yet focused, as all three participants are deeply engaged in their practice. The scene combines elements of balance, harmony, and a shared journey towards mindfulness. This depiction captures the essence of a yoga session that blends personal growth with collective experience.", 
-        "image_path": "images/mixkit-small-group-of-people-doing-yoga-together-43730.png"
+        "image_path": "assets/images/mixkit-small-group-of-people-doing-yoga-together-43730.png"
     },
     {
         "prompt": "In the deep blue expanse of the ocean, two dolphins glide effortlessly, their sleek bodies reflecting the sunlight filtering through the water. The prominent shadows and caustics create a shimmering effect on their skin, capturing the beauty of their natural habitat. Each dolphin moves with a fluid grace, occasionally interacting with gentle nudges, showcasing their playful and social nature. The scene is vibrant and dynamic, with the clear blue background accentuating the dolphins' movements, making it an ideal subject for AI recreation.", 
-        "image_path": "images/mixkit-dolphins-underwater-4133.png"
+        "image_path": "assets/images/mixkit-dolphins-underwater-4133.png"
     },
     {
         "prompt": "A bustling ski slope comes alive with skiers descending a pristine, snow-covered hill, surrounded by towering, snow-draped evergreens. Several figures stand atop the slope, silhouetted against a clear blue sky, preparing to embark on their ski run. The chair lift on the right continuously drops off eager adventurers, adding to the excitement at the hilltop. Each skier, clad in colorful winter gear, carves distinct paths into the textured snow as they weave their way down. The interplay of sunlight and shadows accentuates the myriad tracks etched into the slope, creating a dynamic visual rhythm. The scene captures a vibrant winter wonderland, full of action and the thrill of a perfect ski day.", 
-        "image_path": "images/mixkit-skiers-on-a-snowy-slope-3327.png"
+        "image_path": "assets/images/mixkit-skiers-on-a-snowy-slope-3327.png"
     },
     {
         "prompt": "A determined climber is scaling a massive rock face, showcasing exceptional strength and skill. The person, clad in a teal shirt and dark pants, climbs with precision, their movements measured and deliberate. They are secured by climbing gear, which includes ropes and a harness, emphasizing their commitment to safety. The rugged texture of the sandy-colored rock provides an imposing backdrop, adding drama and scale to the climb. In the distance, other large rock formations and sparse vegetation can be seen under a bright, overcast sky, contributing to the natural and adventurous atmosphere. The scene captures a moment of focus and challenge, highlighting the climber's tenacity and the breathtaking environment.",
-        "image_path": "images/mixkit-alpinist-climbing-a-huge-rock-in-a-desert-43306.png"
+        "image_path": "assets/images/mixkit-alpinist-climbing-a-huge-rock-in-a-desert-43306.png"
     },
     {
         "prompt": "A silver SUV drives along a winding, snow-covered mountain road, with dense pine trees blanketed in snow lining both sides. The scene is serene, with the vehicle moving smoothly, possibly on a winter journey or vacation. As the SUV disappears around the bend, another, darker SUV follows, creating a sense of motion and perspective on the snow-dusted asphalt. The towering, snow-laden rock formation to the right contrasts with the dark green of the pines, highlighting the peacefulness of the wintry landscape.",
-        "image_path": "images/mixkit-curve-on-a-snowy-forest-road-3317.png"
+        "image_path": "assets/images/mixkit-curve-on-a-snowy-forest-road-3317.png"
     },
     {
         "prompt": "A solitary boat glides across the expansive, tranquil expanse of a serene lake. The vessel leaves a gentle wake behind, creating delicate ripples across the mirror-like surface. The water appears a rich shade of teal, seamlessly blending with the sky at the horizon. Silhouettes of distant trees are faintly visible, creating a picturesque backdrop that enhances the solitary journey of the boat. The sky is a calm gradient, shifting from soft oranges near the shore to the pale blues above. In the distance, a few slender poles emerge from the water, remnants of an old structure or natural formation. The mood of the scene is one of peace and solitude, with the boat journeying steadily through the quiet landscape. There is a sense of endless possibilities as the boat moves toward the unseen beyond the frame. The simplicity and stillness of the scene invite contemplation and reflection, encapsulating a perfect moment of quietude on the water.",
-        "image_path": "images/mixkit-motorboat-on-a-large-lake-with-turquoise-blue-waters-4996.png"
+        "image_path": "assets/images/mixkit-motorboat-on-a-large-lake-with-turquoise-blue-waters-4996.png"
     },
     {
         "prompt": "A man wearing grey shorts jumps rope in a gym, weights and gym equipment in the background.",
-        "image_path": "images/gray_short_man.jpg"
+        "image_path": "assets/images/gray_short_man.jpg"
     },
     {
         "prompt": "Flying over a peninsula covered in bushy trees, while discovering the sea around it, painted a beautiful turquoise blue, on a sunny day.",
-        "image_path": "images/peninsula.jpg"
+        "image_path": "assets/images/peninsula.jpg"
     },
     {
         "prompt": "Skillful cyclist doing a wheelie on a bike while riding through a forest, on a dirt road, surrounded by many trees, in the morning.",
-        "image_path": "images/cyclist.jpg"
+        "image_path": "assets/images/cyclist.jpg"
     },
     {
         "prompt": "Some friends dancing and having fun together in circles, at a party surrounded by colored lights at a party, in a fancy old place, in a view from below them.",
-        "image_path": "images/friends.jpg"
+        "image_path": "assets/images/friends.jpg"
     },
     {
         "prompt": "A saxophonist wearing a blazer dances while playing a song in a park.",
-        "image_path": "images/saxophonist.jpg"
+        "image_path": "assets/images/saxophonist.jpg"
     },
     {
         "prompt": "Romantic couple embracing and looking at each other in the middle of a forest, during a break on a road trip through nature.",
-        "image_path": "images/romance.jpg"
+        "image_path": "assets/images/romance.jpg"
     },
     {
         "prompt": "Man dressed in 80's style dances very happily in his kitchen while listening to music on his radio and drinking wine.",
-        "image_path": "images/80s_dance.jpg"
+        "image_path": "assets/images/80s_dance.jpg"
     },
     {
         "prompt": "Pair of jazz musicians performing a song with their saxophone and trombone on an abandoned train.",
-        "image_path": "images/jazz.jpg"
+        "image_path": "assets/images/jazz.jpg"
     },
     {
         "prompt": "A young woman with short hair wearing pink sunglasses chews gum and makes a bubble gum with the city in the background.",
-        "image_path": "images/pink.jpg"
+        "image_path": "assets/images/pink.jpg"
     },
     {
         "prompt": "Natural aerial landscape with a relief covered with abundant trees and vegetation and a thick layer of mist.",
-        "image_path": "images/natural.jpg"
+        "image_path": "assets/images/natural.jpg"
     },
     {
         "prompt": "Loving couple sitting on a log on the shore of a lake outside, sharing an affectionate hug.",
-        "image_path": "images/couple.jpg"
+        "image_path": "assets/images/couple.jpg"
     }
 ]

assets/videos/README.md

@@ -1,4 +1,4 @@
-# FastVideo/videos
+# FastVideo/assets/videos
 
 This folder is used to store **video assets for examples**, primarily **input videos** consumed by scripts under `FastVideo/examples/`.
 

demo/gradio_web_demo.py

@@ -1,195 +0,0 @@
-import argparse
-import os
-import tempfile
-
-import gradio as gr
-import torch
-from diffusers import FlowMatchEulerDiscreteScheduler
-from diffusers.utils import export_to_video
-
-from fastvideo.distill.solver import PCMFMScheduler
-from fastvideo.models.mochi_hf.modeling_mochi import MochiTransformer3DModel
-from fastvideo.models.mochi_hf.pipeline_mochi import MochiPipeline
-
-
-def init_args():
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--prompts", nargs="+", default=[])
-    parser.add_argument("--num_frames", type=int, default=25)
-    parser.add_argument("--height", type=int, default=480)
-    parser.add_argument("--width", type=int, default=848)
-    parser.add_argument("--num_inference_steps", type=int, default=8)
-    parser.add_argument("--guidance_scale", type=float, default=4.5)
-    parser.add_argument("--model_path", type=str, default="data/mochi")
-    parser.add_argument("--seed", type=int, default=12345)
-    parser.add_argument("--transformer_path", type=str, default=None)
-    parser.add_argument("--scheduler_type", type=str, default="pcm_linear_quadratic")
-    parser.add_argument("--lora_checkpoint_dir", type=str, default=None)
-    parser.add_argument("--shift", type=float, default=8.0)
-    parser.add_argument("--num_euler_timesteps", type=int, default=50)
-    parser.add_argument("--linear_threshold", type=float, default=0.1)
-    parser.add_argument("--linear_range", type=float, default=0.75)
-    parser.add_argument("--cpu_offload", action="store_true")
-    return parser.parse_args()
-
-
-def load_model(args):
-    if args.scheduler_type == "euler":
-        scheduler = FlowMatchEulerDiscreteScheduler()
-    else:
-        linear_quadratic = True if "linear_quadratic" in args.scheduler_type else False
-        scheduler = PCMFMScheduler(
-            1000,
-            args.shift,
-            args.num_euler_timesteps,
-            linear_quadratic,
-            args.linear_threshold,
-            args.linear_range,
-        )
-
-    if args.transformer_path:
-        transformer = MochiTransformer3DModel.from_pretrained(args.transformer_path)
-    else:
-        transformer = MochiTransformer3DModel.from_pretrained(args.model_path, subfolder="transformer/")
-
-    pipe = MochiPipeline.from_pretrained(args.model_path, transformer=transformer, scheduler=scheduler)
-    pipe.enable_vae_tiling()
-    # pipe.to(device)
-    # if args.cpu_offload:
-    pipe.enable_sequential_cpu_offload()
-    return pipe
-
-
-def generate_video(
-    prompt,
-    negative_prompt,
-    use_negative_prompt,
-    seed,
-    guidance_scale,
-    num_frames,
-    height,
-    width,
-    num_inference_steps,
-    randomize_seed=False,
-):
-    if randomize_seed:
-        seed = torch.randint(0, 1000000, (1, )).item()
-
-    generator = torch.Generator(device="cuda").manual_seed(seed)
-
-    if not use_negative_prompt:
-        negative_prompt = None
-
-    with torch.autocast("cuda", dtype=torch.bfloat16):
-        output = pipe(
-            prompt=[prompt],
-            negative_prompt=negative_prompt,
-            height=height,
-            width=width,
-            num_frames=num_frames,
-            num_inference_steps=num_inference_steps,
-            guidance_scale=guidance_scale,
-            generator=generator,
-        ).frames[0]
-
-    output_path = os.path.join(tempfile.mkdtemp(), "output.mp4")
-    export_to_video(output, output_path, fps=30)
-    return output_path, seed
-
-
-examples = [
-    "A hand enters the frame, pulling a sheet of plastic wrap over three balls of dough placed on a wooden surface. The plastic wrap is stretched to cover the dough more securely. The hand adjusts the wrap, ensuring that it is tight and smooth over the dough. The scene focuses on the hand’s movements as it secures the edges of the plastic wrap. No new objects appear, and the camera remains stationary, focusing on the action of covering the dough.",
-    "A vintage train snakes through the mountains, its plume of white steam rising dramatically against the jagged peaks. The cars glint in the late afternoon sun, their deep crimson and gold accents lending a touch of elegance. The tracks carve a precarious path along the cliffside, revealing glimpses of a roaring river far below. Inside, passengers peer out the large windows, their faces lit with awe as the landscape unfolds.",
-    "A crowded rooftop bar buzzes with energy, the city skyline twinkling like a field of stars in the background. Strings of fairy lights hang above, casting a warm, golden glow over the scene. Groups of people gather around high tables, their laughter blending with the soft rhythm of live jazz. The aroma of freshly mixed cocktails and charred appetizers wafts through the air, mingling with the cool night breeze.",
-]
-
-args = init_args()
-pipe = load_model(args)
-print("load model successfully")
-with gr.Blocks() as demo:
-    gr.Markdown("# Fastvideo Mochi Video Generation Demo")
-
-    with gr.Group():
-        with gr.Row():
-            prompt = gr.Text(
-                label="Prompt",
-                show_label=False,
-                max_lines=1,
-                placeholder="Enter your prompt",
-                container=False,
-            )
-            run_button = gr.Button("Run", scale=0)
-        result = gr.Video(label="Result", show_label=False)
-
-    with gr.Accordion("Advanced options", open=False):
-        with gr.Group():
-            with gr.Row():
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=1024,
-                    step=32,
-                    value=args.height,
-                )
-                width = gr.Slider(label="Width", minimum=256, maximum=1024, step=32, value=args.width)
-
-            with gr.Row():
-                num_frames = gr.Slider(
-                    label="Number of Frames",
-                    minimum=21,
-                    maximum=163,
-                    value=args.num_frames,
-                )
-                guidance_scale = gr.Slider(
-                    label="Guidance Scale",
-                    minimum=1,
-                    maximum=12,
-                    value=args.guidance_scale,
-                )
-                num_inference_steps = gr.Slider(
-                    label="Inference Steps",
-                    minimum=4,
-                    maximum=100,
-                    value=args.num_inference_steps,
-                )
-
-            with gr.Row():
-                use_negative_prompt = gr.Checkbox(label="Use negative prompt", value=False)
-            negative_prompt = gr.Text(
-                label="Negative prompt",
-                max_lines=1,
-                placeholder="Enter a negative prompt",
-                visible=False,
-            )
-
-            seed = gr.Slider(label="Seed", minimum=0, maximum=1000000, step=1, value=args.seed)
-            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
-            seed_output = gr.Number(label="Used Seed")
-
-    gr.Examples(examples=examples, inputs=prompt)
-
-    use_negative_prompt.change(
-        fn=lambda x: gr.update(visible=x),
-        inputs=use_negative_prompt,
-        outputs=negative_prompt,
-    )
-
-    run_button.click(
-        fn=generate_video,
-        inputs=[
-            prompt,
-            negative_prompt,
-            use_negative_prompt,
-            seed,
-            guidance_scale,
-            num_frames,
-            height,
-            width,
-            num_inference_steps,
-            randomize_seed,
-        ],
-        outputs=[result, seed_output],
-    )
-
-if __name__ == "__main__":
-    demo.queue(max_size=20).launch(server_name="0.0.0.0", server_port=7860)

demo/qualitative_results.md

@@ -1,15 +0,0 @@
-Fast-Hunyuan comparison with original Hunyuan, achieving an 8X diffusion speed boost with the FastVideo framework.
-
-https://github.com/user-attachments/assets/064ac1d2-11ed-4a0c-955b-4d412a96ef30
-
-Fast-Mochi comparison with original Mochi, achieving an 8X diffusion speed boost with the FastVideo framework.
-
-https://github.com/user-attachments/assets/5fbc4596-56d6-43aa-98e0-da472cf8e26c
-
-Comparison between OpenAI Sora, original Hunyuan and FastHunyuan
-
-https://github.com/user-attachments/assets/d323b712-3f68-42b2-952b-94f6a49c4836
-
-Comparison between original FastHunyuan, LLM-INT8 quantized FastHunyuan and NF4 quantized FastHunyuan
-
-https://github.com/user-attachments/assets/cf89efb5-5f68-4949-a085-f41c1ef26c94

docs/examples/examples_index.md

@@ -6,7 +6,7 @@ All documented examples are autogenerated using [generate_examples.py](https://g
 
