[skill] Add add-model and review-add-model-pr skills

Two skills split out of the will/stable-audio PR (#1260) so the model
work and the agent infra evolve on independent review cycles.

`add-model/SKILL.md` (~1600 lines) — the canonical procedure for
porting a new diffusion model into FastVideo, distilled from the Wan,
LTX-2, Hunyuan, Cosmos, Stable Audio, and MagiHuman ports. Covers the
17-row Files-table, 16-step procedure, FastVideo layer/attention
selection rules, parity-test pattern, weight conversion recipe, and
the I2V variant pattern.

`add-model/REVIEW.md` (~1900 lines) — 36 documented failure modes
discovered while porting / re-reviewing those families. Each item is
a self-contained "what we did wrong, what to do instead" entry with a
suggested action on the skill text. Items 1-22 from the Wan / MagiHuman
porting passes; items 23-36 from the Stable Audio port (audio as
first-class workload, hard ban on `from diffusers import` for model
classes at runtime, user-story example docstrings, single-class
kwargs-driven variants, examples accept paths not user-written decode
glue, etc.).

`add-model/add_model_split.md` — plan for breaking the monolithic
1600-line SKILL.md into ~12 satellite docs (deferred until after
the next porting pass validates which sections are most-referenced).

`review-add-model-pr/SKILL.md` (~480 lines, NEW) — reviewer-facing
dual of `add-model`. Walks a PR reviewer through the canonical
surface (per-component checklist for DiT / VAE / encoder /
PipelineConfig / pipeline class / stages / presets+registry / tests /
example), indexes the 36 REVIEW failure modes by where they show up
in a diff (the "Pitfall map"), and produces a structured verdict
(block / nit / follow-up) with a copy-pasteable comment skeleton.
Key design choice: the reviewer's job is NOT to redesign the port,
just to verify the porter followed the canonical procedure honestly
and exercised the parity gate — REVIEW item 16 ("skip-on-missing
parity is a silent no-op trap") gets disproportionate weight because
it's the single most expensive failure mode in past ports.

Both skills registered in `.agents/skills/index.jsonl` as
`status: draft, trust: low` per existing convention.

.agents/STATUS.md

@@ -0,0 +1,94 @@
+# Agent Infrastructure — Status Dashboard
+
+Developer-maintained overview of all agent components and their maturity.
+Use this to understand what exists, how complete it is, and how much to trust it.
+
+_Last synced: 2026-03-02_
+
+> To resync this dashboard, use the workflow: `.agents/workflows/sync-dashboard.md`
+
+---
+
+## Summary
+
+| Category | Total | ✅ Ready | 🟡 Draft | 🔴 Stub | Trust |
+|----------|-------|---------|---------|---------|-------|
+| Skills | 8 | 0 | 8 | 0 | Low — newly created, untested |
+| Workflows (SOPs) | 4 | 0 | 4 | 0 | Low — newly created, untested |
+| Memory files | 4 | 1 | 3 | 0 | Medium — codebase_map is solid |
+| Lessons | 0 | — | — | — | N/A — empty |
+| Exploration logs | 0 | — | — | — | N/A — empty |
+
+---
+
+## Skills (`.agents/skills/`)
+
+| Skill | File | Status | Trust | Tested | Notes |
+|-------|------|--------|-------|--------|-------|
+| Launch Experiment | `launch-experiment.md` | 🟡 Draft | Low | ❌ | Needs dry-run validation |
+| Monitor Experiment | `monitor-experiment.md` | 🟡 Draft | Low | ❌ | Requires W&B API access to test |
+| Summarize Run | `summarize-run.md` | 🟡 Draft | Low | ❌ | Pattern from existing test infra |
+| Log Experiment | `log-experiment.md` | 🟡 Draft | Low | ❌ | Journal formatting only |
+| Evaluate Video Quality | `evaluate-video-quality.md` | 🟡 Draft | Low | ❌ | SSIM section most mature |
+| Index Related Work | `index-related-work.md` | 🟡 Draft | Low | ❌ | Schema defined, no entries yet |
+| Search Related Work | `search-related-work.md` | 🟡 Draft | Low | ❌ | Depends on indexed entries |
+| Skill Template | `SKILL_TEMPLATE.md` | ✅ Ready | High | ✅ | Meta-template, stable |
+
+### Trust Level Definitions
+- **High**: Tested in production, validated against real experiments
+- **Medium**: Logic is sound, partially tested or based on existing patterns
+- **Low**: Newly written, not yet validated
+- **None**: Placeholder only
+
+---
+
+## Workflows / SOPs (`.agents/workflows/`)
+
+| Workflow | File | Status | Trust | Tested | Notes |
+|----------|------|--------|-------|--------|-------|
+| Experiment Lifecycle | `experiment-lifecycle.md` | 🟡 Draft | Low | ❌ | End-to-end flow, untested |
+| Evaluation Development | `evaluation-development.md` | 🟡 Draft | Low | ❌ | Metric dev process |
+| Experiment Journaling | `experiment-journaling.md` | 🟡 Draft | Low | ❌ | Journaling cadence |
+| Lesson Capture | `lesson-capture.md` | 🟡 Draft | Low | ❌ | Post-experiment reflection |
+| Sync Dashboard | `sync-dashboard.md` | 🟡 Draft | Low | ❌ | This dashboard's updater |
+
+---
+
+## Memory (`.agents/memory/`)
+
+| File | Status | Trust | Notes |
+|------|--------|-------|-------|
+| `codebase_map.md` | ✅ Ready | High | Synthesized from full repo research |
+| `experiment_journal.md` | 🟡 Draft | Medium | Schema defined, no entries yet |
+| `evaluation_registry.md` | 🟡 Draft | Medium | SSIM/loss metrics documented |
+| `related_work/README.md` | 🟡 Draft | Medium | Schema defined, no entries yet |
+
+---
+
+## Lessons (`.agents/lessons/`)
+
+| File | Category | Severity | Notes |
+|------|----------|----------|-------|
+
+_No lessons captured yet._
+
+---
+
+## Exploration Logs (`.agents/exploration/`)
+
+| File | Status | Topic | Notes |
+|------|--------|-------|-------|
+
+_No exploration logs yet._
+
+---
+
+## What to Do Next
+
+1. **Validate skills**: Run a minimal training experiment using the
+   `experiment-lifecycle` SOP to test `launch-experiment` → `monitor-experiment`
+   → `summarize-run` end-to-end.
+2. **Index first related work**: Use `index-related-work` to add at least one
+   paper (e.g., the Self-Forcing paper used in the codebase).
+3. **Capture first lesson**: After the validation run, capture any findings.
+4. **Promote to Ready**: As each skill/SOP is tested, update its status here.

.agents/exploration/README.md

@@ -0,0 +1,46 @@
+# Exploration Logs
+
+This directory holds draft procedures and investigation notes for tasks that
+don't yet have a standardized skill or SOP. Each exploration should follow this
+template.
+
+## When to Create an Exploration Log
+
+- You are working on a task with no existing skill or workflow.
+- You are experimenting with a new metric, training technique, or tool.
+- You want to document findings before they are promoted to a standard.
+
+## File Naming
+
+`<topic-slug>.md` — e.g., `fvd-metric-investigation.md`
+
+## Template
+
+```markdown
+# Exploration Log: <Topic>
+
+## Status: draft | under_review | promoted | abandoned
+
+## Context
+<Why this exploration is needed — link to experiment or task if applicable.>
+
+## Progress
+- [ ] Step 1: ...
+- [ ] Step 2: ...
+
+## Findings
+<What you have learned so far.>
+
+## Mistakes / Dead Ends
+<What didn't work and why — these become lessons.>
+
+## Proposed Standardization
+<If this works, describe the skill/SOP/workflow to create.>
+```
+
+## Lifecycle
+
+1. **Create** during exploration mode.
+2. **Update** as you make progress.
+3. **Promote**: If findings are solid, create a skill in `.agents/skills/` or an SOP in `.agents/workflows/`.
+4. **Archive mistakes**: Move failures into `.agents/lessons/`.

.agents/lessons/README.md

@@ -0,0 +1,48 @@
+# Lessons Learned Database
+
+This directory stores documented mistakes, unexpected behaviors, and their fixes.
+Each lesson is a permanent record that helps agents and humans avoid repeating
+past errors.
+
+## When to Create a Lesson
+
+- An experiment failed for a non-obvious reason.
+- A configuration or hyperparameter choice led to wasted compute.
+- A porting, data, or infrastructure issue was discovered and resolved.
+- A workaround was needed for a known framework/library bug.
+
+## File Naming
+
+`<YYYY-MM-DD>_<short-slug>.md` — e.g., `2026-03-02_lr-too-high-for-lora.md`
+
+## Template
+
+```markdown
+---
+date: <ISO-8601>
+experiment: <reference to experiment_journal.md entry, if applicable>
+category: hyperparameter | data | infrastructure | evaluation | porting | other
+severity: critical | important | minor
+---
+
+# <Short Descriptive Title>
+
+## What Happened
+<Description of the problem and its symptoms.>
+
+## Root Cause
+<Analysis of why it happened.>
+
+## Fix / Workaround
+<What resolved the issue.>
+
+## Prevention
+<How to avoid this in the future — updated skills, SOPs, or checks.>
+```
+
+## Usage
+
+- Before starting a task, **search this directory** for relevant lessons.
+- After completing or failing a task, **check if a new lesson should be created**.
+- Periodically review lessons for **patterns** — recurring themes may warrant
+  a new skill, SOP, or codebase fix.