 ## Examples
 
-- [Examples Distillation Index](distillation/examples/examples_distillation_index.md)
-- [Examples Training Index](training/examples/examples_training_index.md)
-- [Examples Inference Index](inference/examples/examples_inference_index.md)
+- [Examples Distillation Index](../distillation/examples/examples_distillation_index.md)
+- [Examples Training Index](../training/examples/examples_training_index.md)
+- [Examples Inference Index](../inference/examples/examples_inference_index.md)
 

docs/generate_examples.py

@@ -2,6 +2,7 @@
 # adapted from vllm: https://github.com/vllm-project/vllm/blob/v0.7.3/docs/source/generate_examples.py
 
 import itertools
+import os
 import re
 from dataclasses import dataclass, field
 from pathlib import Path
@@ -126,8 +127,20 @@ class Example:
         Raises:
             IndexError: If no Markdown files are found in the directory.
         """ # noqa: E501
-        return self.path if self.path.is_file() else list(
-            self.path.glob("*.md")).pop()
+        if self.path.is_file():
+            return self.path
+
+        markdown_files = sorted(self.path.glob("*.md"))
+        if not markdown_files:
+            raise IndexError(f"No Markdown files found in {self.path}")
+
+        readme_files = [
+            f for f in markdown_files if f.name.lower() == "readme.md"
+        ]
+        if readme_files:
+            return readme_files[0]
+
+        return markdown_files[0]
 
     def determine_other_files(self) -> list[Path]:
         """
@@ -521,11 +534,11 @@ def generate_examples(generate_main_index: bool = False) -> None:
             # Add to main index if it exists
             if generate_main_index and examples_index:
                 main_index_dir = examples_index.path.parent
-                rel_path = category_index.path.relative_to(
-                    main_index_dir.parent)
+                rel_path = os.path.relpath(category_index.path,
+                                           start=main_index_dir)
                 examples_index.documents.insert(
                     0,
-                    str(rel_path).replace(".md", ""))
+                    str(rel_path).replace("\\", "/").replace(".md", ""))
 
             # Write the category index file
             with open(category_index.path, "w+") as f:

docs/getting_started/installation.md

@@ -18,12 +18,25 @@ conda activate fastvideo
 pip install fastvideo
 ```
 
+### Using uv
+
+```bash
+# Create and activate a new uv environment
+uv venv --python 3.12 --seed
+source .venv/bin/activate
+
+uv pip install fastvideo
+```
+
 ### From source
 
 ```bash
 git clone https://github.com/hao-ai-lab/FastVideo.git
 cd FastVideo
 pip install -e .
+
+# or if you are using uv
+uv pip install -e .
 ```
 
 Also optionally install flash-attn:

docs/getting_started/quick_start.md

@@ -79,5 +79,6 @@ if __name__ == '__main__':
 
 - [Installation Guide](installation.md) - Detailed installation instructions
 - [Configuration](../inference/configuration.md) - Learn about configuration options
-- [Examples](../inference/examples/) - Explore more examples
+- [Examples](../inference/examples/examples_inference_index.md) - Explore more
+  examples
 - [Optimizations](../inference/optimizations.md) - Performance optimization tips

docs/inference/add_pipeline.md

@@ -101,14 +101,17 @@ Replace standard attention with FastVideo's optimized attention:
 ```python
 # Local attention patterns
 from fastvideo.attention import LocalAttention
-from fastvideo.attention.backends.abstract import _Backend
+from fastvideo.platforms.interface import AttentionBackendEnum
 self.attn = LocalAttention(
     num_heads=num_heads,
     head_size=head_dim,
     dropout_rate=0.0,
     softmax_scale=None,
     causal=False,
-    supported_attention_backends=(_Backend.FLASH_ATTN, _Backend.TORCH_SDPA)
+    supported_attention_backends=(
+        AttentionBackendEnum.FLASH_ATTN,
+        AttentionBackendEnum.TORCH_SDPA,
+    )
 )
 
 # Distributed attention for long sequences
@@ -119,14 +122,21 @@ self.attn = DistributedAttention(
     dropout_rate=0.0,
     softmax_scale=None,
     causal=False,
-    supported_attention_backends=(_Backend.SLIDING_TILE_ATTN, _Backend.FLASH_ATTN, _Backend.TORCH_SDPA)
+    supported_attention_backends=(
+        AttentionBackendEnum.SLIDING_TILE_ATTN,
+        AttentionBackendEnum.FLASH_ATTN,
+        AttentionBackendEnum.TORCH_SDPA,
+    )
 )
 ```
 
 #### Define supported backend selection
 
 ```python
-   _supported_attention_backends = (_Backend.FLASH_ATTN, _Backend.TORCH_SDPA)
+_supported_attention_backends = (
+    AttentionBackendEnum.FLASH_ATTN,
+    AttentionBackendEnum.TORCH_SDPA,
+)
 ```
 
 ### Registering Models

docs/inference/cli.md

@@ -1,104 +1,85 @@
 # FastVideo CLI Inference
 
-The FastVideo CLI provides a quick way to access the FastVideo inference pipeline for video generation. For more advanced usage,
-see the Python interface [here](examples/basic.md).
+The FastVideo CLI exposes the same core inference controls as the Python API.
 
 ## Basic Usage
 
-The basic command to generate a video is:
+Use either:
+
+1. `--model-path` + `--prompt`
+2. `--model-path` + `--prompt-txt` (batch prompts, one line per prompt)
+3. `--config` (JSON/YAML)
 
 ```bash
-fastvideo generate --model-path {MODEL_PATH} --prompt {PROMPT}
+fastvideo generate --model-path Wan-AI/Wan2.1-T2V-1.3B-Diffusers \
+  --prompt "A cat playing with a ball of yarn"
 ```
 
-### Required Parameters
+```bash
+fastvideo generate --model-path Wan-AI/Wan2.1-T2V-1.3B-Diffusers \
+  --prompt-txt prompts.txt
+```
 
-- `--model-path {MODEL_PATH}`: Path to the model or model ID
-- `--prompt {PROMPT}`: Text description for the video you want to generate
+You cannot provide both `--prompt` and `--prompt-txt` in the same run.
 
-## Common Arguments
-
-To see all the options, you can use the `--help` flag:
+## View All Arguments
 
 ```bash
 fastvideo generate --help
 ```
 
-### Hardware Configuration
+Arguments come from:
 
-- `--num-gpus {NUM_GPUS}`: Number of GPUs to use
-- `--tp-size {TP_SIZE}`: Tensor parallelism size (only for the encoder, should not be larger than 1 if text encoder offload is enabled, as layerwise offload + prefetch is faster)
-- `--sp-size {SP_SIZE}`: Sequence parallelism size (Typically should match the number of GPUs)
+- FastVideo runtime args (`FastVideoArgs`)
+- Sampling args (`SamplingParam`)
+- Pipeline config args (`PipelineConfig`)
 
-#### Video Configuration
+## Common Arguments
 
-- `--height {HEIGHT}`: Height of the generated video
-- `--width {WIDTH}`: Width of the generated video
-- `--num-frames {NUM_FRAMES}`: Number of frames to generate
-- `--fps {FPS}`: Frames per second for the saved video
+### Parallelism
 
-#### Generation Parameters
+- `--num-gpus`
+- `--sp-size`
+- `--tp-size`
 
-- `--num-inference-steps {STEPS}`: Number of denoising steps
-- `--negative-prompt {PROMPT}`: Negative prompt to guide generation away from certain concepts
-- `--seed {SEED}`: Random seed for reproducible generation
+### Sampling
 
-#### Output Options
+- `--num-frames`
+- `--height` / `--width`
+- `--num-inference-steps`
+- `--guidance-scale`
+- `--seed`
+- `--negative-prompt`
 
-- `--output-path {PATH}`: Directory to save the generated video
-- `--save-video`: Whether to save the video to disk
-- `--return-frames`: Whether to return the raw frames
+### Output
 
-## Using Configuration Files
+- `--output-path`
+- `--save-video` / `--no-save-video`
+- `--return-frames`
 
-Instead of specifying all parameters on the command line, you can use a configuration file:
+### Offloading and Performance
+
+- `--dit-layerwise-offload`
+- `--use-fsdp-inference`
+- `--text-encoder-cpu-offload`
+- `--image-encoder-cpu-offload`
+- `--vae-cpu-offload`
+- `--enable-torch-compile`
+- `--torch-compile-kwargs`
+
+## Using Config Files
 
 ```bash
-fastvideo generate --config {CONFIG_FILE_PATH}
+fastvideo generate --config config.yaml
 ```
 
-The config file should be in JSON or YAML format with the same parameter names as the CLI options. Command-line arguments will take precedence over settings in the configuration file, allowing you to override specific values while keeping the rest from the config file.
+Config files can be JSON or YAML. CLI flags override config-file values.
 
-Example configuration file (config.json):
-
-```json
-{
-    "model_path": "FastVideo/FastHunyuan-diffusers",
-    "prompt": "A beautiful woman in a red dress walking down a street",
-    "output_path": "outputs/",
-    "num_gpus": 2,
-    "sp_size": 2,
-    "tp_size": 1,
-    "num_frames": 45,
-    "height": 720,
-    "width": 1280,
-    "num_inference_steps": 6,
-    "seed": 1024,
-    "fps": 24,
-    "precision": "bf16",
-    "vae_precision": "fp16",
-    "vae_tiling": true,
-    "vae_sp": true,
-    "vae_config": {
-        "load_encoder": false,
-        "load_decoder": true,
-        "tile_sample_min_height": 256,
-        "tile_sample_min_width": 256
-    },
-    "text_encoder_precisions": [
-        "fp16",
-        "fp16"
-    ],
-    "mask_strategy_file_path": null,
-    "enable_torch_compile": false
-}
-```
-
-Or using YAML format (config.yaml):
+Example `config.yaml`:
 
 ```yaml
 model_path: "FastVideo/FastHunyuan-diffusers"
-prompt: "A beautiful woman in a red dress walking down a street"
+prompt: "A capybara lounging in a hammock"
 output_path: "outputs/"
 num_gpus: 2
 sp_size: 2
@@ -108,44 +89,34 @@ height: 720
 width: 1280
 num_inference_steps: 6
 seed: 1024
-fps: 24
-precision: "bf16"
+dit_precision: "bf16"
 vae_precision: "fp16"
 vae_tiling: true
 vae_sp: true
-vae_config:
-  load_encoder: false
-  load_decoder: true
-  tile_sample_min_height: 256
-  tile_sample_min_width: 256
-text_encoder_precisions:
-  - "fp16"
-  - "fp16"
-mask_strategy_file_path: null
 enable_torch_compile: false
 ```
 
+Notes:
+
+- Use `dit_precision` / `vae_precision` (not `precision`).
+- Nested config objects are supported, for example `vae_config` and
+  `dit_config`.
+
 ## Examples
 
-Generating a simple video:
+Simple generation:
 
 ```bash
-fastvideo generate --model-path FastVideo/FastHunyuan-diffusers --prompt "A cat playing with a ball of yarn" --num-frames 45 --height 720 --width 1280 --num-inference-steps 6 --seed 1024 --output-path outputs/
+fastvideo generate \
+  --model-path FastVideo/FastHunyuan-diffusers \
+  --prompt "A cat playing with a ball of yarn" \
+  --num-frames 45 --height 720 --width 1280 \
+  --num-inference-steps 6 --seed 1024 \
+  --output-path outputs/
 ```
 
-Using a negative prompt to avoid certain elements:
+Config + CLI override:
 
 ```bash
-fastvideo generate --model-path FastVideo/FastHunyuan-diffusers --prompt "A beautiful forest landscape" --negative-prompt "people, buildings, roads"
+fastvideo generate --config config.yaml --prompt "A panda skiing at sunset"
 ```
-
-Combining command line arguments and a configuration file:
-
-```bash
-fastvideo generate --config config.json --prompt "A capybara lounging in a hammock"
-```
-
-## Troubleshooting
-
-- If you encounter CUDA out-of-memory errors, try reducing the video dimensions or number of frames, or the number of inference steps.
-- For reproducible results, set the same seed value between runs.

docs/inference/configuration.md

@@ -1,4 +1,3 @@
-
 # Configuration
 
 ## Multi-GPU Setup
@@ -18,7 +17,8 @@ generator = VideoGenerator.from_pretrained(
 - `PipelineConfig`: Initialization time parameters
 - `SamplingParam`: Generation time parameters
 
-You can customize various parameters when generating videos using the `PipelineConfig` and `SamplingParam` class:
+You can customize generation behavior using `PipelineConfig` and
+`SamplingParam`:
 
 ```python
 from fastvideo import VideoGenerator, SamplingParam, PipelineConfig
@@ -27,12 +27,12 @@ def main():
     model_name = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
     config = PipelineConfig.from_pretrained(model_name)
     config.vae_precision = "fp16"
-    config.dit_cpu_offload = True
 
     # Create the generator
     generator = VideoGenerator.from_pretrained(
         model_name,
         num_gpus=1,
+        dit_layerwise_offload=True,  # FastVideoArgs option
         pipeline_config=config
     )
 
@@ -72,6 +72,34 @@ if __name__ == '__main__':
     main()
 ```
 
+## JSON/YAML Config Files (CLI)
+
+The CLI supports `--config` with JSON or YAML. Command-line arguments override
+config file values.
+
+```bash
+fastvideo generate --config config.yaml
+```
+
+Use CLI argument names as keys (underscore or hyphen is accepted). Example:
+
+```yaml
+model_path: "FastVideo/FastHunyuan-diffusers"
+prompt: "A capybara relaxing in a hammock"
+num_gpus: 2
+sp_size: 2
+num_frames: 45
+height: 720
+width: 1280
+num_inference_steps: 6
+seed: 1024
+dit_precision: "bf16"
+vae_precision: "fp16"
+vae_tiling: true
+vae_sp: true
+enable_torch_compile: false
+```
+
 ## Performance Optimization
 
 For configuring optimizations, please see our [optimizations guide](optimizations.md)

docs/inference/inference_quick_start.md

@@ -61,7 +61,8 @@ python example.py
 
 The generated video will be saved in the current directory under `my_videos/`  
 
-More inference example scripts can be found in `scripts/inference/`
+More inference scripts and recipes can be found in `examples/inference/` and
+`scripts/inference/`.
 
 ## Available Models
 
@@ -103,11 +104,10 @@ Common issues and their solutions:
 If you encounter CUDA out of memory errors:
 
 - Reduce `num_frames` or video resolution
-- Enable memory optimization with `enable_model_cpu_offload`
+- Enable FastVideo offloading options such as `dit_layerwise_offload=True`
+  (single GPU) or `use_fsdp_inference=True` (multi-GPU)
 - Try a smaller model or use distilled versions
 - Use `num_gpus` > 1 if multiple GPUs are available
-- Try enabling FSDP inference with `use_fsdp_inference=True` (may slow down generation)
-- Try enabling DiT layerwise offload with `dit_layerwise_offload=True` (now only a few models support this, but may introduce less overhead than FSDP)
 
 ### Slow Generation
 

docs/inference/optimizations.md

@@ -25,6 +25,9 @@ This page describes the various options for speeding up generation times in Fast
 - Video Sparse Attention: `FASTVIDEO_ATTENTION_BACKEND=VIDEO_SPARSE_ATTN`
 - Sage Attention: `FASTVIDEO_ATTENTION_BACKEND=SAGE_ATTN`
 - Sage Attention 3: `FASTVIDEO_ATTENTION_BACKEND=SAGE_ATTN_THREE`
+- Video MoBA Attention: `FASTVIDEO_ATTENTION_BACKEND=VMOBA_ATTN`
+- Sparse Linear Attention: `FASTVIDEO_ATTENTION_BACKEND=SLA_ATTN`
+- SageSLA Attention: `FASTVIDEO_ATTENTION_BACKEND=SAGE_SLA_ATTN`
 
 ### Configuring Backends
 
@@ -70,22 +73,15 @@ python setup.py install
 
 **`SLIDING_TILE_ATTN`**
 
-```bash
-pip install st_attn==0.0.4
-```
-
-Please see [this page](../attention/sta/index.md) for more installation instructions.
+Sliding Tile Attention is provided by `fastvideo-kernel`.
+See [STA docs](../attention/sta/index.md) for installation details.
 
 ### Video Sparse Attention
 
 **`VIDEO_SPARSE_ATTN`**
 
-```bash
-git submodule update --init --recursive
-python setup_vsa.py install
-```
-
-Please see [this page](../attention/vsa/index.md) for more installation instructions.
+Video Sparse Attention is provided by `fastvideo-kernel`.
+See [VSA docs](../attention/vsa/index.md) for installation details.
 
 ### Sage Attention
 
@@ -115,6 +111,12 @@ Note that Sage Attention 3 requires `python>=3.13`, `torch>=2.8.0`, `CUDA >=12.8
 
 To use Sage Attention 3 in FastVideo, follow the `README.md` in the linked repository to install the package from source.
 
+### V-MoBA / SLA / SageSLA
+
+These backends are model-specific and require the corresponding kernels and
+dependencies. Use the support matrix and model examples to confirm compatibility
+before enabling them.
+
 ## Teacache
 
 TeaCache is an optimization technique supported in FastVideo that can significantly speed up video generation by skipping redundant calculations across diffusion steps. This guide explains how to enable and configure TeaCache for optimal performance in FastVideo.

docs/inference/support_matrix.md

@@ -1,6 +1,10 @@
 # Compatibility Matrix
 
-The table below shows every supported model and optimizations supported for them.
+This page summarizes common model + optimization combinations.
+
+For the canonical, code-level list of model IDs recognized by
+`VideoGenerator.from_pretrained(...)`, see the registrations in
+`fastvideo/registry.py` (`register_configs(...)` entries).
 
 The symbols used have the following meanings:
 
@@ -10,7 +14,9 @@ The symbols used have the following meanings:
 
 ## Models x Optimization
 
-The `HuggingFace Model ID` can be directly pass to `from_pretrained()` methods and FastVideo will use the optimal default parameters when initializing and generating videos.
+The `HuggingFace Model ID` can be passed directly to
+`from_pretrained()`. FastVideo then uses model-specific default settings for
+pipeline initialization and sampling.
 
 <style>
   /* Target tables in this section */
@@ -53,9 +59,9 @@ The `HuggingFace Model ID` can be directly pass to `from_pretrained()` methods a
 | Wan2.1 T2V 14B | `Wan-AI/Wan2.1-T2V-14B-Diffusers` | 480P, 720P | ✅ | ✅* | ✅ | ⭕ | ⭕ |
 | Wan2.1 I2V 480P | `Wan-AI/Wan2.1-I2V-14B-480P-Diffusers` | 480P | ✅ | ✅* | ✅ | ⭕ | ⭕ |
 | Wan2.1 I2V 720P | `Wan-AI/Wan2.1-I2V-14B-720P-Diffusers` | 720P | ✅ | ✅ | ✅ | ⭕ | ⭕ |
-| StepVideo T2V | `FastVideo/stepvideo-t2v-diffusers` | 768px768px204f<br>544px992px204f<br>544px992px136f | ❌ | ❌ | ✅ | ⭕ | ⭕ |
 | TurboWan2.1 T2V 1.3B | `loayrashid/TurboWan2.1-T2V-1.3B-Diffusers` | 480P | ⭕ | ⭕ | ⭕ | ⭕ | ⭕ |
 | TurboWan2.1 T2V 14B | `loayrashid/TurboWan2.1-T2V-14B-Diffusers` | 480P, 720P | ⭕ | ⭕ | ⭕ | ⭕ | ⭕ |
+| TurboWan2.2 I2V A14B | `loayrashid/TurboWan2.2-I2V-A14B-Diffusers` | 480P<br>720P | ⭕ | ⭕ | ⭕ | ⭕ | ⭕ |
 | LongCat T2V 13.6B | See note** | 480P<br>720P | ❌ | ❌ | ❌ | ⭕ | ✅ |
 | Matrix Game 2.0 Base | `FastVideo/Matrix-Game-2.0-Base-Diffusers` | 352x640 | ⭕ | ⭕ | ⭕ | ⭕ | ⭕ |
 | Matrix Game 2.0 GTA | `FastVideo/Matrix-Game-2.0-GTA-Diffusers` | 352x640 | ⭕ | ⭕ | ⭕ | ⭕ | ⭕ |
@@ -63,10 +69,13 @@ The `HuggingFace Model ID` can be directly pass to `from_pretrained()` methods a
 
 **Note**: Wan2.2 TI2V 5B has some quality issues when performing I2V generation. We are working on fixing this issue.
 
-## Special requirements
+## Canonical Supported IDs
 
-### StepVideo T2V
-- The self-attention in text-encoder (step_llm) only supports CUDA capabilities sm_80 sm_86 and sm_90
+The authoritative source for model-ID recognition is
+`fastvideo/registry.py`. If a model ID is registered there, FastVideo can
+resolve default pipeline and sampling configuration for it.
+
+## Special requirements
 
 ### Sliding Tile Attention
 - Currently only Hopper GPUs (H100s) are supported.

docs/utilities/debugging.md

@@ -0,0 +1,75 @@
+# Debugging
+
+This page collects practical debugging steps for FastVideo inference issues.
+
+## Collect Environment Info
+
+From the repository root, run:
+
+```bash
+python collect_env.py
+```
+
+Attach the output when filing a GitHub issue.
+
+## Increase Logging
+
+FastVideo logging level is controlled by environment variables:
+
+```bash
+FASTVIDEO_LOGGING_LEVEL=DEBUG \
+FASTVIDEO_STAGE_LOGGING=1 \
+python your_script.py
+```
+
+Useful variables:
+
+- `FASTVIDEO_LOGGING_LEVEL`: `DEBUG`, `INFO`, `WARNING`, `ERROR`
+- `FASTVIDEO_STAGE_LOGGING`: print per-stage timings during pipeline execution
+- `FASTVIDEO_ATTENTION_BACKEND`: force an attention backend (for example
+  `TORCH_SDPA` or `FLASH_ATTN`)
+
+## Common Failure Modes
+
+### Out-of-memory
+
+Try, in order:
+
+1. Reduce `height`, `width`, `num_frames`, or `num_inference_steps`.
+2. Enable offloading flags such as `dit_layerwise_offload` (single GPU) or
+   `use_fsdp_inference` (multi-GPU).
+3. Enable `vae_cpu_offload`, `image_encoder_cpu_offload`, and
+   `text_encoder_cpu_offload`.
+
+See [Inference Offloading](../inference/offloading.md) for recommended
+combinations.
+
+### Attention backend import errors
+
+If forcing a backend fails, verify optional dependencies are installed:
+
+- `FLASH_ATTN`: `flash-attn`
+- `SLIDING_TILE_ATTN` and `VIDEO_SPARSE_ATTN`: `fastvideo-kernel`
+- `SAGE_ATTN` / `SAGE_ATTN_THREE`: SageAttention packages
+
+As a fallback, use:
+
+```bash
+export FASTVIDEO_ATTENTION_BACKEND=TORCH_SDPA
+```
+
+### Configuration parsing errors
+
+When using `--config`, keep keys aligned with CLI argument names (underscores or
+hyphens are both accepted). For nested config values, use nested objects
+(`vae_config`, `dit_config`) rather than dotted keys.
+
+## Issue Template
+
+When opening an issue, include:
+
+- exact command or Python snippet,
+- model ID/path,
+- full traceback,
+- `collect_env.py` output,
+- whether the problem reproduces with `FASTVIDEO_ATTENTION_BACKEND=TORCH_SDPA`.

examples/inference/basic/basic_cosmos2_5_i2w.py

@@ -20,7 +20,7 @@ def main():
     sampling_param = SamplingParam.from_pretrained(model_path)
 
     # image2world example from official repo
-    image_path = "images/bus_terminal.jpg"
+    image_path = "assets/images/bus_terminal.jpg"
 
     prompt = (
         "A nighttime city bus terminal gradually shifts from stillness to subtle movement. "
@@ -48,4 +48,3 @@ def main():
 if __name__ == "__main__":
     main()
 
-

examples/inference/basic/basic_cosmos2_5_v2w.py

@@ -20,7 +20,7 @@ def main():
     sampling_param = SamplingParam.from_pretrained(model_path)
 
     # video2world example from official repo
-    video_path = "videos/robot_pouring.mp4"
+    video_path = "assets/videos/robot_pouring.mp4"
 
     prompt = (
         "A robotic arm, primarily white with black joints and cables, is shown in a clean, modern indoor setting with a white tabletop. "
@@ -51,4 +51,3 @@ def main():
 if __name__ == "__main__":
     main()
 
-

examples/inference/basic/basic_gamecraft.py

@@ -0,0 +1,119 @@
+# SPDX-License-Identifier: Apache-2.0
+"""
+Basic inference script for HunyuanGameCraft video generation.
+
+HunyuanGameCraft generates game-like videos with camera/action control.
+It takes an optional image input and generates video with camera motion
+based on simple action commands (forward, left, right, backward, rotations).
+
+Available actions:
+    - forward (w): Move camera forward
+    - backward (s): Move camera backward  
+    - left (a): Move camera left (strafe)
+    - right (d): Move camera right (strafe)
+    - left_rot: Rotate camera left (pan)
+    - right_rot: Rotate camera right (pan)
+    - up_rot: Rotate camera up (tilt)
+    - down_rot: Rotate camera down (tilt)
+
+T2V vs I2V:
+    - Default: I2V (uses a default reference image). Set GAMECRAFT_I2V_IMAGE to a
+      URL or path to use a different image.
+    - T2V only (no reference image): run with GAMECRAFT_I2V_IMAGE= (empty).
+"""
+import os
+
+import torch
+
+from fastvideo import VideoGenerator
+from fastvideo.models.camera import create_camera_trajectory
+
+# Model configuration (use GAMECRAFT_MODEL_PATH for local weights)
+MODEL_PATH = os.environ.get("GAMECRAFT_MODEL_PATH", "FastVideo/HunyuanGameCraft-Diffusers")
+
+# Default prompts for demo
+DEFAULT_PROMPTS = {
+    "village": "A charming medieval village with cobblestone streets, thatched-roof houses, and vibrant flower gardens under a bright blue sky.",
+    "temple": "A majestic ancient temple stands under a clear blue sky, its grandeur highlighted by towering Doric columns and intricate architectural details.",
+    "forest": "A lush green forest with tall trees, dappled sunlight filtering through the leaves, and a winding dirt path.",
+    "beach": "A tropical beach with crystal clear turquoise water, white sand, and palm trees swaying in the breeze.",
+}
+
+# I2V: default reference image (URL). Can override with a local path.
+DEFAULT_I2V_IMAGE_URL = (
+    "https://huggingface.co/datasets/huggingface/documentation-images/"
+    "resolve/main/diffusers/astronaut.jpg"
+)
+DEFAULT_I2V_PROMPT = (
+    "An astronaut hatching from an egg, on the surface of the moon, "
+    "the darkness and depth of space realised in the background."
+)
+
+OUTPUT_PATH = "video_samples_gamecraft"
+
+
+def main():
+    # Initialize generator
+    # FastVideo will automatically download weights from HuggingFace
+    generator = VideoGenerator.from_pretrained(
+        MODEL_PATH,
+        num_gpus=1,
+        use_fsdp_inference=True,
+        dit_cpu_offload=True,
+        vae_cpu_offload=True,
+        text_encoder_cpu_offload=True,
+        pin_cpu_memory=True,
+    )
+
+    # Video parameters
+    height = 704
+    width = 1280
+    num_frames = 33
+    action = "forward"
+    action_speed = 0.2
+
+    # Create camera trajectory (Plücker coordinates)
+    camera_states = create_camera_trajectory(
+        action=action,
+        height=height,
+        width=width,