.agents/memory/codebase-map/README.md

@@ -0,0 +1,129 @@
+# FastVideo-WorldModel — Codebase Map
+
+High-level structural index for agent orientation. Updated 2026-03-08.
+
+## Repository Layout
+
+```
+FastVideo-WorldModel/
+├── fastvideo/                 # Core Python package
+│   ├── models/                # Model implementations
+│   │   ├── dits/              #   DiT transformers (wanvideo, ltx2, ...)
+│   │   ├── vaes/              #   VAE models
+│   │   ├── encoders/          #   Text/image encoders (T5, CLIP)
+│   │   ├── schedulers/        #   Noise schedulers
+│   │   ├── upsamplers/        #   Super-resolution models
+│   │   ├── audio/             #   Audio models
+│   │   └── loader/            #   Component loaders for HF repos
+│   ├── configs/               # Configuration system
+│   │   ├── models/            #   Arch configs + param_names_mapping
+│   │   ├── pipelines/         #   Pipeline wiring
+│   │   └── sample/            #   Default sampling parameters
+│   ├── pipelines/             # End-to-end pipelines
+│   │   ├── basic/             #   Per-model pipelines (wan/, ltx2/, ...)
+│   │   └── stages/            #   Reusable pipeline stages
+│   ├── train/                 # Refactored training framework (YAML-driven, preferred)
+│   │   ├── trainer.py         #   Main training loop coordinator
+│   │   ├── entrypoint/        #   Training entrypoint (train.py) + checkpoint conversion
+│   │   ├── methods/           #   Training algorithms (FineTune, DFSFT, DMD2, SelfForcing)
+│   │   │   ├── base.py        #     TrainingMethod ABC
+│   │   │   ├── fine_tuning/   #     FineTuneMethod, DiffusionForcingSFTMethod
+│   │   │   └── distribution_matching/  # DMD2Method, SelfForcingMethod
+│   │   ├── models/            #   Per-role model wrappers (ModelBase, CausalModelBase)
+│   │   │   └── wan/           #     WanModel, WanCausalModel
+│   │   ├── callbacks/         #   Composable hooks (grad_clip, ema, validation)
+│   │   └── utils/             #   Config, builder, checkpoint, optimizer, tracking
+│   ├── training/              # Legacy training infrastructure (being phased out)
+│   │   ├── trackers.py        #   W&B tracker (BaseTracker → WandbTracker)
+│   │   ├── training_utils.py  #   Checkpointing, grad clipping, state dicts
+│   │   ├── training_pipeline.py        # Base training pipeline
+│   │   ├── wan_training_pipeline.py    # Wan T2V training
+│   │   ├── wan_i2v_training_pipeline.py # Wan I2V training
+│   │   ├── distillation_pipeline.py    # Distillation base
+│   │   ├── wan_distillation_pipeline.py # Wan distillation
+│   │   ├── self_forcing_distillation_pipeline.py # Self-forcing distill
+│   │   ├── ltx2_training_pipeline.py   # LTX-2 training
+│   │   └── matrixgame_training_pipeline.py # MatrixGame training
+│   ├── attention/             # Attention backends
+│   ├── distributed/           # Sequence/tensor parallel utilities
+│   ├── layers/                # Tensor-parallel layers
+│   ├── tests/                 # Package-level tests
+│   │   ├── training/          #   Training regression tests (W&B summary comparison)
+│   │   ├── ssim/              #   SSIM visual regression tests
+│   │   ├── encoders/          #   Encoder parity tests
+│   │   └── modal/             #   Modal CI test runner
+│   └── registry.py            # Unified config registry
+├── fastvideo-kernel/          # CUDA/custom kernels (separate build: ./build.sh)
+├── scripts/                   # Utility scripts
+│   ├── distill/               #   Distillation launch scripts
+│   ├── inference/             #   Inference scripts
+│   ├── checkpoint_conversion/ #   Weight conversion tools
+│   ├── finetune/              #   Finetune scripts
+│   └── preprocess/            #   Data preprocessing
+├── examples/                  # Ready-to-run examples
+│   ├── training/              #   Training examples (finetune/, consistency_finetune/)
+│   ├── distill/               #   Distillation examples
+│   ├── inference/             #   Inference examples
+│   └── dataset/               #   Dataset examples
+├── docs/                      # MkDocs documentation source
+│   ├── design/overview.md     #   Architecture overview
+│   ├── training/              #   Training guides
+│   └── contributing/          #   Contributor guides + coding_agents.md
+├── tests/                     # Top-level tests (local_tests/)
+├── AGENTS.md                  # Agent coding guidelines
+└── .agents/                    # Agent infrastructure (you are here)
+```
+
+## Key Training Entrypoints
+
+### New framework (`fastvideo/train/`) — preferred
+
+| Method | Config Example | Launch Pattern |
+|--------|---------------|----------------|
+| FineTune (Wan) | `examples/train/finetune_wan2.1_t2v_1.3B_vsa_*.yaml` | `torchrun -m fastvideo.train.entrypoint.train --config <yaml>` |
+| DFSFT (Wan causal) | `examples/train/dfsft_wan_causal_t2v_1.3B.yaml` | `torchrun -m fastvideo.train.entrypoint.train --config <yaml>` |
+| DMD2 distillation | `examples/train/distill_wan2.1_t2v_1.3B_dmd2.yaml` | `torchrun -m fastvideo.train.entrypoint.train --config <yaml>` |
+| Self-Forcing | `examples/train/self_forcing_wan_causal_t2v_1.3B.yaml` | `torchrun -m fastvideo.train.entrypoint.train --config <yaml>` |
+
+### Legacy pipelines (`fastvideo/training/`) — being phased out
+
+| Pipeline | Entrypoint | Launch Pattern |
+|----------|-----------|----------------|
+| Wan T2V finetune | `fastvideo/training/wan_training_pipeline.py` | `torchrun --nproc_per_node N` |
+| Wan I2V finetune | `fastvideo/training/wan_i2v_training_pipeline.py` | `torchrun --nproc_per_node N` |
+| Wan distillation (DMD) | `fastvideo/training/wan_distillation_pipeline.py` | `torchrun --nproc_per_node N` |
+| Self-forcing distill | `fastvideo/training/wan_self_forcing_distillation_pipeline.py` | `torchrun --nproc_per_node N` |
+| LTX-2 finetune | `fastvideo/training/ltx2_training_pipeline.py` | `torchrun --nproc_per_node N` |
+| MatrixGame | `fastvideo/training/matrixgame_training_pipeline.py` | `torchrun --nproc_per_node N` |
+
+## W&B Integration
+
+- **Tracker classes**: `fastvideo/training/trackers.py`
+  - `WandbTracker` — logs metrics, videos, timing
+  - `SequentialTracker` — fan-out to multiple trackers
+  - `DummyTracker` — no-op for offline/test
+- **Run summary location**: `<output_dir>/tracker/wandb/latest-run/files/wandb-summary.json`
+- **Reference summaries**: `fastvideo/tests/training/*/` (e.g., `a40_reference_wandb_summary.json`)
+- **Environment**: `WANDB_API_KEY`, `WANDB_BASE_URL`, `WANDB_MODE`
+
+## Critical Environment Variables
+
+| Variable | Purpose |
+|----------|---------|
+| `WANDB_API_KEY` | W&B authentication |
+| `WANDB_MODE` | `online` / `offline` |
+| `FASTVIDEO_ATTENTION_BACKEND` | `FLASH_ATTN` / `TORCH_SDPA` |
+| `TOKENIZERS_PARALLELISM` | Set `false` to avoid fork warnings |
+| `HF_HOME` | HuggingFace cache directory |
+
+## Build & Test Commands
+
+```bash
+uv pip install -e .[dev]                          # Editable install
+pre-commit run --all-files                        # Lint/format/spell
+pytest tests/                                     # Top-level tests
+pytest fastvideo/tests/ -v                        # Package tests
+pytest fastvideo/tests/training/Vanilla -srP      # Training loss regression
+pytest fastvideo/tests/ssim/ -vs                  # SSIM visual regression
+cd fastvideo-kernel && ./build.sh                 # Build kernels
+```

.agents/memory/evaluation-registry/README.md

@@ -0,0 +1,327 @@
+# Evaluation Metrics Registry
+
+Living catalog of all evaluation metrics for FastVideo-WorldModel video quality
+assessment. Each metric includes a detailed explanation, implementation status,
+usage instructions, and interpretation guide.
+
+_Last updated: 2026-03-02_
+
+---
+
+## Metric Summary
+
+| Metric | Category | Status | Location | Trust |
+|--------|----------|--------|----------|-------|
+| **FVD** | Distribution | ✅ Implemented | `benchmarks/fvd/` | High |
+| **SSIM** | Reference | ✅ Implemented | `fastvideo/tests/ssim/` | High |
+| **LPIPS** | Perceptual | ✅ Implemented | `scripts/lora_extraction/` | Medium |
+| **Loss trajectory** | Training signal | ✅ Implemented | W&B `train_loss` | Medium |
+| **Grad norm stability** | Training signal | ✅ Implemented | W&B `grad_norm` | Medium |
+| **GameWorld Score** | Multi-dim benchmark | 🟡 External | Matrix-Game repo | Low |
+| **Human preference** | Gold standard | 🔴 Manual | N/A | Highest |
+
+---
+
+## Implemented Metrics
+
+### FVD — Fréchet Video Distance
+
+**Category**: Distribution-level quality metric
+**Status**: ✅ Fully implemented in `benchmarks/fvd/`
+**Trust**: High — standard protocol, I3D feature extractor
+
+#### What It Measures
+FVD measures the distance between the **distribution** of generated videos and
+a distribution of real/reference videos. It works by:
+1. Extracting spatiotemporal features from both real and generated video sets
+   using a pretrained **I3D** (Inflated 3D ConvNet) model.
+2. Modeling each set of features as a multivariate Gaussian (mean + covariance).
+3. Computing the **Fréchet distance** between the two Gaussians.
+
+Lower FVD = generated videos are more statistically similar to real videos.
+
+#### Why It Matters
+- FVD is the **de facto standard** for benchmarking video generation models.
+- It captures both **visual quality** (are individual frames realistic?) and
+  **temporal coherence** (do frames flow naturally?).
+- Matrix-Game 2.0, Open-Sora, and most video generation papers report FVD.
+
+#### Limitations
+- Requires a **large sample set** (standard protocol uses 2048 videos) to
+  produce stable statistics. Small sample sizes yield noisy results.
+- Measures **distributional similarity**, not per-video quality. A model could
+  have low FVD by generating a diverse set of "roughly okay" videos.
+- The I3D model was trained on Kinetics-400 (human actions). It may be less
+  sensitive to domain-specific artifacts in non-human-action videos (e.g.,
+  driving, game environments).
+- Does not directly measure text-video alignment or action controllability.
+
+#### How to Use
+
+```python
+# Programmatic
+from benchmarks.fvd import compute_fvd_with_config, FVDConfig
+
+config = FVDConfig.fvd2048_16f()  # Standard: 2048 videos, 16 frames
+results = compute_fvd_with_config('data/real/', 'outputs/gen/', config)
+print(f"FVD: {results['fvd']:.2f}")
+```
+
+```bash
+# CLI
+python -m benchmarks.fvd.cli \
+    --real-path data/real/ \
+    --gen-path outputs/gen/ \
+    --protocol fvd2048_16f
+```
+
+**Preset protocols**:
+| Protocol | Videos | Frames | Use Case |
+|----------|--------|--------|----------|
+| `fvd2048_16f` | 2048 | 16 | Standard benchmark (papers) |
+| `fvd2048_128f` | 2048 | 128 | Long video evaluation |
+| `quick_test` | 100 | 16 | Fast dev iteration |
+
+**Feature extractors**: `i3d` (default, standard), `clip`, `videomae`
+
+#### Interpretation
+| FVD Range | Interpretation |
+|-----------|---------------|
+| < 100 | Excellent — near-real quality |
+| 100–300 | Good — competitive with SOTA |
+| 300–600 | Fair — noticeable gap from real |
+| > 600 | Poor — significant quality issues |
+
+> FVD values are dataset-dependent. Always compare against baselines evaluated
+> on the same real video distribution.
+
+---
+
+### SSIM — Structural Similarity Index
+
+**Category**: Per-frame reference comparison
+**Status**: ✅ Implemented in `fastvideo/tests/ssim/`
+**Trust**: High — used in CI regression tests
+
+#### What It Measures
+SSIM compares two images (or video frames) based on three components:
+1. **Luminance**: brightness similarity
+2. **Contrast**: dynamic range similarity
+3. **Structure**: spatial pattern similarity
+
+The final score is a value in [0, 1] where 1.0 = identical.
+
+#### Why It Matters
+- Used as a **regression guard** in CI: ensures model updates don't degrade
+  visual output below a threshold.
+- More perceptually meaningful than raw pixel MSE.
+- Fast to compute — suitable for automated testing.
+
+#### Limitations
+- Requires a **pixel-aligned reference** video. Cannot compare videos with
+  different seeds, prompts, or angles.
+- Operates **per-frame** — does not capture temporal coherence.
+- Insensitive to some perceptual artifacts (color shifts, high-frequency noise).
+
+#### How to Use
+
+```bash
+pytest fastvideo/tests/ssim/ -vs
+```
+
+#### Interpretation
+| SSIM Range | Quality |
+|------------|---------|
+| > 0.90 | Excellent — very close to reference |
+| 0.80–0.90 | Good — acceptable for most uses |
+| 0.70–0.80 | Fair — noticeable differences |
+| < 0.70 | Poor — significant divergence |
+
+---
+
+### LPIPS — Learned Perceptual Image Patch Similarity
+
+**Category**: Per-frame perceptual distance
+**Status**: ✅ Implemented in `scripts/lora_extraction/lora_inference_comparison.py`
+**Trust**: Medium — available but only used for LoRA comparison currently
+
+#### What It Measures
+LPIPS uses a pretrained neural network (AlexNet by default) to extract
+deep features from two images and computes the distance between them in
+feature space. Unlike SSIM, LPIPS correlates much more strongly with
+**human perceptual judgments**.
+
+Lower LPIPS = more perceptually similar.
+
+#### Why It Matters
+- Best available automated proxy for **human visual judgments** at the frame
+  level.
+- Captures semantic and structural differences that SSIM misses (e.g., texture
+  changes, minor recoloring).
+- Used for validating LoRA merge quality.
+
+#### Limitations
+- Per-frame metric — no temporal awareness.
+- Requires reference video (paired comparison only).
+- Slightly slower than SSIM due to neural network forward pass.
+
+#### How to Use
+
+```bash
+python scripts/lora_extraction/lora_inference_comparison.py \
+  --base merged_model \
+  --ft path/to/finetuned \
+  --adapter NONE \
+  --output-dir results \
+  --prompt "A cat" \
+  --compute-lpips
+```
+
+#### Interpretation
+| LPIPS Range | Quality |
+|-------------|---------|
+| < 0.10 | Excellent — nearly indistinguishable |
+| 0.10–0.20 | Good — minor perceptual differences |
+| 0.20–0.40 | Fair — noticeable differences |
+| > 0.40 | Poor — clearly different |
+
+---
+
+### Loss Trajectory
+
+**Category**: Training signal proxy
+**Status**: ✅ Active (from W&B `train_loss`)
+**Trust**: Medium — proxy, not direct quality measure
+
+#### What It Measures
+Tracks the training loss over time. A healthy training run shows:
+- **Decreasing loss** over the first hundreds of steps.
+- **Stable gradient norms** (no wild spikes).
+- **Consistent step times** (no infrastructure issues).
+
+#### Why It Matters
+- Cheapest evaluation signal — available in real-time from W&B.
+- Critical for the **30-minute quality check** workflow.
+- At later training stages (when loss becomes meaningful), trajectory shape
+  can predict final model quality.
+
+#### Context: How This Evolves
+The team's experience shows evaluation signals change during a project:
+- **Early stage**: Loss may be flat or meaningless → focus on SSIM & visual
+  inspection instead.
+- **Mid stage**: Loss starts decreasing → trajectory shape becomes useful.
+- **Late stage**: Loss is meaningful → can compare trajectories across runs.
+
+This dynamic is a key insight from the team's workflow: don't over-rely on
+loss early; don't ignore it late.
+
+---
+
+### Grad Norm Stability
+
+**Category**: Training health diagnostic
+**Status**: ✅ Active (from W&B `grad_norm`)
+**Trust**: Medium — diagnostic, not quality metric
+
+#### What It Measures
+The magnitude of gradients during training. Stable grad norms indicate
+healthy optimization. Spikes or NaN values indicate training instability.
+
+#### Alert Thresholds
+| Condition | Meaning |
+|-----------|---------|
+| Stable ~0.3–0.5 | Normal training |
+| Single spike > 3× average | Possible bad batch, monitor |
+| NaN or Inf | 🔴 Training has diverged — stop run |
+| Increasing trend | Learning rate may be too high |
+
+---
+
+## External Benchmarks
+
+### GameWorld Score Benchmark (Matrix-Game)
+
+**Category**: Multi-dimensional evaluation framework for interactive world models
+**Status**: 🟡 External — not implemented in-repo
+**Source**: [Matrix-Game 1.0 benchmark](https://github.com/SkyworkAI/Matrix-Game), used in [Matrix-Game 2.0 paper](https://arxiv.org/abs/2508.13009)
+
+#### What It Measures
+A comprehensive benchmark examining **four critical capabilities**:
+
+| Dimension | What It Evaluates | Example Signals |
+|-----------|-------------------|-----------------|
+| **Visual quality** | Frame-level realism, absence of artifacts | Color fidelity, sharpness, coherence |
+| **Temporal quality** | Smoothness across frames, motion consistency | Jitter, flickering, temporal aliasing |
+| **Action controllability** | Response to input actions (keyboard/mouse) | Action delay, correctness, smoothness |
+| **Physical rule understanding** | Adherence to physics (gravity, collision) | Object persistence, plausible motion |
+
+#### Context from Matrix-Game 2.0
+- Evaluation uses **597-frame composite action sequences** over 32 Minecraft
+  scenes and 16 wild scenes.
+- Action controllability assessment is **Minecraft-specific** — cannot be
+  directly applied to wild/general scenes.
+- The paper notes that models that "collapse" to static frames can
+  paradoxically score higher on consistency metrics — beware of this confound.
+
+#### Relevance to FastVideo
+- Matrix-Game 2.0 is built on SkyReels-V2/Wan2.1 architecture — **same model
+  family as FastVideo**.
+- Their distillation uses DMD-based Self-Forcing — **same technique** as our
+  `self_forcing_distillation_pipeline.py`.
+- GameWorld Score dimensions are a useful framework for thinking about world
+  model quality even outside gaming contexts.
+
+---
+
+## Human Preference Evaluation
+
+**Category**: Gold-standard quality assessment
+**Status**: 🔴 Manual process — no automated implementation
+**Priority**: **Highest** — this is the most important evaluation signal
+**Trust**: Highest — but expensive
+
+### What It Measures
+Human evaluators compare generated videos and rate them on dimensions like:
+- Overall quality and realism
+- Temporal coherence and smoothness
+- Prompt adherence / action correctness
+- Absence of artifacts
+
+#### Why It's the Most Important Metric
+All automated metrics are **proxies** for human judgment. They can be gamed
+or may miss artifacts that humans easily notice. Human preference is the
+ultimate ground truth for video generation quality.
+
+#### Cost & Practicality
+| Approach | Cost | Scale | When to Use |
+|----------|------|-------|-------------|
+| Internal team review | Low | ~10–50 videos | Every major checkpoint |
+| Crowdsource (MTurk, Scale) | Medium | 100+ videos | Pre-release validation |
+| A/B preference test | Medium | Pairs | Comparing two model versions |
+
+#### Recommended Protocol
+1. Sample 10–20 videos from the model at a checkpoint.
+2. Include diverse prompts (easy + hard, short + long).
+3. Have 2–3 evaluators score each video 1–5 on: quality, coherence, fidelity.
+4. Record scores in the experiment journal.
+
+---
+
+## Metrics NOT Used
+
+| Metric | Reason |
+|--------|--------|
+| ~~CLIP-Score~~ | Not used by the team. Measures text-image alignment using CLIP embeddings, but not well-suited for video temporal quality. |
+| Inception Score (IS) | Less informative than FVD for video; primarily an image metric. |
+| PSNR | Pixel-level metric; less perceptually meaningful than SSIM/LPIPS. |
+
+---
+
+## Adding a New Metric
+
+Follow the SOP: `.agents/workflows/evaluation-development.md`
+
+1. Prototype in `.agents/exploration/`
+2. Validate on known-good and known-bad samples
+3. Add to this registry
+4. Update the `evaluate-video-quality` skill