+        num_frames=num_frames,
+        action_speed=action_speed,
+        dtype=torch.bfloat16,
+    )
+    print(f"Camera states shape: {camera_states.shape}")
+
+    # I2V vs T2V: unset GAMECRAFT_I2V_IMAGE -> I2V (default image). Set to "" -> T2V.
+    env_image = os.environ.get("GAMECRAFT_I2V_IMAGE")
+    if env_image is None:
+        image_path = DEFAULT_I2V_IMAGE_URL  # default: I2V
+    elif env_image.strip() == "":
+        image_path = None  # T2V
+    else:
+        image_path = env_image.strip()  # I2V with given URL/path
+
+    is_i2v = image_path is not None
+    prompt = DEFAULT_I2V_PROMPT if is_i2v else DEFAULT_PROMPTS["temple"]
+    print(f"Mode: {'I2V' if is_i2v else 'T2V'}, prompt: {prompt[:60]}...")
+
+    gen_kw = dict(
+        prompt=prompt,
+        negative_prompt="",
+        camera_states=camera_states,
+        height=height,
+        width=width,
+        num_frames=num_frames,
+        num_inference_steps=50,
+        guidance_scale=6.0,
+        seed=42,
+        fps=24,
+        output_path=OUTPUT_PATH,
+        save_video=True,
+    )
+    if is_i2v:
+        gen_kw["image_path"] = image_path
+    generator.generate_video(**gen_kw)
+
+
+if __name__ == "__main__":
+    main()

examples/inference/basic/basic_lingbotworld_base_cam.py

@@ -0,0 +1,49 @@
+from fastvideo import VideoGenerator
+from fastvideo.models.dits.lingbotworld.cam_utils import prepare_camera_embedding
+
+# from fastvideo.configs.sample import SamplingParam
+OUTPUT_PATH = "video_samples_lingbotworld"
+def main():
+    # FastVideo will automatically use the optimal default arguments for the
+    # model.
+    # If a local path is provided, FastVideo will make a best effort
+    # attempt to identify the optimal arguments.
+    generator = VideoGenerator.from_pretrained(
+         "FastVideo/LingBot-World-Base-Cam-Diffusers",
+        # FastVideo will automatically handle distributed setup
+        num_gpus=1,
+        use_fsdp_inference=False, # set to True if GPU is out of memory
+        dit_cpu_offload=True, # DiT need to be offloaded for MoE
+        vae_cpu_offload=False,
+        text_encoder_cpu_offload=True,
+        # Set pin_cpu_memory to false if CPU RAM is limited and there're no frequent CPU-GPU transfer
+        pin_cpu_memory=True,
+        # image_encoder_cpu_offload=False,
+    )
+
+    num_frames = 81
+    prompt = "The video presents a soaring journey through a fantasy jungle. The wind whips past the rider's blue hands gripping the reins, causing the leather straps to vibrate. The ancient gothic castle approaches steadily, its stone details becoming clearer against the backdrop of floating islands and distant waterfalls."
+    image_path = "https://raw.githubusercontent.com/Robbyant/lingbot-world/main/examples/00/image.jpg"
+    action_path = "examples/inference/basic/lingbotworld_examples/00"
+    c2ws_plucker_emb, num_frames = prepare_camera_embedding(
+        action_path=action_path,
+        num_frames=num_frames,
+        height=480,
+        width=832,
+        spatial_scale=8,
+    )
+
+    generator.generate_video(
+        prompt,
+        image_path=image_path,
+        output_path=OUTPUT_PATH,
+        save_video=True,
+        num_frames=num_frames,
+        height=480,
+        width=832,
+        c2ws_plucker_emb=c2ws_plucker_emb,
+    )
+
+
+if __name__ == "__main__":
+    main()

examples/inference/basic/basic_ltx2.py

@@ -1,5 +1,6 @@
 from fastvideo import VideoGenerator
 
+
 PROMPT = (
     "A warm sunny backyard. The camera starts in a tight cinematic close-up "
     "of a woman and a man in their 30s, facing each other with serious "
@@ -16,19 +17,23 @@ PROMPT = (
 
 
 def main() -> None:
+    # Uses FastVideo default sampling settings for LTX2 base.
     generator = VideoGenerator.from_pretrained(
-        "FastVideo/LTX2-Distilled-Diffusers",
+        "Davids048/LTX2-Base-Diffusers",
         num_gpus=1,
     )
 
-    output_path = "outputs_video/ltx2_basic/output_ltx2_distilled_t2v.mp4"
+    output_path = "outputs_video/ltx2_basic/output_ltx2_base_t2v_1088_1920_1.1.mp4"
     generator.generate_video(
         prompt=PROMPT,
         output_path=output_path,
         save_video=True,
+        num_frames=121,
+        height=1088,
+        width=1920,
     )
     generator.shutdown()
 
 
 if __name__ == "__main__":
-    main()
+    main()

examples/inference/basic/basic_ltx2_distilled.py

@@ -0,0 +1,35 @@
+from fastvideo import VideoGenerator
+
+PROMPT = (
+    "A warm sunny backyard. The camera starts in a tight cinematic close-up "
+    "of a woman and a man in their 30s, facing each other with serious "
+    "expressions. The woman, emotional and dramatic, says softly, \"That's "
+    "it... Dad's lost it. And we've lost Dad.\" The man exhales, slightly "
+    "annoyed: \"Stop being so dramatic, Jess.\" A beat. He glances aside, "
+    "then mutters defensively, \"He's just having fun.\" The camera slowly "
+    "pans right, revealing the grandfather in the garden wearing enormous "
+    "butterfly wings, waving his arms in the air like he's trying to take "
+    "off. He shouts, \"Wheeeew!\" as he flaps his wings with full commitment. "
+    "The woman covers her face, on the verge of tears. The tone is deadpan, "
+    "absurd, and quietly tragic."
+)
+import os
+os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "FLASH_ATTN"
+
+def main() -> None:
+    generator = VideoGenerator.from_pretrained(
+        "FastVideo/LTX2-Distilled-Diffusers",
+        num_gpus=4,
+    )
+
+    output_path = "outputs_video/ltx2_basic/output_ltx2_distilled_t2v.mp4"
+    generator.generate_video(
+        prompt=PROMPT,
+        output_path=output_path,
+        save_video=True,
+    )
+    generator.shutdown()
+
+
+if __name__ == "__main__":
+    main()

examples/inference/basic/basic_sd35_t2i.py

@@ -0,0 +1,137 @@
+# SPDX-License-Identifier: Apache-2.0
+
+from __future__ import annotations
+
+import argparse
+import os
+import re
+from typing import List
+
+
+DEFAULT_PROMPTS = [
+    "a photo of a cat",
+    "a cinematic photo of a red panda wearing a tiny backpack, standing on a rainy neon-lit street at night, shallow depth of field, sharp focus, 35mm, bokeh",
+]
+
+
+def _safe_filename(text: str, max_len: int = 100) -> str:
+    """
+    Make a stable, filesystem-friendly filename base.
+    VideoGenerator uses prompt[:100].strip() internally, so we mirror that,
+    but also remove path separators and other problematic characters.
+    """
+    s = text[:max_len].strip()
+    s = s.replace(os.sep, "_")
+    if os.altsep:
+        s = s.replace(os.altsep, "_")
+    s = re.sub(r"\s+", " ", s)
+    s = re.sub(r"[^A-Za-z0-9 .,_-]", "_", s)
+    s = s.strip(" .")
+    return s or "prompt"
+
+
+def _remove_existing_outputs(out_dir: str, filename_base: str) -> None:
+    """
+    Ensure deterministic naming by deleting any existing outputs that would
+    cause VideoGenerator to append suffixes like _1, _2, etc.
+    """
+    if not os.path.isdir(out_dir):
+        return
+
+    pattern = re.compile(rf"^{re.escape(filename_base)}(_\d+)?\.(mp4|png)$")
+    for fn in os.listdir(out_dir):
+        if pattern.match(fn):
+            try:
+                os.remove(os.path.join(out_dir, fn))
+            except FileNotFoundError:
+                pass
+
+
+def parse_args() -> argparse.Namespace:
+    p = argparse.ArgumentParser(description="Run SD3.5 Medium text-to-image with FastVideo VideoGenerator.")
+    p.add_argument("--model-path", default="stabilityai/stable-diffusion-3.5-medium", help="Path to local diffusers-format SD3.5 weights directory.")
+    p.add_argument(
+        "--out-dir",
+        "--outdir",
+        default="outputs/sd35/samples",
+        help="Output directory for generated mp4 files.",
+    )
+    p.add_argument(
+        "--prompt",
+        action="append",
+        default=None,
+        help="Prompt text. Repeat --prompt multiple times to generate multiple samples.",
+    )
+    p.add_argument("--negative", default="lowres, blurry, jpeg artifacts, watermark, text", help="Negative prompt.")
+    p.add_argument(
+        "--backend",
+        default=None,
+        help="Set FASTVIDEO_ATTENTION_BACKEND (e.g. TORCH_SDPA). If omitted, respects the existing env var.",
+    )
+    p.add_argument("--seed", type=int, default=42, help="Base seed. Each prompt uses seed + prompt_idx.")
+    p.add_argument("--height", type=int, default=768, help="Output height.")
+    p.add_argument("--width", type=int, default=768, help="Output width.")
+    p.add_argument("--steps", type=int, default=28, help="Number of inference steps.")
+    p.add_argument("--guidance", type=float, default=6.0, help="Guidance scale.")
+    p.add_argument("--num-gpus", type=int, default=1, help="Number of GPUs to use.")
+    return p.parse_args()
+
+
+def main() -> None:
+    args = parse_args()
+    prompts: List[str] = args.prompt if args.prompt else DEFAULT_PROMPTS
+
+    if args.backend:
+        os.environ["FASTVIDEO_ATTENTION_BACKEND"] = args.backend
+
+    from fastvideo import VideoGenerator
+
+    os.makedirs(args.out_dir, exist_ok=True)
+
+    init_kwargs = {
+        "num_gpus": args.num_gpus,
+        "workload_type": "t2i",
+        "sp_size": 1,
+        "tp_size": 1,
+        "dit_cpu_offload": False,
+        "dit_layerwise_offload": False,
+        "text_encoder_cpu_offload": False,
+        "vae_cpu_offload": False,
+        "image_encoder_cpu_offload": False,
+        "pin_cpu_memory": False,
+        "use_fsdp_inference": False,
+    }
+
+    generator = VideoGenerator.from_pretrained(model_path=args.model_path, **init_kwargs)
+    try:
+        for i, prompt in enumerate(prompts):
+            seed = args.seed + i
+
+            filename_base = f"sd35_{i:02d}_seed{seed}_{_safe_filename(prompt, max_len=80)}"
+            _remove_existing_outputs(args.out_dir, filename_base)
+
+            output_path = os.path.join(args.out_dir, f"{filename_base}.png")
+            print(f"[sd35] prompt_idx={i} seed={seed} output_path={output_path}")
+
+            generation_kwargs = {
+                "output_path": output_path,
+                "height": args.height,
+                "width": args.width,
+                "num_frames": 1,
+                "fps": 1,
+                "num_inference_steps": args.steps,
+                "guidance_scale": args.guidance,
+                "seed": seed,
+                "negative_prompt": args.negative,
+                "save_video": True,
+            }
+
+            generator.generate_video(prompt, **generation_kwargs)
+
+        print(f"[sd35] done. outputs written to: {args.out_dir}")
+    finally:
+        generator.shutdown()
+
+
+if __name__ == "__main__":
+    main()

examples/inference/basic/basic_self_forcing_causal_wan2_2_i2v.py

@@ -31,7 +31,7 @@ def main():
     sampling_param.height = 480
     sampling_param.seed = 1000
 
-    with open("prompts/mixkit_i2v.jsonl", "r") as f:
+    with open("assets/prompts/mixkit_i2v.jsonl", "r") as f:
         prompt_image_pairs = json.load(f)
 
     for prompt_image_pair in prompt_image_pairs:

examples/inference/gradio/serving/gradio_frontend.py

@@ -188,8 +188,9 @@ def load_example_prompts():
     prompt_to_image = {}
     # Try to find the JSON file relative to project root
     possible_json_paths = [
-        Path("prompts/mixkit_i2v.jsonl"),
-        Path(__file__).parent.parent.parent.parent / "prompts" / "mixkit_i2v.jsonl",
+        Path("assets/prompts/mixkit_i2v.jsonl"),
+        Path(__file__).resolve().parents[4] / "assets" / "prompts" /
+        "mixkit_i2v.jsonl",
     ]
     json_path = None
     for path in possible_json_paths:
@@ -201,8 +202,8 @@ def load_example_prompts():
         try:
             with open(json_path, "r", encoding='utf-8') as f:
                 data = json.load(f)
-                # Get the project root directory (parent of prompts directory)
-                project_root = json_path.parent.parent
+                # Resolve paths relative to repository root.
+                project_root = Path(__file__).resolve().parents[4]
                 for item in data:
                     prompt_text = item.get("prompt", "").strip()
                     image_path = item.get("image_path", "")
@@ -736,8 +737,8 @@ def main():
         allowed_paths=[
             os.path.abspath("outputs"), 
             os.path.abspath("fastvideo-logos"),
-            os.path.abspath("prompts"),
-            os.path.abspath("images"),
+            os.path.abspath("assets/prompts"),
+            os.path.abspath("assets/images"),
             os.path.abspath(tempfile.gettempdir()),
             os.path.abspath(os.path.join(tempfile.gettempdir(), "gradio")),
         ]
@@ -747,4 +748,4 @@ def main():
 
 
 if __name__ == "__main__":
-    main() 
+    main() 

examples/inference/sta_mask_search/inference_wan_sta.sh

@@ -5,7 +5,7 @@ export FASTVIDEO_ATTENTION_BACKEND=SLIDING_TILE_ATTN
 export MODEL_BASE=Wan-AI/Wan2.1-T2V-14B-Diffusers
 
 base_port=29503
-num_gpu=1
+num_gpu=8
 gpu_ids=$(seq 0 $((num_gpu-1)))
 skip_time_steps=12
 

examples/training/consistency_finetune/causal_ode_init/finetune_ode_init.sh

@@ -4,7 +4,7 @@ export WANDB_BASE_URL="https://api.wandb.ai"
 export WANDB_MODE=online
 export TOKENIZERS_PARALLELISM=false
 
-MODEL_PATH="Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
+MODEL_PATH="wlsaidhi/SFWan2.1-T2V-1.3B-Diffusers"
 DATA_DIR="data/crush-smol_processed_t2v_1_3b_ode_init/"
 VALIDATION_DATASET_FILE="$(dirname "$0")/validation.json"
 NUM_GPUS=1
@@ -14,7 +14,6 @@ NUM_GPUS=1
 training_args=(
   --tracker_project_name "wan_ode_init"
   --output_dir "wan_ode_init_crush_smol"
-  --override_transformer_cls_name "CausalWanTransformer3DModel"
   --wandb_run_name "wan_ode_init_crush_smol"
   --max_train_steps 6000
   --train_batch_size 1
@@ -34,7 +33,7 @@ parallel_args=(
   --sp_size 1
   --tp_size 1
   --hsdp_replicate_dim 1
-  --hsdp_shard_dim 1
+  --hsdp_shard_dim $NUM_GPUS
 )
 
 # Model arguments
@@ -51,20 +50,17 @@ dataset_args=(
 
 # Validation arguments
 validation_args=(
-  --log_validation
-  --validation_dataset_file "$VALIDATION_DATASET_FILE"
-  --validation_steps 50
-  --validation_sampling_steps "50"
-  --validation_guidance_scale "6.0"
+  --log-visualization
+  --visualization-steps 100
 )
 
 # Optimizer arguments
 optimizer_args=(
-  --learning_rate 6e-6
+  --learning_rate 1e-5
   --mixed_precision "bf16"
   --weight_only_checkpointing_steps 1000
   --training_state_checkpointing_steps 1000
-  --weight_decay 1e-4
+  --weight_decay 0.01
   --max_grad_norm 1.0
 )
 

examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/I2V-14B-VSA.slurm

@@ -44,8 +44,8 @@ VALIDATION_DATASET_FILE=your_validation_dataset_file
 # export CUDA_VISIBLE_DEVICES=4,5
 # IP=[MASTER NODE IP]
 
-# Training arguments
-training_args=(
+# Core training arguments
+core_training_args=(
   --tracker_project_name wan_i2v_VSA
   --output_dir "checkpoints/wan_i2v_finetune_VSA"
   --max_train_steps 4000
@@ -53,10 +53,14 @@ training_args=(
   --train_sp_batch_size 1
   --gradient_accumulation_steps 1
   --num_latent_t 21
+  --enable_gradient_checkpointing_type "full"
+)
+
+# Validation generation shape arguments (used during training validation)
+validation_generation_shape_args=(
   --num_height 720
   --num_width 1280
   --num_frames 81
-  --enable_gradient_checkpointing_type "full"
 )
 
 # Parallel arguments
@@ -127,8 +131,9 @@ srun torchrun \
     "${parallel_args[@]}" \
     "${model_args[@]}" \
     "${dataset_args[@]}" \
-    "${training_args[@]}" \
+    "${core_training_args[@]}" \
+    "${validation_generation_shape_args[@]}" \
     "${optimizer_args[@]}" \
     "${validation_args[@]}" \
     "${miscellaneous_args[@]}" \
-    "${vsa_args[@]}"  
+    "${vsa_args[@]}"  

examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/README.md

@@ -6,9 +6,7 @@ These are e2e example scripts for finetuning Wan2.1 T2V with VSA to accelerate i
 
 ## Make sure you have installed VSA
 
-```bash
-pip install vsa
-```
+Go to fastvideo-kernel/README.md for instructions.
 
 ### Download the synthetic dataset:
 

examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/T2V-1.3B-VSA.slurm

@@ -44,8 +44,8 @@ VALIDATION_DATASET_FILE=your_validation_dataset_file
 # export CUDA_VISIBLE_DEVICES=4,5
 # IP=[MASTER NODE IP]
 
-# Training arguments
-training_args=(
+# Core training arguments
+core_training_args=(
   --tracker_project_name wan_t2v_VSA
   --output_dir "checkpoints/wan_t2v_finetune_VSA"
   --max_train_steps 4000
@@ -53,10 +53,14 @@ training_args=(
   --train_sp_batch_size 1
   --gradient_accumulation_steps 1
   --num_latent_t 21
+  # --enable_gradient_checkpointing_type "full" # if OOM enable this
+)
+
+# Validation generation shape arguments (used during training validation)
+validation_generation_shape_args=(
   --num_height 480
   --num_width 832
   --num_frames 81
-  # --enable_gradient_checkpointing_type "full" # if OOM enable this
 )
 
 # Parallel arguments
@@ -127,8 +131,9 @@ srun torchrun \
     "${parallel_args[@]}" \
     "${model_args[@]}" \
     "${dataset_args[@]}" \
-    "${training_args[@]}" \
+    "${core_training_args[@]}" \
+    "${validation_generation_shape_args[@]}" \
     "${optimizer_args[@]}" \
     "${validation_args[@]}" \
     "${miscellaneous_args[@]}" \
-    "${vsa_args[@]}"  
+    "${vsa_args[@]}"  

examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/T2V-1.3B-VSA_4_gpu_debug.sh

@@ -0,0 +1,114 @@
+
+
+# Basic info
+export WANDB_MODE="online"
+export NCCL_P2P_DISABLE=1
+export TORCH_NCCL_ENABLE_MONITORING=0
+export TRITON_CACHE_DIR="/tmp/triton_cache_${USER}_$$"
+export MASTER_ADDR="${MASTER_ADDR:-127.0.0.1}"
+export MASTER_PORT="${MASTER_PORT:-29500}"
+export NODE_RANK=0
+export TOKENIZERS_PARALLELISM=false
+export WANDB_BASE_URL="https://api.wandb.ai"
+export FASTVIDEO_ATTENTION_BACKEND=VIDEO_SPARSE_ATTN
+export WANDB_API_KEY="your_wandb_api_key_here" # TODO: Replace with your actual key or load from a secure location
+
+# Configs
+NUM_GPUS=4
+MODEL_PATH="Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
+DATA_DIR=data/Wan-Syn_77x448x832_600k
+VALIDATION_DATASET_FILE=examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/validation_4.json
+
+# Core training arguments
+core_training_args=(
+  --tracker_project_name wan_t2v_VSA
+  --output_dir "checkpoints/wan_t2v_finetune_VSA"
+  --max_train_steps 4000
+  --train_batch_size 1
+  --train_sp_batch_size 1
+  --gradient_accumulation_steps 1
+  --num_latent_t 20
+  --enable_gradient_checkpointing_type "full" # if OOM enable this
+)
+
+# Validation generation shape arguments (used during training validation)
+validation_generation_shape_args=(
+  --num_height 448
+  --num_width 832
+  --num_frames 77
+)
+
+# Parallel arguments
+parallel_args=(
+  --num_gpus $NUM_GPUS
+  --sp_size 1
+  --tp_size 1
+  --hsdp_replicate_dim $NUM_GPUS
+  --hsdp_shard_dim 1
+)
+
+# Model arguments
+model_args=(
+  --model_path "$MODEL_PATH"
+  --pretrained_model_name_or_path "$MODEL_PATH"
+)
+
+# Dataset arguments
+dataset_args=(
+  --data_path "$DATA_DIR"
+  --dataloader_num_workers 4
+)
+
+# Validation arguments
+validation_args=(
+  --log_validation
+  --validation_dataset_file "$VALIDATION_DATASET_FILE"
+  --validation_steps 200
+  --validation_sampling_steps "50"
+  --validation_guidance_scale "5.0"
+)
+
+# Optimizer arguments
+optimizer_args=(
+  --learning_rate 1e-6
+  --mixed_precision "bf16"
+  --weight_only_checkpointing_steps 1000
+  --training_state_checkpointing_steps 1000
+  --weight_decay 0.01
+  --max_grad_norm 1.0
+)
+
+# Miscellaneous arguments
+miscellaneous_args=(
+  --inference_mode False
+  --checkpoints_total_limit 3
+  --training_cfg_rate 0.1
+  --dit_precision "fp32"
+  --ema_start_step 0
+  --flow_shift 1
+  --seed 1000
+)
+
+# VSA arguments
+vsa_args=(
+  --VSA_decay_rate 0.03
+  --VSA_decay_interval_steps 50
+  --VSA_sparsity 0.9
+)
+
+torchrun \
+  --nnodes 1 \
+  --nproc_per_node "$NUM_GPUS" \
+  --node_rank "$NODE_RANK" \
+  --rdzv_backend c10d \
+  --rdzv_endpoint "$MASTER_ADDR:$MASTER_PORT" \
+  fastvideo/training/wan_training_pipeline.py \
+  "${parallel_args[@]}" \
+  "${model_args[@]}" \
+  "${dataset_args[@]}" \
+  "${core_training_args[@]}" \
+  "${validation_generation_shape_args[@]}" \
+  "${optimizer_args[@]}" \
+  "${validation_args[@]}" \
+  "${miscellaneous_args[@]}" \
+  "${vsa_args[@]}"

examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/T2V-14B-VSA.slurm

@@ -44,8 +44,8 @@ VALIDATION_DATASET_FILE=your_validation_dataset_file
 # export CUDA_VISIBLE_DEVICES=4,5
 # IP=[MASTER NODE IP]
 
-# Training arguments
-training_args=(
+# Core training arguments
+core_training_args=(
   --tracker_project_name wan_t2v_VSA
   --output_dir "checkpoints/wan_t2v_finetune_VSA"
   --max_train_steps 4000
@@ -53,10 +53,14 @@ training_args=(
   --train_sp_batch_size 1
   --gradient_accumulation_steps 1
   --num_latent_t 21
+  --enable_gradient_checkpointing_type "full"
+)
+
+# Validation generation shape arguments (used during training validation)
+validation_generation_shape_args=(
   --num_height 720
   --num_width 1280
   --num_frames 81
-  --enable_gradient_checkpointing_type "full"
 )
 
 # Parallel arguments
@@ -127,8 +131,9 @@ srun torchrun \
     "${parallel_args[@]}" \
     "${model_args[@]}" \
     "${dataset_args[@]}" \
-    "${training_args[@]}" \
+    "${core_training_args[@]}" \
+    "${validation_generation_shape_args[@]}" \
     "${optimizer_args[@]}" \
     "${validation_args[@]}" \
     "${miscellaneous_args[@]}" \
-    "${vsa_args[@]}"  
+    "${vsa_args[@]}"  

examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/download_dataset.sh

@@ -1,7 +1,8 @@
 #!/bin/bash
+# Note: for debugging, you do not need to download them all and just control c.
+
 
 # 480P dataset
-python scripts/huggingface/download_hf.py --repo_id "FastVideo/Wan-Syn_77x448x832_600k" --local_dir "FastVideo/Wan-Syn_77x448x832_600k" --repo_type "dataset"
-
+python scripts/huggingface/download_hf.py --repo_id "FastVideo/Wan-Syn_77x448x832_600k" --local_dir "data/Wan-Syn_77x448x832_600k" --repo_type "dataset"
 # 720P dataset
-python scripts/huggingface/download_hf.py --repo_id "FastVideo/Wan-Syn_77x768x1280_250k" --local_dir "FastVideo/Wan-Syn_77x768x1280_250k" --repo_type "dataset"
+python scripts/huggingface/download_hf.py --repo_id "FastVideo/Wan-Syn_77x768x1280_250k" --local_dir "data/Wan-Syn_77x768x1280_250k" --repo_type "dataset"

examples/training/finetune/Wan2.1-VSA/Wan-Syn-Data/validation_4.json

@@ -0,0 +1,36 @@
+{
+  "data": [
+    {
+      "caption": "In the video, a woman is elegantly showcasing her earrings, bringing attention to their intricate design with a gentle touch of her fingers. She is bathed in ambient purple and pink lighting, which casts a soft glow on her delicate features and enhances the vivid tones of her lipstick and eye makeup. Her hair is styled to frame her face smoothly, emphasizing the contours of her jawline and cheekbones. The background features a blurred neon light, adding an artistic and modern touch to the overall aesthetic.",
+      "video_path": "Fashion/mixkit-face-of-an-elegant-and-captivating-woman-41914_clip_1.mp4",
+      "num_inference_steps":  3,
+      "height": 448,
+      "width": 832,
+      "num_frames": 61
+    },
+    {
+      "caption": "In the video, a lone rider guides a majestic horse across an expansive, open field as the sun sets in the background. The rider, dressed in a classic blue shirt and wide-brimmed hat, sits confidently in the saddle, silhouetted against the warm glow of the evening sky. The horse moves gracefully, its mane and tail flowing with each step, creating a sense of harmony between horse and rider. Surrounding the pair, towering trees form a natural border, their leaves gently rustling in the breeze. The shadows lengthen on the ground, accentuating the serene and timeless feel of the scene. The distant hills and wooden fences frame the horizon, adding depth to the tranquil landscape. A few horses graze peacefully in the background, blending into the pastoral setting. The overall ambiance evokes a sense of calmness and quietude, capturing a perfect moment in the golden light of dusk.",
+      "video_path": "Man/mixkit-a-rancher-riding-a-horse-at-sunset-1143_clip_1.mp4",
+      "num_inference_steps":  3,
+      "height": 448,
+      "width": 832,
+      "num_frames": 61
+    },
+    {
+      "caption": "In a dimly lit, eerie setting, a mysterious pink bottle labeled \"Authentic 100% organic POISON\" sits prominently in the foreground, casting a menacing aura. The bottle is accentuated by green fog, which swirls lightly around it, enhancing its sinister allure. Behind it, a shadowy golden bottle adorned with a spider emblem subtly emerges, adding an extra layer of mystery to the scene. Dim candles provide faint, flickering light, which complements the dark atmosphere, making the setting ideal for an illusion of hidden dangers.",
+      "video_path": "smoke/mixkit-poison-in-halloween-ritual-33879_clip_1.mp4",
+      "num_inference_steps":  3,
+      "height": 448,
+      "width": 832,
+      "num_frames": 61
+    },
+    {