.agents/memory/experiment-journal/README.md

@@ -0,0 +1,21 @@
+# Experiment Journal
+
+Living log of all experiments. Each entry captures what was tried, the result,
+and any insights. Newest entries go at the top.
+
+_No experiments logged yet. Use the `log-experiment` skill to add entries._
+
+<!-- TEMPLATE — copy and fill for each new experiment:
+
+## [YYYY-MM-DD] Experiment: <name>
+- **Hypothesis**: <what you expected to learn>
+- **Config**: model=..., lr=..., sp_size=..., gpus=..., script=...
+- **W&B run**: <run_id or URL>
+- **Duration**: <total wall time>
+- **Key metrics**: loss=..., step_time=..., grad_norm=...
+- **Checkpoint**: <path>
+- **Insight**: <what was learned>
+- **Status**: running | completed | failed | abandoned
+- **Related lessons**: `.agents/lessons/<filename>.md`
+
+-->

.agents/memory/index.jsonl

@@ -0,0 +1,4 @@
+{"name": "codebase-map", "description": "High-level structural index of the FastVideo-WorldModel repository", "path": "codebase-map/README.md", "status": "ready", "trust": "high"}
+{"name": "evaluation-registry", "description": "Catalog of all evaluation metrics with detailed explanations, implementation status, and usage guides", "path": "evaluation-registry/README.md", "status": "draft", "trust": "medium"}
+{"name": "experiment-journal", "description": "Living log of all experiments with hypotheses, configs, metrics, and insights", "path": "experiment-journal/README.md", "status": "draft", "trust": "medium"}
+{"name": "related-work", "description": "Index of related papers, repos, and blog posts with structured comparisons to FastVideo", "path": "related-work/README.md", "status": "draft", "trust": "low"}

.agents/memory/related-work/README.md

@@ -0,0 +1,34 @@
+# Related Work Index
+
+Each file in this directory is a structured summary of a related paper, repo,
+or blog post relevant to FastVideo-WorldModel training.
+
+## File Format
+
+Each file is named `<slug>.md` and follows this structure:
+
+```markdown
+---
+title: <paper/repo title>
+source: <URL or citation>
+type: paper | repo | blog
+date_indexed: <ISO-8601>
+tags: [world-model, distillation, evaluation, reward-shaping, ...]
+---
+
+## Summary
+<1-2 paragraph summary of the work.>
+
+## Key Differences from FastVideo
+- <Bullet points comparing their approach to ours.>
+
+## Actionable Insights
+- <What we could adopt or adapt.>
+```
+
+## How to Add New Entries
+
+Use the `index-related-work` skill, or manually create a file following the
+template above.
+
+_No related work indexed yet._

.agents/onboarding/README.md

@@ -0,0 +1,76 @@
+# Agent Onboarding — FastVideo-WorldModel
+
+Welcome, agent. This is the **master onboarding** guide. Follow the steps below,
+then check if a **domain-specific onboarding** exists for your task.
+
+## Domain-Specific Onboarding
+
+If your task falls into one of these areas, read the specialized guide **after**
+completing the general steps below:
+
+| Domain | Guide | When to Use |
+|--------|-------|-------------|
+| **WorldModel Training** | `worldmodel-training/README.md` | Training, finetuning, distillation, experiment management |
+
+---
+
+## Step 1: Understand the Codebase
+
+Read these files to build your context:
+
+| Priority | File | What you learn |
+|----------|------|----------------|
+| 1 | `AGENTS.md` | Coding guidelines, build/test commands, PR conventions |
+| 2 | `docs/design/overview.md` | Architecture: models, pipelines, configs, registry |
+| 3 | `fastvideo/train/` | Refactored training framework (YAML-driven, modular methods/models/callbacks) |
+| 4 | `docs/training/overview.md` | Training data flow and preprocessing |
+| 5 | `docs/training/finetune.md` | Training arguments, parallelism, LoRA, validation |
+| 6 | `docs/contributing/coding_agents.md` | How to add model pipelines with agent assistance |
+
+## Step 2: Discover Available Resources
+
+Read these two index files to see what skills and memory modules exist:
+
+- **`.agents/skills/index.jsonl`** — catalog of all agent skills (name + description)
+- **`.agents/memory/index.jsonl`** — catalog of all memory modules (name + description)
+
+Each entry has a `path` field pointing to the full content. Only load the
+full README.md for modules relevant to your current task.
+
+## Step 3: Check for Existing Skills & SOPs
+
+Before writing new code or procedures:
+
+1. **Skills**: Read `.agents/skills/index.jsonl` — find a matching skill by description.
+2. **Workflows/SOPs**: Browse `.agents/workflows/` — step-by-step procedures for common tasks.
+3. **Lessons**: Browse `.agents/lessons/` — known pitfalls and their fixes.
+
+If a skill or SOP exists for your task, **use it**. If not, you are in **exploration mode** — see Step 4.
+
+## Step 4: Exploration Mode
+
+If no existing skill/SOP covers your task:
+
+1. Document your progress in `.agents/exploration/<topic>.md` using the template in `.agents/exploration/README.md`.
+2. At the end of your session, reflect:
+   - **What worked** → propose a new skill or SOP in the exploration log.
+   - **What failed** → create a lesson in `.agents/lessons/`.
+3. Flag the exploration log for human review.
+
+## Quick Reference
+
+```
+.agents/
+├── ONBOARDING.md          ← you are here
+├── STATUS.md              ← dashboard: completeness & trust of all components
+├── skills/                ← reusable agent skills
+├── workflows/             ← SOPs and procedures
+├── memory/                ← persistent context (folder per topic + index.jsonl)
+│   ├── index.jsonl
+│   ├── codebase-map/
+│   ├── experiment-journal/
+│   ├── evaluation-registry/
+│   └── related-work/
+├── lessons/               ← mistakes and fixes
+└── exploration/           ← draft procedures
+```

.agents/onboarding/worldmodel-training/README.md

@@ -0,0 +1,302 @@
+# WorldModel Training — Agent Onboarding
+
+Specialized onboarding for agents working on FastVideo-WorldModel training,
+distillation, and evaluation. Read the master onboarding (`.agents/onboarding/README.md`)
+first, then come here.
+
+---
+
+## Domain Context
+
+FastVideo-WorldModel trains **interactive world models** — video generation systems
+that respond to user actions (keyboard/mouse) in real-time. The architecture is
+based on **Wan2.1** (SkyReels-V2) DiT models with causal attention for
+auto-regressive streaming generation.
+
+**Key techniques you will work with:**
+- Full finetuning and LoRA on Wan / LTX-2 / MatrixGame models
+- DMD-based distillation (few-step generation)
+- Self-Forcing distillation (causal streaming)
+- Diffusion-Forcing SFT (DFSFT) for causal models
+- VSA (Variable Sparsity Acceleration) for efficient training
+
+---
+
+## Training Code: Two Generations
+
+### New modular framework: `fastvideo/train/` (preferred)
+
+The refactored training code uses a **YAML-only config-driven** architecture
+with composable methods, per-role models, and a callback system. All new
+training work should use this framework.
+
+### Legacy pipelines: `fastvideo/training/` (deprecated)
+
+The old monolithic pipeline classes (`WanTrainingPipeline`,
+`DistillationPipeline`, etc.) still exist but are being phased out. The new
+framework imports select utilities from `fastvideo/training/` for backward
+compatibility (EMA, gradient clipping, checkpoint wrappers).
+
+---
+
+## Essential Reading (Training-Specific)
+
+Read these **in order** before touching any training code:
+
+| # | File | What You Learn |
+|---|------|----------------|
+| 1 | `docs/training/overview.md` | Training data flow: raw video → text embeddings + video latents → training |
+| 2 | `docs/training/finetune.md` | Training arguments, parallelism (SP/TP), LoRA, validation settings |
+| 3 | `docs/training/data_preprocess.md` | How to preprocess datasets into the expected format |
+| 4 | `docs/design/overview.md` | Architecture: models, pipelines, configs, registry |
+
+---
+
+## New Training Framework (`fastvideo/train/`)
+
+### Architecture Overview
+
+```
+fastvideo/train/
+├── __init__.py                    → exports Trainer
+├── trainer.py                     → main training loop coordinator
+├── entrypoint/
+│   ├── train.py                   → YAML-only training entrypoint
+│   └── dcp_to_diffusers.py        → checkpoint conversion utility
+├── methods/                       → training algorithms (TrainingMethod ABC)
+│   ├── base.py                    → TrainingMethod base class
+│   ├── fine_tuning/
+│   │   ├── finetune.py            → FineTuneMethod (supervised finetuning)
+│   │   └── dfsft.py               → DiffusionForcingSFTMethod (causal)
+│   ├── distribution_matching/
+│   │   ├── dmd2.py                → DMD2Method (distribution matching distill)
+│   │   └── self_forcing.py        → SelfForcingMethod (causal streaming)
+│   ├── knowledge_distillation/    → (stub, not yet implemented)
+│   └── consistency_model/         → (stub, not yet implemented)
+├── models/                        → per-role model instances
+│   ├── base.py                    → ModelBase & CausalModelBase (ABC)
+│   └── wan/
+│       ├── wan.py                 → WanModel (non-causal)
+│       └── wan_causal.py          → WanCausalModel (causal streaming)
+├── callbacks/                     → training hooks & monitoring
+│   ├── callback.py                → Callback base class + CallbackDict
+│   ├── grad_clip.py               → GradNormClipCallback
+│   ├── ema.py                     → EMACallback (shadow weights)
+│   └── validation.py              → ValidationCallback (sampling + eval)
+└── utils/                         → configuration, building, checkpointing
+    ├── builder.py                 → build_from_config() (config → runtime)
+    ├── checkpoint.py              → CheckpointManager (DCP-based)
+    ├── config.py                  → load_run_config() (YAML → RunConfig)
+    ├── training_config.py         → TypedConfig dataclasses
+    ├── optimizer.py               → build_optimizer_and_scheduler()
+    ├── instantiate.py             → resolve_target() + instantiate()
+    ├── tracking.py                → build_tracker() (W&B, etc.)
+    ├── dataloader.py              → dataloader utilities
+    ├── module_state.py            → apply_trainable()
+    └── moduleloader.py            → load_module_from_path()
+```
+
+### Key Concepts
+
+**TrainingMethod** (`methods/base.py`): Abstract base class for all training
+algorithms. Owns role models (student, teacher, critic), manages checkpoint
+state, and defines the training step interface.
+
+**ModelBase** (`models/base.py`): Per-role model wrapper. Each role (student,
+teacher, critic) gets its own `ModelBase` instance owning a `transformer` and
+`noise_scheduler`. `CausalModelBase` extends this for streaming models.
+
+**Callback system** (`callbacks/`): Composable hooks for gradient clipping,
+EMA, validation, etc. Configured via YAML, dispatched by `CallbackDict`.
+
+**Config system** (`utils/config.py`, `utils/training_config.py`): YAML files
+are parsed into typed `RunConfig` dataclass trees. Models and methods use
+`_target_` fields for instantiation (similar to Hydra).
+
+### Training Flow
+
+```
+run_training_from_config(config_path)
+  → load_run_config()           # YAML → RunConfig
+  → init_distributed()          # TP/SP setup
+  → build_from_config()         # instantiate models, method, dataloader
+  → Trainer.run()               # main loop:
+      ├─ callbacks.on_train_start()
+      ├─ checkpoint_manager.maybe_resume()
+      ├─ for step in range(max_steps):
+      │    ├─ method.single_train_step(batch)
+      │    ├─ method.backward()
+      │    ├─ callbacks.on_before_optimizer_step()
+      │    ├─ method.optimizers_schedulers_step()
+      │    ├─ tracker.log(metrics, step)
+      │    ├─ callbacks.on_training_step_end()
+      │    └─ checkpoint_manager.maybe_save(step)
+      ├─ callbacks.on_train_end()
+      └─ checkpoint_manager.save_final()
+```
+
+### Training Methods
+
+| Method | Class | Use Case |
+|--------|-------|----------|
+| **FineTune** | `FineTuneMethod` | Single-role supervised finetuning |
+| **DFSFT** | `DiffusionForcingSFTMethod` | Diffusion-forcing SFT with inhomogeneous timesteps |
+| **DMD2** | `DMD2Method` | Multi-role distribution matching distillation (student + teacher + critic) |
+| **Self-Forcing** | `SelfForcingMethod` | Extends DMD2 for causal student rollouts |
+
+### Launching Training (New Framework)
+
+Training is launched via `torchrun` with a single YAML config:
+
+```bash
+torchrun --nproc_per_node <N_GPUS> \
+  -m fastvideo.train.entrypoint.train \
+  --config examples/train/<config>.yaml
+```
+
+### Example YAML Configs
+
+| Config | Method | Description |
+|--------|--------|-------------|
+| `examples/train/finetune_wan2.1_t2v_1.3B_vsa_phase3.4_0.9sparsity.yaml` | FineTune | Wan 1.3B finetuning with VSA sparsity |
+| `examples/train/distill_wan2.1_t2v_1.3B_dmd2.yaml` | DMD2 | Wan 1.3B distillation (student + teacher + critic) |
+| `examples/train/dfsft_wan_causal_t2v_1.3B.yaml` | DFSFT | Causal Wan 1.3B diffusion-forcing SFT |
+| `examples/train/self_forcing_wan_causal_t2v_1.3B.yaml` | Self-Forcing | Causal streaming distillation |
+
+### Checkpointing (New Framework)
+
+**CheckpointManager** (`utils/checkpoint.py`) saves via `torch.distributed.checkpoint`:
+
+```
+output_dir/
+└─ checkpoint-{step}/
+   ├─ dcp/                    # DCP state dict
+   ├─ config.json             # resolved training config
+   └─ .fastvideo_metadata.json
+```
+
+Checkpoint state includes: role model weights, per-role optimizers/schedulers,
+CUDA RNG state, and callback state (e.g., EMA shadow weights).
+
+### Config Structure
+
+A YAML config defines the full training pipeline:
+
+```yaml
+models:
+  student:
+    _target_: fastvideo.train.models.wan.WanModel
+    model_path: ...
+    trainable: true
+  teacher:  # optional, for distillation
+    _target_: fastvideo.train.models.wan.WanModel
+    model_path: ...
+    trainable: false
+
+method:
+  _target_: fastvideo.train.methods.fine_tuning.FineTuneMethod
+  # method-specific params...
+
+training:
+  distributed: { num_gpus: 8, tp_size: 1, sp_size: 8 }
+  data: { data_path: ..., batch_size: 1 }
+  optimizer: { lr: 1e-5, lr_scheduler: constant_with_warmup }
+  loop: { max_train_steps: 1000 }
+  checkpoint: { output_dir: ./outputs }
+  tracker: { trackers: [wandb], project_name: ... }
+
+callbacks:
+  grad_clip:
+    _target_: fastvideo.train.callbacks.GradNormClipCallback
+    max_grad_norm: 1.0
+  validation:
+    _target_: fastvideo.train.callbacks.ValidationCallback
+    validation_steps: 100
+```
+
+---
+
+## Legacy Training Pipelines (`fastvideo/training/`)
+
+> **Note:** Use the new `fastvideo/train/` framework for new work. This section
+> is retained for reference on existing pipelines not yet migrated.
+
+| Pipeline | Entrypoint | Use Case |
+|----------|-----------|----------|
+| Wan T2V finetune | `fastvideo/training/wan_training_pipeline.py` | Standard text-to-video finetune / LoRA |
+| Wan I2V finetune | `fastvideo/training/wan_i2v_training_pipeline.py` | Image-to-video (first frame conditioned) |
+| MatrixGame finetune | `fastvideo/training/matrixgame_training_pipeline.py` | Action-conditioned world model |
+| LTX-2 finetune | `fastvideo/training/ltx2_training_pipeline.py` | LTX-2 architecture finetuning |
+| Wan DMD distillation | `fastvideo/training/wan_distillation_pipeline.py` | Few-step distillation via DMD |
+| Self-Forcing distill | `fastvideo/training/wan_self_forcing_distillation_pipeline.py` | Causal streaming distillation |
+
+---
+
+## Key Infrastructure
+
+### W&B Integration
+- **Tracker**: `fastvideo/training/trackers.py` — `WandbTracker` class
+- **New framework tracker**: `fastvideo/train/utils/tracking.py` — `build_tracker()`
+- **Env vars**: `WANDB_API_KEY`, `WANDB_BASE_URL`, `WANDB_MODE`
+
+### Parallelism
+- **SP** (Sequence Parallel): splits video frames across GPUs — `sp_size: N`
+- **TP** (Tensor Parallel): splits model layers across GPUs — `tp_size: N`
+- Typical configs: SP=2–8, TP=1–2
+
+---
+
+## Evaluation (for training runs)
+
+Read `.agents/memory/evaluation-registry/README.md` for the full metric catalog.
+
+**Quick summary for training agents:**
+| Metric | When to Use | Trust |
+|--------|-------------|-------|
+| **Loss trajectory** | Every run, real-time from W&B | Medium |
+| **SSIM** | When comparing against reference outputs | High |
+| **FVD** | For benchmarking model quality (`benchmarks/fvd/`) | High |
+| **LPIPS** | LoRA merge validation | Medium |
+| **Human preference** | Major checkpoints | Highest |
+
+---
+
+## Common Workflows
+
+| Task | Skill / SOP |
+|------|-------------|
+| Launch a training run | `.agents/skills/launch-experiment/SKILL.md` |
+| Monitor a running experiment | `.agents/skills/monitor-experiment/SKILL.md` |
+| Summarize final results | `.agents/skills/summarize-run/SKILL.md` |
+| Full experiment lifecycle | `.agents/workflows/experiment-lifecycle.md` |
+| Capture lessons from failures | `.agents/workflows/lesson-capture.md` |
+
+---
+
+## World Model–Specific Concepts
+
+### Action Injection (MatrixGame)
+The MatrixGame pipeline adds **action modules** to each DiT block, enabling
+frame-level mouse/keyboard input conditioning. The action sequence is injected
+per-frame alongside the latent video tokens.
+
+### Causal Architecture
+For streaming generation, the model uses **causal attention** (each frame only
+attends to previous frames). This enables auto-regressive chunk-by-chunk
+generation — critical for real-time interactive world models.
+
+### Self-Forcing Distillation
+A **data-free** distillation method where the student model is trained to
+generate coherent video sequences by being forced to use its own previous
+outputs (rather than ground-truth) as context. This produces models robust to
+their own error accumulation during long auto-regressive generation.
+
+### DMD Distillation (Distribution Matching Distillation)
+Reduces inference steps from ~50 to 3–4 by training a student model to match
+the output distribution of the teacher model. Uses a critic network to estimate
+distribution divergence.
+
+### Diffusion-Forcing SFT (DFSFT)
+Supervised finetuning with **inhomogeneous timesteps** across chunks — each
+chunk in a causal sequence can have a different noise level, training the model
+to handle mixed-fidelity contexts.

.agents/scripts/sync-skills.sh

@@ -0,0 +1,96 @@
+#!/usr/bin/env bash
+# Sync .agents/skills/ into .claude/skills/ via per-skill symlinks.
+#
+# Why: Claude Code only scans .claude/skills/ and ~/.claude/skills/ for
+# user-invocable skills (no skillsPath config exists — see
+# https://code.claude.com/docs/en/skills.md). This repo's skills live
+# in .agents/skills/ so they travel with the repo and stay under git.
+# Run this once after cloning (or after adding/removing a skill) to
+# expose them to Claude Code without maintaining a parallel tree.
+#
+# Usage:
+#   .agents/scripts/sync-skills.sh
+#
+# Idempotent and safe to re-run. Prunes stale symlinks whose source
+# has been removed from .agents/skills/. Leaves hand-written
+# .claude/skills/<name>/ directories untouched (only symlinks are
+# managed).
+
+set -euo pipefail
+
+REPO_ROOT="$(git -C "$(dirname "$0")" rev-parse --show-toplevel)"
+SRC_DIR="$REPO_ROOT/.agents/skills"
+DST_DIR="$REPO_ROOT/.claude/skills"
+
+if [[ ! -d "$SRC_DIR" ]]; then
+    echo "Error: $SRC_DIR does not exist." >&2
+    exit 1
+fi
+
+mkdir -p "$DST_DIR"
+
+linked=0
+unchanged=0
+skipped=0
+pruned=0
+
+link_skill() {
+    local name="$1"
+    local src="$SRC_DIR/$name"
+    local dst="$DST_DIR/$name"
+    # Relative target keeps symlinks portable across clones.
+    local rel="../../.agents/skills/$name"
+
+    if [[ -L "$dst" ]]; then
+        if [[ "$(readlink "$dst")" == "$rel" ]]; then
+            unchanged=$((unchanged + 1))
+            return
+        fi
+        rm "$dst"
+    elif [[ -e "$dst" ]]; then
+        echo "Skipped (not a symlink): .claude/skills/$name" >&2
+        skipped=$((skipped + 1))
+        return
+    fi
+
+    ln -s "$rel" "$dst"
+    echo "Linked: .claude/skills/$name -> $rel"
+    linked=$((linked + 1))
+}
+
+prune_stale() {
+    local link="$1"
+    local target
+    target="$(readlink "$link")"
+    case "$target" in
+        ../../.agents/skills/*) ;;
+        *) return ;;
+    esac
+    local name="${target##*/}"
+    if [[ ! -d "$SRC_DIR/$name" ]]; then
+        rm "$link"
+        echo "Pruned stale: .claude/skills/$(basename "$link")"
+        pruned=$((pruned + 1))
+    fi
+}
+
+for src in "$SRC_DIR"/*/; do
+    [[ -d "$src" ]] || continue
+    name="$(basename "$src")"
+    # Only treat directories that actually contain a SKILL.md as skills.
+    [[ -f "$src/SKILL.md" ]] || continue
+    link_skill "$name"
+done
+
+shopt -s nullglob
+for link in "$DST_DIR"/*; do
+    [[ -L "$link" ]] || continue
+    prune_stale "$link"
+done
+shopt -u nullglob
+
+printf "\nSummary: %d linked, %d unchanged, %d pruned" "$linked" "$unchanged" "$pruned"
+if [[ "$skipped" -gt 0 ]]; then
+    printf ", %d skipped (non-symlink collision)" "$skipped"
+fi
+printf "\n"