+      "caption": "The video opens with a tranquil scene in the heart of a dense forest, emphasizing two large, textured tree trunks in the foreground framing the view. Sunlight filters through the canopy above, casting intricate patterns of light and shadow on the trees and the ground. Between the tree trunks, a clear view of a calm, muddy river unfolds, its surface shimmering under the gentle sunlight. The riverbank is decorated with a variety of small bushes and vibrant foliage, subtly transitioning into the deep greens of tall, leafy plants. In the background, the dense forest looms, filled with dark, towering trees, their branches intertwining to form an intricate canopy. The scene is bathed in the soft glow of the sun, creating a serene and picturesque setting. Occasional sunbeams pierce through the foliage, adding a magical aura to the landscape. The vibrant reds and oranges of the smaller plants add contrast, bringing warmth to the earthy tones of the scenery. Overall, this harmonious blend of natural elements creates a peaceful and idyllic forest setting.",
+      "video_path": "forest/mixkit-view-of-a-river-between-two-old-trees-560_clip_1.mp4",
+      "num_inference_steps":  3,
+      "height": 448,
+      "width": 832,
+      "num_frames": 61
+    }
+  ]
+}

examples/training/finetune/ltx2/.gitignore

@@ -0,0 +1,3 @@
+*.err
+*.out
+*slurm_logs/*

examples/training/finetune/ltx2/README.md

@@ -0,0 +1,25 @@
+# LTX-2 Crush-Smol Example
+# TODO: Update this doc.
+
+These are e2e example scripts for finetuning LTX-2 on the crush-smol dataset.
+
+## Execute the following commands from `FastVideo/` to run training:
+
+### Download crush-smol dataset:
+
+`bash examples/training/finetune/ltx2/download_dataset.sh`
+
+#### Or use the scripts at scripts/dataset_preparation to download and prepare the dataset
+
+### Preprocess the videos and captions into latents:
+
+`bash examples/training/finetune/ltx2/preprocess_ltx2_data_t2v_new.sh`
+
+### Edit the following file and run finetuning:
+
+`bash examples/training/finetune/ltx2/finetune_t2v.sh`
+
+Notes:
+- Update `DATASET_PATH` in the preprocess script to point to your merged dataset root (`videos/` + `videos2caption.json`).
+- `MODEL_PATH` should point to a local LTX-2 diffusers-style directory that contains `model_index.json` and `text_encoder/gemma`.
+

examples/training/finetune/ltx2/download_dataset.sh

@@ -0,0 +1,3 @@
+# #!/bin/bash
+#
+python scripts/huggingface/download_hf.py --repo_id "wlsaidhi/crush-smol-merged" --local_dir "data/crush-smol" --repo_type "dataset"

examples/training/finetune/ltx2/finetune_t2v.sh

@@ -0,0 +1,93 @@
+#!/bin/bash
+
+export WANDB_BASE_URL="https://api.wandb.ai"
+export WANDB_MODE=online
+export TOKENIZERS_PARALLELISM=false
+
+MODEL_PATH="FastVideo/LTX2-Distilled-Diffusers"
+DATA_DIR="data/crush-smol"
+VALIDATION_DATASET_FILE="$(dirname "$0")/validation.json"
+echo  VALIDATION_DATASET_FILE: $VALIDATION_DATASET_FILE
+NUM_GPUS=4
+HEIGHT=1088
+WIDTH=1920
+FRAMES=121
+
+training_args=(
+  --tracker_project_name "ltx2_t2v_finetune"
+  --output_dir "checkpoints/ltx2_t2v_finetune"
+  --max_train_steps 5000
+  --train_batch_size 1
+  --train_sp_batch_size 1
+  --gradient_accumulation_steps 1
+  --num_latent_t 16
+  --num_height $HEIGHT
+  --num_width $WIDTH
+  --num_frames $FRAMES
+  --enable_gradient_checkpointing_type "full"
+  --mode "finetuning"
+)
+
+parallel_args=(
+  --num_gpus $NUM_GPUS
+  --sp_size 2
+  --tp_size 1
+  --hsdp_replicate_dim 1
+  --hsdp_shard_dim $NUM_GPUS
+)
+
+model_args=(
+  --model_path $MODEL_PATH
+  --pretrained_model_name_or_path $MODEL_PATH
+)
+
+dataset_args=(
+  --data_path $DATA_DIR
+  --dataloader_num_workers 1
+)
+
+validation_args=(
+  --log_validation
+  --validation_dataset_file $VALIDATION_DATASET_FILE
+  --validation_steps 20
+  --validation_sampling_steps "8"
+  --validation_guidance_scale "1.0"
+)
+
+optimizer_args=(
+  --learning_rate 1e-5
+  --mixed_precision "bf16"
+  --weight_only_checkpointing_steps 1000
+  --training_state_checkpointing_steps 1000
+  --weight_decay 1e-4
+  --max_grad_norm 1.0
+  --lr_scheduler "linear"
+)
+
+miscellaneous_args=(
+  --inference_mode False
+  --checkpoints_total_limit 3
+  --dit_precision "fp32"
+  --dit_cpu_offload False
+  --dit_layerwise_offload False
+  --text_encoder_cpu_offload False
+  --image_encoder_cpu_offload False
+  --vae_cpu_offload False
+  --ltx2-first-frame-conditioning-p 0.0
+)
+
+# NOTE: Setting this environment variable to TORCH_SDPA to avoid the issue of stacking that failed in flash attn. 
+
+
+torchrun \
+  --nnodes 1 \
+  --master_port 29501 \
+  --nproc_per_node $NUM_GPUS \
+    fastvideo/training/ltx2_training_pipeline.py \
+    "${parallel_args[@]}" \
+    "${model_args[@]}" \
+    "${dataset_args[@]}" \
+    "${training_args[@]}" \
+    "${optimizer_args[@]}" \
+    "${validation_args[@]}" \
+    "${miscellaneous_args[@]}"

examples/training/finetune/ltx2/finetune_t2v_lora.sh

@@ -0,0 +1,80 @@
+#!/bin/bash
+
+export WANDB_BASE_URL="https://api.wandb.ai"
+export WANDB_MODE=online
+export TOKENIZERS_PARALLELISM=false
+
+MODEL_PATH="FastVideo/LTX2-Distilled-Diffusers"
+DATA_DIR="data/crush-smol"
+VALIDATION_DATASET_FILE="$(dirname "$0")/validation.json"
+NUM_GPUS=1
+
+training_args=(
+  --tracker_project_name "ltx2_t2v_lora_finetune"
+  --output_dir "checkpoints/ltx2_t2v_lora_finetune"
+  --max_train_steps 2000
+  --train_batch_size 1
+  --train_sp_batch_size 1
+  --gradient_accumulation_steps 8
+  --num_latent_t 10
+  --num_height 480
+  --num_width 832
+  --num_frames 77
+  --ltx2-first-frame-conditioning-p 0.1
+  --enable_gradient_checkpointing_type "full"
+)
+
+parallel_args=(
+  --num_gpus $NUM_GPUS
+  --sp_size $NUM_GPUS
+  --tp_size 1
+  --hsdp_replicate_dim 1
+  --hsdp_shard_dim $NUM_GPUS
+)
+
+model_args=(
+  --model_path $MODEL_PATH
+  --pretrained_model_name_or_path $MODEL_PATH
+)
+
+dataset_args=(
+  --data_path $DATA_DIR
+  --dataloader_num_workers 1
+)
+
+validation_args=(
+  --log_validation
+  --validation_dataset_file $VALIDATION_DATASET_FILE
+  --validation_steps 200
+  --validation_sampling_steps "50"
+  --validation_guidance_scale "3.0"
+)
+
+optimizer_args=(
+  --learning_rate 2e-4
+  --mixed_precision "bf16"
+  --weight_only_checkpointing_steps 1000
+  --training_state_checkpointing_steps 1000
+  --weight_decay 1e-4
+  --max_grad_norm 1.0
+  --lora_training True
+  --lora_rank 16
+  --lora_alpha 16
+)
+
+miscellaneous_args=(
+  --inference_mode False
+  --checkpoints_total_limit 3
+)
+
+torchrun \
+  --nnodes 1 \
+  --nproc_per_node $NUM_GPUS \
+    fastvideo/training/ltx2_training_pipeline.py \
+    "${parallel_args[@]}" \
+    "${model_args[@]}" \
+    "${dataset_args[@]}" \
+    "${training_args[@]}" \
+    "${optimizer_args[@]}" \
+    "${validation_args[@]}" \
+    "${miscellaneous_args[@]}"

examples/training/finetune/ltx2/preprocess_ltx2_data_t2v_new.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+
+GPU_NUM=4
+MODEL_PATH="FastVideo/LTX2-Distilled-Diffusers"
+DATASET_PATH="data/crush-smol"
+OUTPUT_DIR="$DATASET_PATH"
+WITH_AUDIO=true
+
+
+torchrun  --nproc_per_node=$GPU_NUM \
+    --master_port=29513 \
+    -m fastvideo.pipelines.preprocess.v1_preprocessing_new \
+    --model_path $MODEL_PATH \
+    --mode preprocess \
+    --workload_type t2v \
+    --preprocess.video_loader_type torchvision \
+    --preprocess.dataset_type merged \
+    --preprocess.dataset_path $DATASET_PATH \
+    --preprocess.dataset_output_dir $OUTPUT_DIR \
+    --preprocess.with_audio $WITH_AUDIO \
+    --preprocess.preprocess_video_batch_size 1 \
+    --preprocess.dataloader_num_workers 0 \
+    --preprocess.max_height 1088  \
+    --preprocess.max_width 1920 \
+    --preprocess.num_frames 121 \
+    --preprocess.train_fps 24 \
+    --preprocess.video_length_tolerance_range 5

examples/training/finetune/ltx2/validation.json

@@ -0,0 +1,16 @@
+{
+  "data": [
+    {
+      "caption": "A large metal cylinder is seen pressing down on a pile of Oreo cookies, flattening them as if they were under a hydraulic press."
+    },
+    {
+      "caption": "A large metal cylinder is seen compressing colorful clay into a compact shape, demonstrating the power of a hydraulic press."
+    },
+    {
+      "caption": "A large metal cylinder is seen pressing down on a pile of colorful candies, flattening them as if they were under a hydraulic press. The candies are crushed and broken into small pieces, creating a mess on the table."
+    },
+    {
+      "caption": "A massive steel piston descends onto a stack of chocolate chip cookies, crushing them into crumbs as though they are being compressed by a hydraulic press."
+    }
+  ]
+}

fastvideo-kernel/pyproject.toml

@@ -9,7 +9,7 @@ build-backend = "scikit_build_core.build"
 
 [project]
 name = "fastvideo-kernel"
-version = "0.2.5"
+version = "0.2.6"
 description = "Unified CUDA kernels for FastVideo"
 readme = "README.md"
 requires-python = ">=3.10"

fastvideo-kernel/python/fastvideo_kernel/block_sparse_attn.py

@@ -287,9 +287,8 @@ def block_sparse_attn(
     block_sparse_fwd, block_sparse_bwd = _get_sm90_ops()
     if (not _force_triton()) and _is_sm90() and (block_sparse_fwd is not None) and (block_sparse_bwd is not None):
         return block_sparse_attn_sm90(q, k, v, block_map, variable_block_sizes)
-    # Triton path: generally assumes q/k/v share the same padded length
-    if q.shape[2] != k.shape[2] or q.shape[2] != v.shape[2]:
-        raise RuntimeError("Triton fallback requires q/k/v to have the same padded length.")
+    # Triton path: supports q_seq_len != kv_seq_len as long as both are padded
+    # to a multiple of the block size (64 tokens).
     return block_sparse_attn_triton(q, k, v, block_map, variable_block_sizes)
 
 

fastvideo-kernel/python/fastvideo_kernel/ops.py

@@ -141,12 +141,6 @@ def video_sparse_attn(
         # Use autograd-enabled wrapper so backward works (and still uses SM90 kernel when available)
         out_s = block_sparse_attn(q, k, v, mask, variable_block_sizes)[0]
     else:
-        if q_seq_len != kv_seq_len:
-            raise RuntimeError(
-                "q/k have different lengths, but the compiled CUDA kernel (block_sparse_fwd) "
-                "is not available. The Triton fallback currently requires q and k/v to have "
-                "the same padded length."
-            )
         # Triton-only forward (kept for environments without the wrapper deps)
         out_s, _ = triton_block_sparse_attn_forward(q, k, v, idx, num, variable_block_sizes)
 

fastvideo-kernel/python/fastvideo_kernel/triton_kernels/block_sparse_attn_triton.py

@@ -29,7 +29,7 @@ configs = [
 
 
 # ──────────────────────────── SPARSE ADDITION BEGIN ───────────────────────────
-@triton.autotune(configs, key=["N_CTX", "HEAD_DIM"])
+@triton.autotune(configs, key=["N_CTX_Q", "HEAD_DIM"])
 @triton.jit
 def _attn_fwd_sparse(
         Q,
@@ -60,7 +60,8 @@ def _attn_fwd_sparse(
         stride_on,
         Z,
         H,
-        N_CTX,  #
+        N_CTX_Q,  #
+        N_CTX_KV,  #
         HEAD_DIM: tl.constexpr,  #
         BLOCK_M: tl.constexpr,
         BLOCK_N: tl.constexpr,
@@ -75,24 +76,29 @@ def _attn_fwd_sparse(
     off_hz = tl.program_id(1)  # fused (batch, head)
     b = off_hz // H
     h = off_hz % H
-    q_tiles = N_CTX // BLOCK_M
+    q_tiles = N_CTX_Q // BLOCK_M
     meta_base = ((b * H + h) * q_tiles + q_blk)
 
     kv_blocks = tl.load(q2k_num + meta_base)  # int32
     kv_ptr = q2k_index + meta_base * max_kv_blks  # ptr to list
 
     # ----- base pointers -----
-    qvk_off = (b.to(tl.int64) * stride_qz + h.to(tl.int64) * stride_qh)
+    # Note: when q and kv have different sequence lengths, their per-(batch,head)
+    # strides differ, so we must compute separate base offsets.
+    q_off = (b.to(tl.int64) * stride_qz + h.to(tl.int64) * stride_qh)
+    k_off = (b.to(tl.int64) * stride_kz + h.to(tl.int64) * stride_kh)
+    v_off = (b.to(tl.int64) * stride_vz + h.to(tl.int64) * stride_vh)
+    o_off = (b.to(tl.int64) * stride_oz + h.to(tl.int64) * stride_oh)
 
-    Q_ptr = tl.make_block_ptr(base=Q + qvk_off,
-                              shape=(N_CTX, HEAD_DIM),
+    Q_ptr = tl.make_block_ptr(base=Q + q_off,
+                              shape=(N_CTX_Q, HEAD_DIM),
                               strides=(stride_qm, stride_qk),
                               offsets=(q_blk * BLOCK_M, 0),
                               block_shape=(BLOCK_M, HEAD_DIM),
                               order=(1, 0))
 
-    K_base = tl.make_block_ptr(base=K + qvk_off,
-                               shape=(HEAD_DIM, N_CTX),
+    K_base = tl.make_block_ptr(base=K + k_off,
+                               shape=(HEAD_DIM, N_CTX_KV),
                                strides=(stride_kk, stride_kn),
                                offsets=(0, 0),
                                block_shape=(HEAD_DIM, BLOCK_N),
@@ -100,15 +106,15 @@ def _attn_fwd_sparse(
 
     v_order: tl.constexpr = (0, 1) if V.dtype.element_ty == tl.float8e5 else (1,
                                                                               0)
-    V_base = tl.make_block_ptr(base=V + qvk_off,
-                               shape=(N_CTX, HEAD_DIM),
+    V_base = tl.make_block_ptr(base=V + v_off,
+                               shape=(N_CTX_KV, HEAD_DIM),
                                strides=(stride_vk, stride_vn),
                                offsets=(0, 0),
                                block_shape=(BLOCK_N, HEAD_DIM),
                                order=v_order)
 
-    O_ptr = tl.make_block_ptr(base=Out + qvk_off,
-                              shape=(N_CTX, HEAD_DIM),
+    O_ptr = tl.make_block_ptr(base=Out + o_off,
+                              shape=(N_CTX_Q, HEAD_DIM),
                               strides=(stride_om, stride_on),
                               offsets=(q_blk * BLOCK_M, 0),
                               block_shape=(BLOCK_M, HEAD_DIM),
@@ -150,7 +156,7 @@ def _attn_fwd_sparse(
     # ----- epilogue -----
     m_i += tl.math.log2(l_i)
     acc = acc / l_i[:, None]
-    tl.store(M + off_hz * N_CTX + offs_m, m_i)
+    tl.store(M + off_hz * N_CTX_Q + offs_m, m_i)
     tl.store(O_ptr, acc.to(Out.type.element_ty))
 
 
@@ -201,7 +207,7 @@ def _attn_bwd_dkdv(
         stride_tok,
         stride_d,  #
         H,
-        N_CTX,
+        N_CTX_KV,
         BLOCK_M1: tl.constexpr,  #
         BLOCK_N1: tl.constexpr,  #
         HEAD_DIM: tl.constexpr,  #
@@ -221,8 +227,8 @@ def _attn_bwd_dkdv(
     off_hz = tl.program_id(2)  # fused (batch, head)
     b = off_hz // H
     h = off_hz % H
-    q_tiles = N_CTX // BLOCK_N1
-    meta_base = ((b * H + h) * q_tiles + kv_blk)
+    kv_tiles = N_CTX_KV // BLOCK_N1
+    meta_base = ((b * H + h) * kv_tiles + kv_blk)
 
     q_blocks = tl.load(k2q_num + meta_base)  # int32
     q_ptr = k2q_index + meta_base * max_q_blks  # ptr to list
@@ -302,16 +308,21 @@ def _attn_bwd_dq(
 
     kv_blocks = tl.load(q2k_num + meta_base)  # int32
     kv_ptr = q2k_index + meta_base * max_kv_blks  # ptr to list
-    block_size = tl.load(variable_block_sizes + q_blk)
 
     for blk_idx in range(kv_blocks * 2):
-        block_sparse_offset = (tl.load(kv_ptr + blk_idx // 2).to(tl.int32) * 2 +
-                               blk_idx % 2) * step_n * stride_tok
+        kv_idx = tl.load(kv_ptr + blk_idx // 2).to(tl.int32)
+        # variable_block_sizes is defined per KV block (tile). Mask must therefore
+        # use kv_idx (not q_blk). Also, because we split each 64-token block into
+        # two 32-token halves, the mask must account for the half-block offset.
+        block_size = tl.load(variable_block_sizes + kv_idx).to(tl.int32)
+        half = (blk_idx % 2).to(tl.int32)
+        block_sparse_offset = (kv_idx * 2 + half) * step_n * stride_tok
         kT = tl.load(kT_ptrs + block_sparse_offset)
         vT = tl.load(vT_ptrs + block_sparse_offset)
         qk = tl.dot(q, kT)
         p = tl.math.exp2(qk - m)
-        mask = tl.arange(0, BLOCK_N2) < block_size.to(tl.int32)
+        offs_in_block = half * step_n + tl.arange(0, BLOCK_N2)
+        mask = offs_in_block < block_size
         p = tl.where(mask[None, :], p, 0.0)
         # Compute dP and dS.
         dp = tl.dot(do, vT).to(tl.float32)
@@ -467,19 +478,235 @@ def _attn_bwd(
     tl.store(dq_ptrs, dq)
 
 
+@triton.jit
+def _attn_bwd_dkdv_kernel(
+        Q,
+        K,
+        V,
+        sm_scale,  #
+        DO,  #
+        DK,
+        DV,  #
+        M,
+        D,
+        k2q_index,
+        k2q_num,
+        max_q_blks,
+        variable_block_sizes,
+        # shared token/dim strides (assumed contiguous along token and dim)
+        stride_tok,
+        stride_d,  #
+        # batch/head strides (may differ between Q and KV)
+        stride_qz,
+        stride_qh,
+        stride_kz,
+        stride_kh,
+        stride_vz,
+        stride_vh,
+        stride_doz,
+        stride_doh,
+        stride_dkz,
+        stride_dkh,
+        stride_dvz,
+        stride_dvh,
+        H,
+        N_CTX_Q,
+        N_CTX_KV,
+        BLOCK_M1: tl.constexpr,  #
+        BLOCK_N1: tl.constexpr,  #
+        HEAD_DIM: tl.constexpr):
+    """
+    Backward kernel that computes dK and dV for each KV block (64 tokens).
+    Grid:
+      pid0: kv_blk in [0, N_CTX_KV/BLOCK_N1)
+      pid2: fused (batch, head) in [0, B*H)
+    """
+    bhid = tl.program_id(2)
+    b = bhid // H
+    h = bhid % H
+    kv_blk = tl.program_id(0)
+
+    q_adj = (b.to(tl.int64) * stride_qz + h.to(tl.int64) * stride_qh)
+    kv_adj_k = (b.to(tl.int64) * stride_kz + h.to(tl.int64) * stride_kh)
+    kv_adj_v = (b.to(tl.int64) * stride_vz + h.to(tl.int64) * stride_vh)
+    do_adj = (b.to(tl.int64) * stride_doz + h.to(tl.int64) * stride_doh)
+    dk_adj = (b.to(tl.int64) * stride_dkz + h.to(tl.int64) * stride_dkh)
+    dv_adj = (b.to(tl.int64) * stride_dvz + h.to(tl.int64) * stride_dvh)
+
+    Q = Q + q_adj
+    K = K + kv_adj_k
+    V = V + kv_adj_v
+    DO = DO + do_adj
+    DK = DK + dk_adj
+    DV = DV + dv_adj
+
+    # M and D (delta) are always sized by Q length.
+    M = M + (bhid * N_CTX_Q).to(tl.int64)
+    D = D + (bhid * N_CTX_Q).to(tl.int64)
+
+    offs_k = tl.arange(0, HEAD_DIM)
+    start_n = kv_blk * BLOCK_N1
+    offs_n = start_n + tl.arange(0, BLOCK_N1)
+
+    # load K and V: they stay in SRAM throughout the inner loop.
+    k = tl.load(K + offs_n[:, None] * stride_tok + offs_k[None, :] * stride_d)
+    v = tl.load(V + offs_n[:, None] * stride_tok + offs_k[None, :] * stride_d)
+
+    dv_acc = tl.zeros([BLOCK_N1, HEAD_DIM], dtype=tl.float32)
+    dk_acc = tl.zeros([BLOCK_N1, HEAD_DIM], dtype=tl.float32)
+
+    num_steps = N_CTX_Q // BLOCK_M1
+    dk_acc, dv_acc = _attn_bwd_dkdv(
+        dk_acc,
+        dv_acc,
+        Q,
+        k,
+        v,
+        sm_scale,
+        DO,
+        M,
+        D,
+        k2q_index,
+        k2q_num,
+        max_q_blks,
+        variable_block_sizes,
+        stride_tok,
+        stride_d,
+        H,
+        N_CTX_KV,
+        BLOCK_M1=BLOCK_M1,
+        BLOCK_N1=BLOCK_N1,
+        HEAD_DIM=HEAD_DIM,
+        start_n=start_n,
+        start_m=0,
+        num_steps=num_steps,
+    )
+
+    dv_ptrs = DV + offs_n[:, None] * stride_tok + offs_k[None, :] * stride_d
+    tl.store(dv_ptrs, dv_acc)
+
+    dk_acc *= sm_scale
+    dk_ptrs = DK + offs_n[:, None] * stride_tok + offs_k[None, :] * stride_d
+    tl.store(dk_ptrs, dk_acc)
+
+
+@triton.jit
+def _attn_bwd_dq_kernel(
+        Q,
+        K,
+        V,
+        DO,  #
+        DQ,
+        M,
+        D,
+        q2k_index,
+        q2k_num,
+        max_kv_blks,
+        variable_block_sizes,
+        # shared token/dim strides (assumed contiguous along token and dim)
+        stride_tok,
+        stride_d,  #
+        # batch/head strides (may differ between Q and KV)
+        stride_qz,
+        stride_qh,
+        stride_kz,
+        stride_kh,
+        stride_vz,
+        stride_vh,
+        stride_doz,
+        stride_doh,
+        stride_dqz,
+        stride_dqh,
+        H,
+        N_CTX_Q,
+        BLOCK_M2: tl.constexpr,  #
+        BLOCK_N2: tl.constexpr,  #
+        HEAD_DIM: tl.constexpr):
+    """
+    Backward kernel that computes dQ for each Q block (64 tokens).
+    Grid:
+      pid0: q_blk in [0, N_CTX_Q/BLOCK_M2)
+      pid2: fused (batch, head) in [0, B*H)
+    """
+    LN2 = 0.6931471824645996  # = ln(2)
+    bhid = tl.program_id(2)
+    b = bhid // H
+    h = bhid % H
+    q_blk = tl.program_id(0)
+
+    q_adj = (b.to(tl.int64) * stride_qz + h.to(tl.int64) * stride_qh)
+    kv_adj_k = (b.to(tl.int64) * stride_kz + h.to(tl.int64) * stride_kh)
+    kv_adj_v = (b.to(tl.int64) * stride_vz + h.to(tl.int64) * stride_vh)
+    do_adj = (b.to(tl.int64) * stride_doz + h.to(tl.int64) * stride_doh)
+    dq_adj = (b.to(tl.int64) * stride_dqz + h.to(tl.int64) * stride_dqh)
+
+    Q = Q + q_adj
+    K = K + kv_adj_k
+    V = V + kv_adj_v
+    DO = DO + do_adj
+    DQ = DQ + dq_adj
+
+    M = M + (bhid * N_CTX_Q).to(tl.int64)
+    D = D + (bhid * N_CTX_Q).to(tl.int64)
+
+    offs_k = tl.arange(0, HEAD_DIM)
+    start_m = q_blk * BLOCK_M2
+    offs_m = start_m + tl.arange(0, BLOCK_M2)
+
+    q = tl.load(Q + offs_m[:, None] * stride_tok + offs_k[None, :] * stride_d)
+    do = tl.load(DO + offs_m[:, None] * stride_tok + offs_k[None, :] * stride_d)
+    m = tl.load(M + offs_m)[:, None]
+
+    dq_acc = tl.zeros([BLOCK_M2, HEAD_DIM], dtype=tl.float32)
+    num_steps = 0  # unused in _attn_bwd_dq
+    dq_acc = _attn_bwd_dq(
+        dq_acc,
+        q,
+        K,
+        V,
+        do,
+        m,
+        D,
+        q2k_index,
+        q2k_num,
+        max_kv_blks,
+        variable_block_sizes,
+        stride_tok,
+        stride_d,
+        H,
+        N_CTX_Q,
+        BLOCK_M2=BLOCK_M2,
+        BLOCK_N2=BLOCK_N2,
+        HEAD_DIM=HEAD_DIM,
+        start_m=start_m,
+        start_n=0,
+        num_steps=num_steps,
+    )
+
+    dq_ptrs = DQ + offs_m[:, None] * stride_tok + offs_k[None, :] * stride_d
+    dq_acc *= LN2
+    tl.store(dq_ptrs, dq_acc)
+
+
 # ──────────────────────────── SPARSE ADDITION BEGIN ───────────────────────────
 def triton_block_sparse_attn_forward(q, k, v, q2k_index, q2k_num,
                                      variable_block_sizes):
-    B, H, T, D = q.shape
+    B, H, Tq, D = q.shape
+    Tkv = k.shape[2]
     sm_scale = 1.0 / math.sqrt(D)
     max_kv_blks = q2k_index.shape[-1]
-    assert T % 64 == 0, f"T must be a multiple of 64, but got {T}"
+    assert Tq % 64 == 0, f"q length must be a multiple of 64, but got {Tq}"
+    assert Tkv % 64 == 0, f"kv length must be a multiple of 64, but got {Tkv}"
     assert q2k_num.shape[
-        -1] == T // 64, f"shape mismatch, T // 64 = {T // 64}, q2k_num.shape[-2] = {q2k_num.shape[-2]}"
+        -1] == Tq // 64, f"shape mismatch, Tq // 64 = {Tq // 64}, q2k_num.shape[-2] = {q2k_num.shape[-2]}"
+    assert variable_block_sizes.numel() == Tkv // 64, (
+        f"shape mismatch, variable_block_sizes must have length {Tkv // 64}, "
+        f"got {variable_block_sizes.numel()}"
+    )
     o = torch.empty_like(q)
-    M = torch.empty((B, H, T), dtype=torch.float32, device=q.device)
+    M = torch.empty((B, H, Tq), dtype=torch.float32, device=q.device)
 
-    grid = lambda _: (triton.cdiv(T, 64), B * H, 1)
+    grid = lambda _: (triton.cdiv(Tq, 64), B * H, 1)
     _attn_fwd_sparse[grid](q,
                            k,
                            v,
@@ -508,7 +735,8 @@ def triton_block_sparse_attn_forward(q, k, v, q2k_index, q2k_num,
                            o.stride(3),
                            B,
                            H,
-                           T,
+                           Tq,
+                           Tkv,
                            HEAD_DIM=D,
                            STAGE=3)
 
@@ -518,21 +746,21 @@ def triton_block_sparse_attn_forward(q, k, v, q2k_index, q2k_num,
 def triton_block_sparse_attn_backward(do, q, k, v, o, M, q2k_index, q2k_num,
                                       k2q_index, k2q_num, variable_block_sizes):
     assert do.is_contiguous()
-    assert q.stride() == k.stride() == v.stride() == o.stride() == do.stride()
 
-    B, H, T, D = q.shape
+    B, H, Tq, D = q.shape
+    Tkv = k.shape[2]
     sm_scale = 1.0 / math.sqrt(D)
     dq = torch.empty_like(q)
     dk = torch.empty_like(k)
     dv = torch.empty_like(v)
-    BATCH, N_HEAD, N_CTX = q.shape[:3]
+    BATCH, N_HEAD = q.shape[:2]
     BLOCK_M1, BLOCK_N1, BLOCK_M2, BLOCK_N2 = 32, 64, 64, 32
     RCP_LN2 = 1.4426950408889634  # = 1.0 / ln(2)
     arg_k = k
     arg_k = arg_k * (sm_scale * RCP_LN2)
     PRE_BLOCK = 64
-    assert N_CTX % PRE_BLOCK == 0
-    pre_grid = (N_CTX // PRE_BLOCK, BATCH * N_HEAD)
+    assert Tq % PRE_BLOCK == 0
+    pre_grid = (Tq // PRE_BLOCK, BATCH * N_HEAD)
     delta = torch.empty_like(M)
     _attn_bwd_preprocess[pre_grid](
         o,
@@ -540,7 +768,7 @@ def triton_block_sparse_attn_backward(do, q, k, v, o, M, q2k_index, q2k_num,
         delta,  #
         BATCH,
         N_HEAD,
-        N_CTX,  #
+        Tq,  #
         BLOCK_M=PRE_BLOCK,
         HEAD_DIM=D  #
     )
@@ -548,36 +776,75 @@ def triton_block_sparse_attn_backward(do, q, k, v, o, M, q2k_index, q2k_num,
     max_q_blks = k2q_index.shape[-1]
     max_kv_blks = q2k_index.shape[-1]
 
-    grid = (N_CTX // BLOCK_N1, 1, BATCH * N_HEAD)