.agents/skills/SKILL_TEMPLATE.md

@@ -0,0 +1,57 @@
+---
+name: <skill-name>
+description: <one-line description — Codex uses this for implicit invocation matching>
+---
+
+# <Skill Name>
+
+## Purpose
+<Why this skill exists and when to use it.>
+
+## Prerequisites
+- <What must be true before using this skill>
+
+## Inputs
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `param1` | Yes | ... |
+
+## Steps
+
+1. **Step 1 title**
+   - Detail...
+
+2. **Step 2 title**
+   - Detail...
+
+## Outputs
+- <What this skill produces>
+
+## Example Usage
+
+```
+<Example invocation or prompt snippet>
+```
+
+## References
+- <Links to relevant files in the codebase>
+
+---
+
+## Folder Structure
+
+Each skill lives in its own directory under `.agents/skills/`:
+
+```
+.agents/skills/<skill-name>/
+├── SKILL.md          # Required: instructions + metadata (this file)
+├── scripts/          # Optional: executable helper scripts
+├── references/       # Optional: documentation, papers
+└── assets/           # Optional: templates, resources
+```
+
+After creating a new skill, add an entry to `.agents/skills/index.jsonl`:
+
+```json
+{"name": "<skill-name>", "description": "<description>", "path": "<skill-name>/SKILL.md", "status": "draft", "trust": "low"}
+```

.agents/skills/add-model/add_model_split.md

@@ -0,0 +1,230 @@
+# add-model skill — split plan
+
+The current `SKILL.md` is **1605 lines** in a single file. This document
+proposes splitting it into ~12 satellite docs with a much shorter
+top-level index, following the idiomatic Anthropic-skills pattern of
+"short procedural index + targeted satellites."
+
+## Why split
+
+1. **Cold-load cost.** The whole 1605-line file is loaded into the
+   model's context every time the skill fires. Most porting sessions
+   only need a fraction of that — e.g. a VAE-only contribution doesn't
+   need the I2V-variant section or the weight-conversion recipe. A
+   shorter `SKILL.md` index + lazy-loaded satellites keeps the live
+   context lean.
+
+2. **Reviewability.** A single 1605-line markdown file is hard to
+   diff. Splits keep changes scoped — adding the "audio workload"
+   section (REVIEW item 25) becomes a new file in `how-to/` rather
+   than a 100-line insert in the middle of an existing megafile.
+
+3. **Findability.** Section anchors in a single file are hard to
+   navigate; per-topic files surface in `ls .agents/skills/add-model/`
+   and `grep -r` returns clean per-file matches.
+
+4. **Component-only contributions** (REVIEW item 23) become a
+   first-class workflow document instead of an awkward "but skip
+   half the steps" exception inside the main file.
+
+## Current section inventory
+
+| Section | Approx lines | Stays in SKILL.md? | Target file |
+|---|---|---|---|
+| Purpose | 8 | Yes | SKILL.md |
+| When to use / not to use | 25 | Yes | SKILL.md |
+| Prerequisites | 40 | Yes | SKILL.md |
+| Inputs (table) | 15 | Yes | SKILL.md |
+| FastVideo's single architecture | 40 | No | how-to/architecture.md |
+| Files you will create or touch (table) | 25 | Yes (link to how-to) | SKILL.md + how-to/files_table.md |
+| Steps (1–16) | 440 | **Index only** in SKILL.md (3-line per step + link) | how-to/steps_<phase>.md ×4 |
+| Standard stages — subclass targets | 20 | No | how-to/architecture.md |
+| FastVideo layers and attention | 110 | No | how-to/layers_and_attention.md |
+| Parity test pattern | 165 | No | how-to/parity_testing.md |
+| Parallel component porting | 100 | No | how-to/parity_testing.md |
+| Weight conversion | 155 | No | how-to/weight_conversion.md |
+| `register_configs` cheatsheet | 25 | No | how-to/registry_cheatsheet.md |
+| Adding an I2V variant | 270 | No | how-to/i2v_variant.md |
+| Distributed support | 35 | No | how-to/distributed_support.md |
+| Common pitfalls | 50 | No | reference/pitfalls.md |
+| Outputs | 15 | Yes | SKILL.md |
+| Example prompt snippet | 35 | No | reference/example_prompt.md |
+| References | 50 | No | reference/links.md |
+| Changelog | 20 | No | reference/changelog.md |
+
+## Target tree
+
+```
+.agents/skills/add-model/
+├── SKILL.md                              ~250 lines  — procedural index
+├── REVIEW.md                             unchanged
+├── add_model_split.md                    this doc
+├── how-to/
+│   ├── architecture.md                   ~70 lines   — single-architecture diagram + standard stages
+│   ├── files_table.md                    ~50 lines   — annotated 17-row Files table
+│   ├── steps_1_setup.md                  ~80 lines   — steps 1–5 (gather, study, reuse, convert, clone)
+│   ├── steps_2_components.md             ~120 lines  — step 6 + parallel component porting
+│   ├── steps_3_pipeline.md               ~100 lines  — steps 7–11 (config, stages, pipeline class, presets, registry)
+│   ├── steps_4_validation.md             ~120 lines  — steps 12–16 (smoke, pipeline parity, SSIM, cleanup, ask)
+│   ├── parity_testing.md                 ~250 lines  — conventions + component template + pipeline-level gate + subagent prompt
+│   ├── weight_conversion.md              ~200 lines  — decision tree + recipe + reference scripts + gotchas
+│   ├── layers_and_attention.md           ~150 lines  — linear/attention/primitive selection rules
+│   ├── i2v_variant.md                    ~270 lines  — full I2V add section verbatim (cleanly extractable)
+│   ├── distributed_support.md            ~50 lines   — SP/TP/VAE-tiling rules
+│   ├── registry_cheatsheet.md            ~30 lines   — register_configs fields
+│   ├── component_only_contributions.md   ~120 lines  — NEW (REVIEW #23): VAE-only / encoder-only PR shape
+│   └── audio_workload.md                 ~150 lines  — NEW (REVIEW #25): audio output, T2A workload, no-SSIM metrics
+├── reference/
+│   ├── pitfalls.md                       ~80 lines   — the 14-item pitfalls list
+│   ├── example_prompt.md                 ~40 lines   — the example user prompt snippet
+│   ├── links.md                          ~50 lines   — file/repo references
+│   └── changelog.md                      ~30 lines   — change history table
+└── seed-ssim-references/                 unchanged
+```
+
+Total: ~2200 lines across 17 files (vs current 1605 lines × 1 file).
+The line growth is intentional — each file gets a short "Purpose +
+Status + Prerequisites" header so it can be loaded without context
+from the others.
+
+## SKILL.md target shape (~250 lines)
+
+```markdown
+# Add a Model to FastVideo
+
+## Purpose
+[8 lines, unchanged]
+
+## When to use / When not to use
+[25 lines, unchanged]
+
+## Prerequisites — gather inputs (blocking)
+[40 lines, unchanged]
+
+## Inputs
+[15-line table, unchanged]
+
+## Files you will create or touch
+The full table lives in `how-to/files_table.md`. Quick sketch:
+- Component files (model + config + __init__ export): rows 1–6.
+- Pipeline files (config + class + stages + presets): rows 7–10.
+- Registry: row 11.
+- Tests (smoke + pipeline parity + per-component parity + SSIM): rows 12–15.
+- Conversion script (only if not Diffusers format): row 16.
+- Example: row 17.
+
+For component-only contributions (just a VAE / encoder), see
+`how-to/component_only_contributions.md` — you can skip rows 7–13 + 15
++ 17.
+
+## Steps (procedural index)
+
+1. **Gather inputs (blocking).** See `how-to/steps_1_setup.md`.
+2. **Study the reference implementation.** See `how-to/steps_1_setup.md`.
+3. **Decide what to reuse.** See `how-to/steps_1_setup.md`.
+4. **Convert weights to Diffusers format.** Only if not Diffusers-format.
+   See `how-to/weight_conversion.md`.
+5. **Clone the official repo for parity testing.** See `how-to/steps_1_setup.md`.
+6. **Port components in parallel via subagents.** See
+   `how-to/steps_2_components.md` + `how-to/parity_testing.md`.
+7. **Create the PipelineConfig.** See `how-to/steps_3_pipeline.md`.
+8. **Build or pick the stages.** See `how-to/architecture.md` (standard
+   stages catalog) + `how-to/steps_3_pipeline.md`.
+9. **Write the pipeline class.** See `how-to/steps_3_pipeline.md`.
+10. **Define presets.** See `how-to/steps_3_pipeline.md`.
+11. **Register in `fastvideo/registry.py`.** See `how-to/registry_cheatsheet.md`.
+12. **Smoke-test the pipeline.** See `how-to/steps_4_validation.md`.
+13. **Full-pipeline parity + example (gated).** See
+    `how-to/parity_testing.md` (the pipeline-level section is the
+    handoff gate).
+14. **Add SSIM regression.** See `how-to/steps_4_validation.md`. For
+    audio, see `how-to/audio_workload.md` (no SSIM analog).
+15. **Clean up the cloned reference repo.** See `how-to/steps_4_validation.md`.
+16. **Ask about tests + perf data.** See `how-to/steps_4_validation.md`.
+
+## Pre-handoff checklist
+[New section addressing REVIEW item 16 — bans skip-only parity at handoff]
+- [ ] `pytest tests/local_tests/<bucket>/test_<family>_*parity*.py -v` produces non-skip PASS for each non-reused component.
+- [ ] `pytest tests/local_tests/pipelines/test_<family>_pipeline_parity.py -v` produces non-skip PASS.
+- [ ] `python examples/inference/basic/basic_<family>.py` writes a non-corrupt mp4 (or .wav for audio).
+- [ ] Conversion has actually been run (the parity tests skip if not).
+
+## Outputs
+[15 lines, unchanged]
+
+## Common pitfalls
+See `reference/pitfalls.md` for the full 14-item list. Most-cited:
+- #1: `EntryClass` missing → pipeline silently invisible.
+- #11: raw `nn.Linear` in DiT/VAE hot paths → use `ReplicatedLinear`.
+- #16: skip-only parity → see pre-handoff checklist above.
+
+## See also
+- `reference/example_prompt.md` — example user prompt for invoking this skill.
+- `reference/links.md` — file/repo references.
+- `reference/changelog.md` — change history.
+```
+
+## Migration steps
+
+1. **Create the new files** with content extracted from current
+   `SKILL.md` (no semantic edits — just relocation + cross-link edits).
+2. **Add per-file headers** of the form:
+
+   ```markdown
+   # <topic>
+
+   **Part of:** add-model skill
+   **When to read:** <one-liner — e.g. "during step 6 / parallel
+   component porting">
+   **Prerequisites:** <links to other files needed first>
+
+   ---
+
+   <content>
+   ```
+
+3. **Rewrite SKILL.md** as the index. Each step gets the new 3-line
+   format: title + 2-line summary + link to `how-to/<file>.md`.
+4. **Add the pre-handoff checklist** (addresses REVIEW item 16) — the
+   single load-bearing addition this split enables.
+5. **Add `how-to/component_only_contributions.md`** (addresses REVIEW
+   item 23).
+6. **Add `how-to/audio_workload.md`** (addresses REVIEW items 25 + 28).
+7. **Update REVIEW.md** to mark items 16, 23, 25, 28 as resolved by
+   the split.
+8. **Re-run a sample skill invocation** (e.g. on a hypothetical new
+   port) to validate the split — check that the model only loads
+   the index + 2-3 satellite files for a typical port.
+
+## Risks / open questions
+
+- **Breaks existing prompts.** Anyone with an in-flight skill
+  invocation may have memorized section anchors that move. Mitigate
+  by leaving anchor stubs in SKILL.md for one cycle (forwarding
+  comments).
+- **Cross-link maintenance.** Each rename / move requires updating
+  cross-refs. Mitigate by keeping the file tree shallow (one
+  `how-to/` and one `reference/` directory only).
+- **Discoverability of new files.** A porter who only reads
+  `SKILL.md` may not realize `audio_workload.md` exists. Mitigate by
+  having the index's step-3 line for "audio variant of step X" link
+  explicitly to the audio doc, and by keeping the "See also" section
+  visible.
+- **What counts as "idiomatic"?** Anthropic skills tend toward
+  ~150-300 line single-file or ~3-5 file splits. 17 files is on the
+  large side. We might collapse further if some satellites are
+  always-loaded-together (e.g. merge `architecture.md` +
+  `registry_cheatsheet.md` if they're never read independently).
+  Decide post-prototype.
+
+## Next step
+
+Implement the split as a separate PR (don't bundle with the
+will/stable-audio first-class VAE work or the will/magi MagiHuman
+port). Sequence:
+
+1. Land the split as-is (mechanical relocation, zero semantic
+   change). Verify that running the skill against a known port
+   produces equivalent guidance.
+2. Then layer in the REVIEW-item edits (16, 23, 25, 28) as content
+   changes inside the new structure.

.agents/skills/evaluate-video-quality/SKILL.md

@@ -0,0 +1,128 @@
+---
+name: evaluate-video-quality
+description: Evaluate generated video quality using available metrics (SSIM, loss trajectory, caption consistency)
+---
+
+# Evaluate Video Quality
+
+## Purpose
+Assess the quality of videos generated by a training run. Combines multiple
+signals to give a holistic quality assessment. This skill is **evolving** —
+new metrics will be added as they are developed.
+
+## Prerequisites
+- Generated videos available locally or via W&B artifacts.
+- For SSIM: reference videos from official implementations.
+- For caption consistency: LLM access (optional, stub for now).
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `video_paths` | Yes | List of paths to generated videos |
+| `reference_paths` | No | Paths to reference videos (for SSIM) |
+| `prompts` | No | Prompts used to generate videos (for caption check) |
+| `loss_summary` | No | Path to W&B summary JSON (for loss trajectory) |
+| `metrics` | No | Which metrics to run (default: all available) |
+
+## Available Metrics
+
+Check `.agents/memory/evaluation-registry/README.md` for the current catalog.
+
+### SSIM (Active)
+
+Leverages the existing infrastructure in `fastvideo/tests/ssim/`.
+
+```bash
+pytest fastvideo/tests/ssim/ -vs --video-path <generated> --reference-path <reference>
+```
+
+Or use the SSIM utility directly:
+
+```python
+from fastvideo.tests.ssim.ssim_utils import compute_ssim
+score = compute_ssim(generated_video, reference_video)
+# score > 0.85 is typically "acceptable"
+```
+
+**Interpretation**:
+| SSIM Range | Quality |
+|------------|---------|
+| > 0.90 | Excellent — very close to reference |
+| 0.80–0.90 | Good — acceptable for most uses |
+| 0.70–0.80 | Fair — noticeable differences |
+| < 0.70 | Poor — significant quality issues |
+
+### Loss Trajectory (Active)
+
+Analyze the loss curve shape from W&B summary:
+
+```python
+import json
+with open(loss_summary_path) as f:
+    summary = json.load(f)
+
+final_loss = summary["train_loss"]
+runtime = summary["_runtime"]
+steps = summary["_step"]
+```
+
+**Early-stage heuristics** (first 500 steps):
+- Loss should be decreasing (even slightly).
+- Grad norm should be stable (no wild oscillations).
+- If loss is flat or increasing, flag for review.
+
+### Caption Consistency (Draft — Not Yet Calibrated)
+
+Use an LLM to evaluate whether the video content matches the input prompt.
+
+```
+Prompt: "A golden retriever playing in the snow"
+Video: <path>
+
+Score the video on:
+1. Object presence (is there a golden retriever?)
+2. Action accuracy (is it playing?)
+3. Environment match (is there snow?)
+4. Overall coherence (does it look natural?)
+
+Each 1-5, total /20.
+```
+
+> ⚠️ This metric is in **draft** status. Results should not be treated as
+> ground truth until calibrated against human judgments.
+
+## Steps
+
+1. **Identify available metrics** — Check `.agents/memory/evaluation-registry/README.md`.
+2. **Run each metric** — Collect scores.
+3. **Aggregate** — Produce a combined quality report.
+4. **Log** — Update the experiment journal with quality results.
+
+## Outputs
+
+```markdown
+## Video Quality Report: <experiment_name>
+
+| Metric | Score | Threshold | Status |
+|--------|-------|-----------|--------|
+| SSIM (avg) | 0.87 | > 0.80 | ✅ Pass |
+| Loss trajectory | decreasing | decreasing | ✅ Pass |
+| Caption consistency | 16/20 | > 14/20 | ✅ Pass |
+
+### Per-Video Scores
+| Video | SSIM | Caption |
+|-------|------|---------|
+| video_001.mp4 | 0.89 | 17/20 |
+| video_002.mp4 | 0.85 | 15/20 |
+```
+
+## References
+- `fastvideo/tests/ssim/` — SSIM test infrastructure
+- `fastvideo/tests/training/Vanilla/test_training_loss.py` — loss comparison
+- `.agents/memory/evaluation-registry/README.md` — metric catalog
+
+## Changelog
+| Date | Change |
+|------|--------|
+| 2026-03-02 | Initial version with SSIM, loss trajectory, caption consistency stub |

.agents/skills/index-related-work/SKILL.md

@@ -0,0 +1,94 @@
+---
+name: index-related-work
+description: Ingest a paper or repository into the related work index
+---
+
+# Index Related Work
+
+## Purpose
+Create a structured summary of a related paper, repository, or blog post and
+add it to `.agents/memory/related-work/` for future reference. This builds the
+agent's knowledge base for making informed decisions about training, evaluation,
+and architecture choices.
+
+## Prerequisites
+- Access to the paper/repo (URL, PDF, or local clone).
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `source` | Yes | URL, citation, or local path |
+| `type` | Yes | `paper`, `repo`, or `blog` |
+| `tags` | No | List of tags (default: inferred from content) |
+
+## Steps
+
+### 1. Extract key information
+
+For **papers**: Read abstract, method section, experimental setup, and results.
+For **repos**: Read README, key source files, and training scripts.
+For **blogs**: Read the full post.
+
+Focus on:
+- What problem does it solve?
+- What architecture/technique is used?
+- How does it relate to FastVideo's approach?
+
+### 2. Create the index entry
+
+Write to `.agents/memory/related-work/<slug>.md`:
+
+```markdown
+---
+title: <title>
+source: <URL or citation>
+type: paper | repo | blog
+date_indexed: <ISO-8601>
+tags: [world-model, distillation, evaluation, ...]
+---
+
+## Summary
+<1-2 paragraph summary.>
+
+## Key Differences from FastVideo
+- <comparison points>
+
+## Actionable Insights
+- <what we could adopt or adapt>
+```
+
+### 3. Update the catalog
+
+If `.agents/memory/related-work/_catalog.md` exists, append the new entry.
+If not, create it:
+
+```markdown
+# Related Work Catalog
+
+| Slug | Title | Type | Tags | Date |
+|------|-------|------|------|------|
+| <slug> | <title> | <type> | <tags> | <date> |
+```
+
+## Outputs
+- New file in `.agents/memory/related-work/<slug>.md`.
+- Updated catalog.
+
+## Example Usage
+
+```
+Index the Self-Forcing paper:
+
+  source: https://arxiv.org/abs/2406.xxxxx
+  type: paper
+  tags: [world-model, self-forcing, distillation]
+```
+
+## References
+- `.agents/memory/related-work/README.md` — schema documentation
+
+## Changelog
+| Date | Change |
+|------|--------|
+| 2026-03-02 | Initial version |

.agents/skills/index.jsonl

@@ -0,0 +1,9 @@
+{"name": "launch-experiment", "description": "Generate and execute a training launch command for FastVideo models", "path": "launch-experiment/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "monitor-experiment", "description": "Poll a running W&B training run for progress and emit structured alerts", "path": "monitor-experiment/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "summarize-run", "description": "Extract a W&B run summary into a structured experiment report", "path": "summarize-run/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "log-experiment", "description": "Append or update an experiment entry in the experiment journal", "path": "log-experiment/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "evaluate-video-quality", "description": "Evaluate generated video quality using available metrics (SSIM, loss trajectory, caption consistency)", "path": "evaluate-video-quality/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "index-related-work", "description": "Ingest a paper or repository into the related work index", "path": "index-related-work/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "search-related-work", "description": "Query the related work index for relevant papers, repos, or comparisons", "path": "search-related-work/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "seed-ssim-references", "description": "Run a new or updated fastvideo/tests/ssim/ test on Modal, pull generated videos, and upload them to FastVideo/ssim-reference-videos so the test has a regression baseline", "path": "seed-ssim-references/SKILL.md", "status": "draft", "trust": "low"}
+{"name": "review-add-model-pr", "description": "Review a PR that adds a new model / pipeline (or major variant) to FastVideo. Walks the canonical surface, indexes the 36 documented add-model failure modes by where they show up in a diff, and produces a structured verdict (block / nit / follow-up).", "path": "review-add-model-pr/SKILL.md", "status": "draft", "trust": "low"}

.agents/skills/launch-experiment/SKILL.md

@@ -0,0 +1,127 @@
+---
+name: launch-experiment
+description: Generate and execute a training launch command for FastVideo models
+---
+
+# Launch Experiment
+
+## Purpose
+Construct a fully-specified `torchrun` training command for a FastVideo model
+given a target pipeline, dataset, and hyperparameter overrides. This skill
+automates the boilerplate of setting environment variables, picking the right
+entrypoint, and applying defaults from the closest example script.
+
+## Prerequisites
+- The repo is cloned and `fastvideo` is installed (`uv pip install -e .[dev]`).
+- Dataset is preprocessed (see `docs/training/data_preprocess.md`).
+- `WANDB_API_KEY` is set in the environment (or `WANDB_MODE=offline` for local).
+- GPU resources are available (multi-GPU requires NCCL).
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `pipeline` | Yes | Training pipeline type: `finetune`, `distill-dmd`, `self-forcing`, `lora`, `consistency` |
+| `model` | Yes | Model family: `wan-t2v-1.3B`, `wan-i2v-14B`, `ltx2`, `matrixgame` |
+| `data_path` | Yes | Path to preprocessed dataset (parquet) |
+| `num_gpus` | Yes | Number of GPUs |
+| `overrides` | No | Dict of hyperparameter overrides (any CLI arg) |
+| `output_dir` | No | Output directory (default: `outputs/<model>_<pipeline>`) |
+| `run_name` | No | W&B run name (default: auto-generated) |
+
+## Steps
+
+### 1. Identify the training entrypoint
+
+| Pipeline | Entrypoint |
+|----------|-----------|
+| `finetune` (Wan T2V) | `fastvideo/training/wan_training_pipeline.py` |
+| `finetune` (Wan I2V) | `fastvideo/training/wan_i2v_training_pipeline.py` |
+| `finetune` (LTX-2) | `fastvideo/training/ltx2_training_pipeline.py` |
+| `finetune` (MatrixGame) | `fastvideo/training/matrixgame_training_pipeline.py` |
+| `distill-dmd` | `fastvideo/training/wan_distillation_pipeline.py` |
+| `self-forcing` | `fastvideo/training/wan_self_forcing_distillation_pipeline.py` |
+
+### 2. Resolve default hyperparameters
+
+Find the closest example script in `examples/training/` for the model:
+
+| Model | Example Script Directory |
+|-------|-------------------------|
+| `wan-t2v-1.3B` | `examples/training/finetune/wan_t2v_1.3B/crush_smol/` |
+| `wan-i2v-14B` | `examples/training/finetune/wan_i2v_14B_480p/crush_smol/` |
+| `ltx2` | `examples/training/finetune/ltx2/` |
+| `matrixgame` | `examples/training/finetune/MatrixGame2.0/` |
+| `distill-dmd` | `scripts/distill/v1_distill_dmd_wan.sh` |
+
+Read the script to extract default values for:
+- `--learning_rate`, `--train_batch_size`, `--sp_size`, `--tp_size`
+- `--num_latent_t`, `--num_height`, `--num_width`, `--num_frames`
+- `--gradient_accumulation_steps`, `--max_train_steps`
+- `--mixed_precision`, `--weight_decay`, `--max_grad_norm`
+- `--validation_steps`, `--validation_sampling_steps`
+
+### 3. Set environment variables
+
+```bash
+export WANDB_API_KEY="${WANDB_API_KEY}"
+export WANDB_BASE_URL="https://api.wandb.ai"
+export FASTVIDEO_ATTENTION_BACKEND=FLASH_ATTN
+export TOKENIZERS_PARALLELISM=false
+export TRITON_CACHE_DIR=/tmp/triton_cache
+```
+
+### 4. Construct the torchrun command
+
+```bash
+torchrun --nnodes 1 --nproc_per_node <num_gpus> \
+    <entrypoint> \
+    --pretrained_model_name_or_path <model_hf_id> \
+    --data_path "<data_path>" \
+    --output_dir "<output_dir>" \
+    --wandb_run_name "<run_name>" \
+    --tracker_project_name "<project_name>" \
+    --log_validation \
+    <...all hyperparameters...>
+```
+
+### 5. Log to experiment journal
+
+After launching, append an entry to `.agents/memory/experiment-journal/README.md`:
+
+```markdown
+## [YYYY-MM-DD] Experiment: <run_name>
+- **Hypothesis**: <user-provided or auto-generated>
+- **Config**: model=<model>, lr=<lr>, sp_size=<sp>, gpus=<n>, script=<entrypoint>
+- **W&B run**: <pending — will be updated by monitor skill>
+- **Status**: running
+```
+
+## Outputs
+- A ready-to-execute shell command.
+- An experiment journal entry.
+
+## Example Usage
+
+```
+Launch a Wan T2V 1.3B finetune on 4 GPUs with lr=5e-5 and max_train_steps=1000:
+
+  pipeline: finetune
+  model: wan-t2v-1.3B
+  data_path: data/crush_smol_preprocessed/
+  num_gpus: 4
+  overrides:
+    learning_rate: 5e-5
+    max_train_steps: 1000
+```
+
+## References
+- `examples/training/finetune/wan_t2v_1.3B/crush_smol/finetune_t2v.sh`
+- `scripts/distill/v1_distill_dmd_wan.sh`
+- `docs/training/finetune.md` (training arguments table)
+- `fastvideo/training/trackers.py` (tracker initialization)
+
+## Changelog
+| Date | Change |
+|------|--------|
+| 2026-03-02 | Initial version |