-    _attn_bwd[grid](
+    # dK/dV kernel: grid over KV blocks
+    grid_kv = (Tkv // BLOCK_N1, 1, BATCH * N_HEAD)
+    _attn_bwd_dkdv_kernel[grid_kv](
         q,
         arg_k,
         v,
         sm_scale,
         do,
-        dq,
         dk,
-        dv,  #
+        dv,
         M,
-        delta,  #
-        q2k_index,
-        q2k_num,
-        max_kv_blks,
+        delta,
         k2q_index,
         k2q_num,
         max_q_blks,
         variable_block_sizes,
+        q.stride(2),
+        q.stride(3),
         q.stride(0),
         q.stride(1),
-        q.stride(2),
-        q.stride(3),  #
+        arg_k.stride(0),
+        arg_k.stride(1),
+        v.stride(0),
+        v.stride(1),
+        do.stride(0),
+        do.stride(1),
+        dk.stride(0),
+        dk.stride(1),
+        dv.stride(0),
+        dv.stride(1),
         N_HEAD,
-        N_CTX,  #
+        Tq,
+        Tkv,
         BLOCK_M1=BLOCK_M1,
-        BLOCK_N1=BLOCK_N1,  #
+        BLOCK_N1=BLOCK_N1,
+        HEAD_DIM=D,
+    )
+
+    # dQ kernel: grid over Q blocks
+    grid_q = (Tq // BLOCK_M2, 1, BATCH * N_HEAD)
+    _attn_bwd_dq_kernel[grid_q](
+        q,
+        arg_k,
+        v,
+        do,
+        dq,
+        M,
+        delta,
+        q2k_index,
+        q2k_num,
+        max_kv_blks,
+        variable_block_sizes,
+        q.stride(2),
+        q.stride(3),
+        q.stride(0),
+        q.stride(1),
+        arg_k.stride(0),
+        arg_k.stride(1),
+        v.stride(0),
+        v.stride(1),
+        do.stride(0),
+        do.stride(1),
+        dq.stride(0),
+        dq.stride(1),
+        N_HEAD,
+        Tq,
         BLOCK_M2=BLOCK_M2,
-        BLOCK_N2=BLOCK_N2,  #
-        HEAD_DIM=D  #
+        BLOCK_N2=BLOCK_N2,
+        HEAD_DIM=D,
     )
 
     return dq, dk, dv

fastvideo-kernel/python/fastvideo_kernel/version.py

@@ -1 +1 @@
-__version__ = "0.2.5"
+__version__ = "0.2.6"

fastvideo/attention/backends/STA_configuration.py

@@ -350,7 +350,7 @@ def select_best_mask_strategy(
 
 
 def save_mask_search_results(
-        mask_search_final_result: list[dict[str, list[float]]],
+        mask_search_final_result: list[Any],
         prompt: str,
         mask_strategies: list[str],
         output_dir: str = 'output/mask_search_result/') -> str | None:
@@ -358,8 +358,9 @@ def save_mask_search_results(
         print("No mask search results to save")
         return None
 
-    # Create result dictionary with defaultdict for nested lists
-    mask_search_dict: dict[str, dict[str, list[list[float]]]] = {
+    # Create result dictionary with nested lists:
+    # [timesteps][layers][heads].
+    mask_search_dict: dict[str, dict[str, list[list[list[float]]]]] = {
         "L2_loss": defaultdict(list),
         "L1_loss": defaultdict(list)
     }
@@ -371,23 +372,79 @@ def save_mask_search_results(
         masks_list = [int(x) for x in mask.split(',')]
         selected_masks.append(masks_list)
 
+    def _to_float_list(loss_values: Any, loss_name: str) -> list[float]:
+        if isinstance(loss_values, np.ndarray):
+            loss_values = loss_values.tolist()
+        if not isinstance(loss_values, list | tuple):
+            raise ValueError(
+                f"{loss_name} must be a sequence of numeric values")
+
+        float_values = []
+        for loss in loss_values:
+            try:
+                float_values.append(float(loss))
+            except (TypeError, ValueError) as exc:
+                raise ValueError(
+                    f"Invalid {loss_name} value {loss!r}: expected a number"
+                ) from exc
+        return float_values
+
+    def _extract_timestep_layer_losses(step_data: Any, loss_name: str,
+                                       strategy_idx: int) -> list[list[float]]:
+        if isinstance(step_data, dict):
+            layer_data_list = [step_data]
+        elif isinstance(step_data, list):
+            layer_data_list = step_data
+        else:
+            return []
+
+        timestep_layer_losses: list[list[float]] = []
+        for layer_data in layer_data_list:
+            if not isinstance(layer_data, dict) or loss_name not in layer_data:
+                continue
+
+            raw_losses = layer_data[loss_name]
+            if isinstance(raw_losses, np.ndarray):
+                raw_losses = raw_losses.tolist()
+            if not isinstance(raw_losses, list | tuple):
+                raise ValueError(f"{loss_name} must be a list or tuple")
+            raw_losses = list(raw_losses)
+
+            if raw_losses and isinstance(raw_losses[0], list | tuple
+                                         | np.ndarray):
+                if strategy_idx >= len(raw_losses):
+                    raise ValueError(
+                        f"Missing strategy index {strategy_idx} in {loss_name}")
+                strategy_losses = raw_losses[strategy_idx]
+            else:
+                if strategy_idx > 0:
+                    continue
+                strategy_losses = raw_losses
+
+            timestep_layer_losses.append(
+                _to_float_list(strategy_losses, loss_name))
+
+        return timestep_layer_losses
+
     # Process each mask strategy
     for i, mask_strategy in enumerate(selected_masks):
         mask_strategy_str = str(mask_strategy)
-        # Process L2 loss
-        step_results: list[list[float]] = []
-        for step_data in mask_search_final_result:
-            if isinstance(step_data, dict) and "L2_loss" in step_data:
-                layer_losses = [float(loss) for loss in step_data["L2_loss"]]
-                step_results.append(layer_losses)
-        mask_search_dict["L2_loss"][mask_strategy_str] = step_results
+        l2_step_results: list[list[list[float]]] = []
+        l1_step_results: list[list[list[float]]] = []
 
-        step_results = []
         for step_data in mask_search_final_result:
-            if isinstance(step_data, dict) and "L1_loss" in step_data:
-                layer_losses = [float(loss) for loss in step_data["L1_loss"]]
-                step_results.append(layer_losses)
-        mask_search_dict["L1_loss"][mask_strategy_str] = step_results
+            l2_layer_losses = _extract_timestep_layer_losses(
+                step_data, "L2_loss", i)
+            if l2_layer_losses:
+                l2_step_results.append(l2_layer_losses)
+
+            l1_layer_losses = _extract_timestep_layer_losses(
+                step_data, "L1_loss", i)
+            if l1_layer_losses:
+                l1_step_results.append(l1_layer_losses)
+
+        mask_search_dict["L2_loss"][mask_strategy_str] = l2_step_results
+        mask_search_dict["L1_loss"][mask_strategy_str] = l1_step_results
 
     # Create the output directory if it doesn't exist
     os.makedirs(output_dir, exist_ok=True)

fastvideo/attention/backends/flash_attn.py

@@ -92,11 +92,59 @@ class FlashAttentionImpl(AttentionImpl):
         value: torch.Tensor,
         attn_metadata: FlashAttnMetadata,
     ):
+
+        def _key_padding_mask_from_attn_mask(attn_mask: torch.Tensor,
+                                             key_len: int) -> torch.Tensor:
+            # Normalize attn_mask to [B, key_len] where True means valid token.
+            if attn_mask.dim() == 4:
+                attn_mask = attn_mask[:, 0, 0, :]
+            elif attn_mask.dim() == 3:
+                attn_mask = attn_mask[:, 0, :]
+            elif attn_mask.dim() != 2:
+                raise ValueError(
+                    f"Unsupported attn_mask shape for FLASH_ATTN: {attn_mask.shape}"
+                )
+
+            if attn_mask.dtype == torch.bool:
+                key_padding_mask = attn_mask
+            else:
+                # SDPA additive mask convention: valid=0, masked=-inf/large negative.
+                key_padding_mask = attn_mask >= 0
+
+            if key_padding_mask.shape[-1] != key_len:
+                raise ValueError(
+                    "Invalid key padding mask length for FLASH_ATTN: "
+                    f"expected {key_len}, got {key_padding_mask.shape[-1]}")
+            return key_padding_mask
+
         if attn_metadata is not None and hasattr(
                 attn_metadata,
                 "attn_mask") and attn_metadata.attn_mask is not None:
-            from fastvideo.attention.utils.flash_attn_no_pad import flash_attn_no_pad
+            from fastvideo.attention.utils.flash_attn_no_pad import (
+                flash_attn_no_pad, flash_attn_varlen_qk_no_pad)
             attn_mask = attn_metadata.attn_mask
+
+            # flash_attn_no_pad packs q/k/v as one tensor and assumes equal q/k
+            # sequence lengths. Cross-attention can violate this.
+            if query.shape[1] != key.shape[1]:
+                query_padding_mask = torch.ones(
+                    (query.shape[0], query.shape[1]),
+                    dtype=torch.bool,
+                    device=query.device,
+                )
+                key_padding_mask = _key_padding_mask_from_attn_mask(
+                    attn_mask, key.shape[1]).to(device=key.device)
+                return flash_attn_varlen_qk_no_pad(
+                    query,
+                    key,
+                    value,
+                    query_padding_mask=query_padding_mask,
+                    key_padding_mask=key_padding_mask,
+                    causal=self.causal,
+                    dropout_p=0.0,
+                    softmax_scale=self.softmax_scale,
+                )
+
             qkv = torch.stack([query, key, value], dim=2)
 
             attn_mask = F.pad(attn_mask, (qkv.shape[1] - attn_mask.shape[1], 0),

fastvideo/attention/utils/flash_attn_no_pad.py

@@ -97,3 +97,60 @@ def flash_attn_no_pad_v3(qkv,
                        "b s (h d) -> b s h d",
                        h=nheads)
     return output
+
+
+def flash_attn_varlen_qk_no_pad(
+    query,
+    key,
+    value,
+    query_padding_mask,
+    key_padding_mask,
+    causal=False,
+    dropout_p=0.0,
+    softmax_scale=None,
+    deterministic=False,
+):
+    from flash_attn.bert_padding import pad_input, unpad_input
+
+    try:
+        from flash_attn_interface import flash_attn_varlen_func as flash_attn_varlen_func_impl
+    except ImportError:
+        from flash_attn import flash_attn_varlen_func as flash_attn_varlen_func_impl
+
+    if flash_attn_varlen_func_impl is None:
+        raise ImportError("FlashAttention varlen backend not available")
+
+    batch_size, q_seqlen, nheads, _ = query.shape
+
+    query_unpad, q_indices, cu_seqlens_q, max_seqlen_q, _ = unpad_input(
+        rearrange(query, "b s h d -> b s (h d)"), query_padding_mask)
+    key_unpad, _, cu_seqlens_k, max_seqlen_k, _ = unpad_input(
+        rearrange(key, "b s h d -> b s (h d)"), key_padding_mask)
+    value_unpad, _, _, _, _ = unpad_input(
+        rearrange(value, "b s h d -> b s (h d)"), key_padding_mask)
+
+    query_unpad = rearrange(query_unpad, "nnz (h d) -> nnz h d", h=nheads)
+    key_unpad = rearrange(key_unpad, "nnz (h d) -> nnz h d", h=nheads)
+    value_unpad = rearrange(value_unpad, "nnz (h d) -> nnz h d", h=nheads)
+
+    output_unpad = flash_attn_varlen_func_impl(
+        query_unpad,
+        key_unpad,
+        value_unpad,
+        cu_seqlens_q,
+        cu_seqlens_k,
+        max_seqlen_q,
+        max_seqlen_k,
+        dropout_p=dropout_p,
+        softmax_scale=softmax_scale,
+        causal=causal,
+        deterministic=deterministic,
+    )
+
+    output = rearrange(
+        pad_input(rearrange(output_unpad, "nnz h d -> nnz (h d)"), q_indices,
+                  batch_size, q_seqlen),
+        "b s (h d) -> b s h d",
+        h=nheads,
+    )
+    return output

fastvideo/configs/configs.py

@@ -86,6 +86,7 @@ class PreprocessConfig:
 
     # Model configuration
     training_cfg_rate: float = 0.0
+    with_audio: bool = False
 
     # framework configuration
     seed: int = 42
@@ -190,6 +191,10 @@ class PreprocessConfig:
                                      type=float,
                                      default=PreprocessConfig.training_cfg_rate,
                                      help="Training CFG rate")
+        preprocess_args.add_argument(f"--{prefix_with_dot}with-audio",
+                                     action=StoreBoolean,
+                                     default=PreprocessConfig.with_audio,
+                                     help="Whether to extract and encode audio")
         preprocess_args.add_argument(f"--{prefix_with_dot}seed",
                                      type=int,
                                      default=PreprocessConfig.seed,