.agents/skills/log-experiment/SKILL.md

@@ -0,0 +1,87 @@
+---
+name: log-experiment
+description: Append or update an experiment entry in the experiment journal
+---
+
+# Log Experiment
+
+## Purpose
+Create or update an entry in `.agents/memory/experiment-journal/README.md` to maintain
+a living record of all experiments and their outcomes.
+
+## Prerequisites
+- `.agents/memory/experiment-journal/README.md` exists.
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `name` | Yes | Experiment name / identifier |
+| `hypothesis` | No | What you expected to learn |
+| `config` | Yes | Key config: model, lr, sp_size, gpus, script |
+| `wandb_run` | No | W&B run ID or URL |
+| `duration` | No | Total wall time |
+| `metrics` | No | Key metrics dict (loss, step_time, grad_norm) |
+| `checkpoint` | No | Path to checkpoint |
+| `insight` | No | What was learned |
+| `status` | Yes | `running`, `completed`, `failed`, `abandoned` |
+| `lessons` | No | Paths to related lesson files |
+
+## Steps
+
+### 1. Check for existing entry
+
+Search `.agents/memory/experiment-journal/README.md` for an entry with the same name.
+If found, update it instead of creating a duplicate.
+
+### 2. Format the entry
+
+```markdown
+## [YYYY-MM-DD] Experiment: <name>
+- **Hypothesis**: <hypothesis or "N/A">
+- **Config**: model=<model>, lr=<lr>, sp_size=<sp>, gpus=<n>, script=<script>
+- **W&B run**: <wandb_run or "pending">
+- **Duration**: <duration or "in progress">
+- **Key metrics**: loss=<loss>, step_time=<step_time>, grad_norm=<grad_norm>
+- **Checkpoint**: <checkpoint or "N/A">
+- **Insight**: <insight or "pending">
+- **Status**: <status>
+- **Related lessons**: <lessons or "none">
+```
+
+### 3. Insert at the top of the journal
+
+New entries go at the top of the file (after the header), so the most recent
+experiments are always visible first.
+
+### 4. Warn on duplicates
+
+If a similar experiment name exists with `status: completed`, warn that this
+may be a repeat. If it's `status: running`, assume this is an update.
+
+## Outputs
+- Updated `.agents/memory/experiment-journal/README.md`.
+
+## Example Usage
+
+```
+Log a completed experiment:
+
+  name: wan-t2v-finetune-lr5e5-sp4
+  config: model=wan-t2v-1.3B, lr=5e-5, sp_size=4, gpus=4
+  wandb_run: fastvideo/training/run_abc123
+  duration: 2h 15m
+  metrics: {loss: 0.065, step_time: 2.3, grad_norm: 0.35}
+  checkpoint: outputs/wan_finetune/checkpoint-1000
+  insight: LR 5e-5 converges 30% faster than 1e-5 with no quality loss
+  status: completed
+```
+
+## References
+- `.agents/memory/experiment-journal/README.md` — journal file
+- `.agents/workflows/experiment-lifecycle.md` — when to log
+
+## Changelog
+| Date | Change |
+|------|--------|
+| 2026-03-02 | Initial version |

.agents/skills/monitor-experiment/SKILL.md

@@ -0,0 +1,134 @@
+---
+name: monitor-experiment
+description: Poll a running W&B training run for progress and emit structured alerts
+---
+
+# Monitor Experiment
+
+## Purpose
+Continuously (or on-demand) check a running experiment's W&B metrics and emit
+alerts for anomalies. Supports the "30-minute quality check" paradigm: after
+the first 30 minutes of a long training run, produce a checkpoint quality
+report before committing more resources.
+
+## Prerequisites
+- `WANDB_API_KEY` is set in the environment.
+- The experiment is actively logging to W&B (not in `WANDB_MODE=offline`).
+- For offline mode: read from local `wandb-summary.json` instead.
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `run_id` | Yes* | W&B run ID (e.g., `entity/project/run_id`) |
+| `output_dir` | Yes* | Local output directory (for offline mode fallback) |
+| `poll_interval` | No | Seconds between polls (default: 60) |
+| `alert_on` | No | List of alert conditions to enable (default: all) |
+
+\* One of `run_id` or `output_dir` is required.
+
+## Steps
+
+### 1. Connect to the run
+
+**Online mode** (preferred):
+
+```python
+import wandb
+api = wandb.Api()
+run = api.run("<run_id>")
+```
+
+**Offline fallback**:
+
+```python
+import json
+summary_path = f"{output_dir}/tracker/wandb/latest-run/files/wandb-summary.json"
+with open(summary_path) as f:
+    summary = json.load(f)
+```
+
+### 2. Track key metrics
+
+| Metric | W&B Key | Description |
+|--------|---------|-------------|
+| Training loss | `train_loss` | Primary training loss |
+| Gradient norm | `grad_norm` | Gradient magnitude |
+| Step time | `step_time` | Wall-clock seconds per step |
+| Learning rate | `learning_rate` | Current LR |
+| Avg step time | `avg_step_time` | Running average step time |
+| Validation videos | `validation_videos_*` | Generated validation samples |
+
+### 3. Evaluate alert conditions
+
+| Alert | Condition | Severity |
+|-------|-----------|----------|
+| **Loss spike** | `current_loss > 3 × rolling_avg_loss` | 🔴 Critical |
+| **NaN/Inf gradient** | `grad_norm` is NaN or Inf | 🔴 Critical |
+| **Step time regression** | `step_time > 2 × baseline_step_time` | 🟡 Warning |
+| **No progress** | No new W&B logs for > 10 minutes | 🟡 Warning |
+| **Loss plateau** | Loss change < 1% over last 100 steps | 🟢 Info |
+
+### 4. Emit structured status
+
+Output format (agent-consumable):
+
+```json
+{
+  "run_id": "...",
+  "step": 500,
+  "metrics": {
+    "train_loss": 0.078,
+    "grad_norm": 0.41,
+    "step_time": 2.5,
+    "learning_rate": 1e-6
+  },
+  "alerts": [
+    {"type": "loss_spike", "severity": "critical", "message": "Loss jumped to 0.45 (avg: 0.08)"}
+  ],
+  "status": "running"
+}
+```
+
+### 5. 30-Minute Quality Check
+
+After the first 30 minutes of wall-clock time:
+1. Summarize the loss curve shape (decreasing? at what rate?).
+2. Check if validation videos have been generated.
+3. Report step count, loss at start vs. current, and estimated time to completion.
+4. Produce a go/no-go recommendation.
+
+```markdown
+## 30-Minute Check: <run_name>
+- **Steps completed**: 150
+- **Loss**: 0.12 → 0.08 (↓ 33%)
+- **Grad norm**: stable at ~0.4
+- **Step time**: 2.5s/step (consistent)
+- **Validation videos**: 5 generated at step 100
+- **Recommendation**: ✅ Continue — loss is decreasing normally
+```
+
+## Outputs
+- Structured JSON status updates.
+- Alert messages for anomalous conditions.
+- 30-minute checkpoint quality report.
+
+## Example Usage
+
+```
+Monitor W&B run "fastvideo/Wan_distillation/abc123":
+
+  run_id: fastvideo/Wan_distillation/abc123
+  poll_interval: 120
+  alert_on: [loss_spike, nan_gradient, step_time_regression]
+```
+
+## References
+- `fastvideo/training/trackers.py` — `WandbTracker` implementation
+- `fastvideo/tests/training/Vanilla/test_training_loss.py` — how summaries are compared
+- `fastvideo/tests/training/Vanilla/a40_reference_wandb_summary.json` — reference summary format
+
+## Changelog
+| Date | Change |
+|------|--------|
+| 2026-03-02 | Initial version |

.agents/skills/review-add-model-pr/SKILL.md

@@ -0,0 +1,482 @@
+---
+name: review-add-model-pr
+description: Use when reviewing a PR that adds a new model / pipeline (or a major variant like I2V/V2V/DMD) to FastVideo under `fastvideo/pipelines/basic/<family>/`. Walks a reviewer through the canonical surface the porter should have touched, the parity bar, and the 36 documented failure modes from prior ports. Returns a structured review verdict (block / nit / follow-up).
+---
+
+# Review an add-model PR
+
+## Purpose
+
+Catch the failure modes that have actually shipped in past add-model PRs
+(documented in `.agents/skills/add-model/REVIEW.md`) **before merge**,
+without re-litigating design choices that the `add-model` skill has
+already settled. The reviewer's job is *not* to redesign the port — it
+is to verify the porter did the things the skill prescribes, exercised
+the parity gate honestly, and surfaced the right knobs to end users.
+
+This skill assumes the PR claims to add a model. For PRs adding only
+sampling-preset tweaks, a single pipeline kwarg, or a new test against
+an existing pipeline, this skill is overkill — comment scoped to those
+specific changes instead.
+
+## When to use
+
+- A new pipeline directory under `fastvideo/pipelines/basic/<family>/`.
+- A new variant pipeline (e.g. `<family>_i2v_pipeline.py`,
+  `<family>_dmd_pipeline.py`) sibling-added to an existing family.
+- A "first-class component" PR that adds a new DiT, VAE, or encoder
+  port without (yet) wiring a pipeline (REVIEW item 23).
+- Re-review of a port that previously skipped the post-parity hot-path
+  pass (REVIEW item 33) or shipped with placeholder diffusers imports
+  (REVIEW item 30).
+
+## When not to use
+
+- PR only edits sampling defaults in an existing
+  `<family>/presets.py` → review the values against the model card
+  inline, no skill needed.
+- PR only adds a new SSIM reference → use the
+  `seed-ssim-references` skill instead.
+- PR refactors shared infra (`fastvideo/layers/`, `fastvideo/attention/`,
+  `fastvideo/pipelines/stages/` base classes) without touching a
+  family directory → that's not an add-model PR.
+
+## Required reading before starting the review
+
+1. **The PR description.** What family is being added, what variants
+   (T2V / I2V / V2V / DMD / T2A / …), what the porter claims is parity-
+   verified, and which of the four prereqs (official repo URL, HF
+   weights path, HF token, target `model_family`) they collected.
+2. **`.agents/skills/add-model/SKILL.md` Files-table** (rows 1–17) — the
+   canonical surface a model port touches. You'll cross-reference this
+   against the diff in step 2 below.
+3. **`.agents/skills/add-model/REVIEW.md` summary table** — the 36
+   failure modes. The "Pitfall map" section below indexes them by where
+   they typically show up in a diff.
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `pr` | Yes | GitHub PR number, URL, or local branch name. |
+| `model_family` | Recommended | The snake_case family slug (e.g. `stable_audio`, `magi_human`). Lets you grep-scope the review. |
+| `official_repo` | Recommended | URL of the upstream reference, so you can sanity-check the parity test references it correctly. |
+| `parity_run_log` | Optional | If the porter attached a parity run log, read it before reading code. |
+
+## Steps
+
+### 1. Verify the PR scope is shaped like an add-model PR
+
+Run `gh pr view <pr> --json files | jq -r '.files[].path'` (or
+`git diff --name-only origin/main...HEAD`). Confirm the diff touches
+**at least 6 of the 17 Files-table rows** in `add-model/SKILL.md`.
+Typical shape:
+
+| If you see... | Expect... | If missing... |
+|---|---|---|
+| `fastvideo/models/dits/<family>.py` | `fastvideo/configs/models/dits/<family>.py` + export in `__init__.py` | Block: DiT config + export missing (Files-table rows 2 + 3). |
+| `fastvideo/pipelines/basic/<family>/<family>_pipeline.py` | `presets.py`, `registry.py` edit, smoke test, parity test, example | Block on whichever is missing — the porter will be told the same by the skill. |
+| `<family>_i2v_pipeline.py` (sibling-added) | A new preset for the I2V workload + the registry entry pointing at the I2V HF repo | Block: I2V variants get their own preset + workload tag (REVIEW item 8). |
+| Standalone DiT/VAE/encoder under `fastvideo/models/<bucket>/` with no pipeline | A "first-class component" PR (REVIEW item 23) | Confirm with the author that the consuming pipeline PR is in flight or planned. Don't block. |
+| Edits under `fastvideo/configs/pipelines/base.py`, `fastvideo/api/sampling_param.py`, `fastvideo/pipelines/pipeline_batch_info.py` | New pipeline-call kwargs the family needs | Sanity-check the field's default doesn't change behavior for other families. Often a footgun (e.g. `0 or default` Python truthiness). |
+
+### 2. Walk the diff in dependency order, not file order
+
+Components have a strict dependency order — review them in this order
+so you can spot mismatches as they cascade:
+
+1. **DiT** (`fastvideo/models/dits/<family>.py` + config) — foundational.
+2. **VAE** — only one usually exists; if not, this is the second-biggest
+   review surface.
+3. **Encoder(s) / conditioner** — text encoder is usually shared; new
+   conditioners (e.g. `MultiConditioner`-style) need careful look.
+4. **`PipelineConfig`** subclass.
+5. **Pipeline class** — wires modules + stages.
+6. **Stages** — most should be standard; mod-specific subclasses
+   warrant careful review.
+7. **Presets + registry** — usually mechanical; check workload type
+   selection against REVIEW item 28.
+8. **Tests** — both component parity and pipeline parity.
+9. **Example script** — last, but the bar is high (REVIEW items 31, 35).
+
+For each component, run the Per-component checklist (next section).
+
+### 3. Apply the Per-component checklist
+
+For each new file in `fastvideo/models/<bucket>/<family>.py`:
+
+#### DiT review (`fastvideo/models/dits/<family>.py`)
+
+- [ ] **No raw `nn.Linear`** in QKV/MLP/proj/embedder paths. All
+  projections should be `ReplicatedLinear` from
+  `fastvideo.layers.linear`. Exceptions are the legacy MoE-packed
+  case (REVIEW item 11) — must be flagged in a comment with the
+  weight-layout justification.
+- [ ] **No raw SDPA / `flash_attn_func`** calls. Self-attention should
+  be `DistributedAttention` (or `LocalAttention` for cross-attention).
+  See `add-model/SKILL.md` "Attention layers" table.
+- [ ] **No raw `nn.LayerNorm` in modulation paths.** Should be
+  `FP32LayerNorm` / `RMSNorm` / `LayerNormScaleShift` from
+  `fastvideo.layers.layernorm`.
+- [ ] **No `from diffusers import` or `from transformers import
+  <ModelClass>`** anywhere outside of test files (REVIEW item 30, the
+  hard ban). Tokenizers are the *only* allowed `from transformers`
+  runtime import in production code.
+- [ ] **`from_official_state_dict()` (or equivalent)** is provided so
+  the consuming pipeline can load the published checkpoint without
+  going through `diffusers.from_pretrained`.
+- [ ] **Param-mapping deviations from upstream are localized** — e.g.
+  if the model uses `gamma`/`beta` and FastVideo's `FP32LayerNorm`
+  uses `weight`/`bias`, the remap happens *in the loader*, not by
+  renaming the layers.
+- [ ] **Partial / non-standard rotary** (e.g. halves-swap vs
+  interleaved-pair) is kept local with a one-line WHY comment, not a
+  copy of FastVideo's `_apply_rotary_emb` with edits.
+
+#### VAE review (`fastvideo/models/vaes/<family or arch>.py`)
+
+- [ ] **File name matches the convention** (REVIEW item 29): name
+  after the *arch* if the VAE is shared across families
+  (`oobleck.py`, `autoencoder_kl.py`); name after the *family* if it's
+  specific (`wanvae.py`).
+- [ ] **`from_pretrained()` (or `from_official_state_dict`)** loads
+  weights from the published HF path *without* a Diffusers
+  intermediary (REVIEW item 30 again).
+- [ ] **Per-channel `latents_mean`/`latents_std` are reshaped with
+  explicit `.view(1, -1, 1, 1, 1)`** when applied (REVIEW item 22) —
+  silent broadcasting along the wrong dim is a stealth bug.
+- [ ] **Normalization-convention mismatches with the wrapper**
+  (REVIEW item 20) — if upstream `decode()` does
+  `z = z*std + mean` *internally* but the FastVideo wrapper expects
+  pre-denormalized input, the parity test must compensate or the
+  decode parity will look broken.
+- [ ] **Pipeline-glue wrapper** (e.g. `fastvideo/models/vaes/<family>_audio.py`
+  for lazy-load semantics) exists if the pipeline needs lazy-load /
+  hide-from-named_parameters semantics (REVIEW item 26).
+
+#### Conditioner / encoder review
+
+- [ ] If the conditioner mixes text + numeric (e.g. duration), it
+  produces the **DiT-ready (cross_attn_cond, cross_attn_mask,
+  global_embed) triple** in a single helper, not via inline cat-ing
+  in the conditioning stage.
+- [ ] **T5 / Llama / SigLIP TP wiring** uses the encoder bucket's TP
+  primitives (`QKVParallelLinear`, `MergedColumnParallelLinear`,
+  `RowParallelLinear`) — not `ReplicatedLinear`.
+- [ ] If the conditioner intentionally hides its T5 from
+  `named_parameters()` (so the SA-style checkpoint loader doesn't try
+  to match upstream-absent T5 keys), that exclusion is documented in
+  a comment.
+
+#### `PipelineConfig` review (`fastvideo/configs/pipelines/<family>.py`
+or `fastvideo/pipelines/basic/<family>/pipeline_configs.py`)
+
+- [ ] **Subclasses `PipelineConfig`** from
+  `fastvideo.configs.pipelines.base`.
+- [ ] **`vae_config`, `dit_config`, `text_encoder_configs`** are
+  defaulted via `field(default_factory=...)` (mutable-default-arg
+  Python rule).
+- [ ] **The text-encoder slot** matches what the pipeline actually
+  uses. If the pipeline owns its own conditioner (no FastVideo-loaded
+  text encoder), the text-encoder tuples should be `tuple()` and the
+  parent's length-equality validator should still pass — the porter
+  must zero out *all four* of `text_encoder_configs`,
+  `text_encoder_precisions`, `preprocess_text_funcs`,
+  `postprocess_text_funcs` together (we've seen this break before).
+- [ ] **Component-bucket inheritance** (REVIEW item 24) — if the new
+  config goes under `vaes/`, it must subclass `VAEConfig` not
+  `EncoderConfig`. The bucket directory determines the base.
+- [ ] **`__post_init__`** is used only to flip `load_encoder` /
+  `load_decoder` / etc. on the child configs, not to do any heavy
+  build.
+
+#### Pipeline class review (`fastvideo/pipelines/basic/<family>/<family>_pipeline.py`)
+
+- [ ] **`EntryClass` is a single class**, not a list.
+- [ ] **`_required_config_modules`** lists exactly what the loader
+  reads from `model_index.json`. Missing keys cause silent loading
+  degradation.
+- [ ] **`load_modules()`** does not have any `from diffusers import`
+  / `from transformers import <ModelClass>` for production
+  components. If the porter explicitly opted into a temporary
+  diffusers shim, REJECT — the right move is to ship the native port
+  or hold the pipeline back (REVIEW item 30).
+- [ ] **`torch.backends.*` flags** (TF32, cuDNN benchmark) — if set,
+  they're set **once in `load_modules`**, not per-call inside a stage
+  (REVIEW item 33). Mid-run flips invalidate the cuDNN algorithm
+  cache and amplify A2A SDE drift.
+- [ ] **`create_pipeline_stages()`** uses standard stages from
+  `fastvideo.pipelines.stages` where possible, only subclassing when
+  the math diverges. New stage classes live in
+  `fastvideo/pipelines/basic/<family>/stages/` and are re-exported in
+  `fastvideo/pipelines/stages/__init__.py`.
+- [ ] **One pipeline class for kwargs-driven variants** (REVIEW item
+  34) — T2A/A2A/inpaint that share weights/components shouldn't be
+  split into three classes. Triggers to split: separate
+  `_required_config_modules`, separate HF repo, divergent forward
+  signatures, separate `WorkloadType`. Reject the split unless one
+  of those applies.
+
+#### Stages review (`fastvideo/pipelines/basic/<family>/stages/*.py`)
+
+- [ ] **No `init_audio_strength = 0` / "0 or default" footguns** —
+  Python truthy-or fallbacks (`x or default`) silently swallow `0`,
+  `0.0`, `""`. Use `x if x is not None else default`.
+- [ ] **Loud-fail on malformed kwarg combos** — e.g. inpaint without
+  mask should raise `ValueError`, not silently fall through to T2A.
+- [ ] **Hot-path discipline** (REVIEW item 33): no per-step
+  `torch.zeros_like(...)` or `torch.randn_like(...)` inside the
+  sampler loop; pre-allocate buffers + reuse via `.normal_()` /
+  `.copy_()`.
+- [ ] **No dead `batch.extra` writes** — every key written should be
+  read by a downstream stage.
+- [ ] **Stage docstring** is one or two lines. No "Mirrors upstream X"
+  / "Vendored from Y" provenance (REVIEW item 36); upstream
+  comparison belongs in the parity test, not the production docstring.
+
+#### Presets + registry review
+
+- [ ] **Sampling defaults** match the published model card example
+  block. Track the *model* defaults, not the upstream library's
+  *generic* defaults (this distinction shipped a 100% drift before
+  for Stable Audio).
+- [ ] **`workload_type`** matches what the model actually does. If
+  the family is audio (T2A/A2A/AV) and `WorkloadType` doesn't yet
+  have those values (REVIEW item 28), accept `"t2v"` as a placeholder
+  but require a TODO comment + the porter to file a follow-up.
+- [ ] **Registry detector** matches both the HF path and the
+  pipeline class name (`_class_name` from `model_index.json`).
+- [ ] **`ALL_PRESETS`** export exists and is added to
+  `_register_presets()`'s group tuple in `fastvideo/registry.py`.
+
+#### Tests review — **the most load-bearing review surface**
+
+REVIEW item 16 is the single most expensive failure mode. A skipped
+test reads as green in CI. Confirm explicitly:
+
+- [ ] **Component parity tests exist** for every non-reused
+  component: DiT, VAE (if new), encoder (if new). Find them under
+  `tests/local_tests/<bucket>/test_<family>_*.py`.
+- [ ] **Each component parity test produces a non-skip pass on the
+  reviewer's machine** if at all possible. If the porter says "I ran
+  it locally", ask for the diff numbers in the PR description.
+- [ ] **Pipeline parity test exists** at
+  `tests/local_tests/pipelines/test_<family>_pipeline_parity.py` AND
+  has a non-skip pass — *not* a "skipped because the official clone
+  is missing" pass.
+- [ ] **Smoke test exists** at `test_<family>_pipeline_smoke.py` — no
+  GPU, just import + registry + preset wiring. CI can run this even
+  if local-only parity tests skip.
+- [ ] **Parity reference is the official upstream**, not diffusers
+  (REVIEW item 30). Diffusers parity is acceptable as a *secondary*
+  test only when (a) the published weights load through both and (b)
+  the official repo is also imported and compared.
+- [ ] **Tolerances are scope-appropriate** (REVIEW item 21): single-
+  block + single-kernel = `atol=1e-4`; full-DiT cross-kernel = `0.1`
+  with a complementary `abs_mean drift < 5%` check; bare `assert_close`
+  alone with a loose tolerance is a smell.
+- [ ] **Stub helpers for upstream private DSL deps** (REVIEW items
+  17, 18, 32) — if the upstream has `magi_compiler`-style imports
+  that aren't on PyPI, a small `tests/local_tests/helpers/<family>_upstream.py`
+  shim is fine. But if the deps it bypasses become real installs,
+  the shim must be deleted (item 32 — no zombie no-op shims).
+- [ ] **GQA-aware kernel routing in stub paths** (REVIEW item 19) —
+  if the parity test routes upstream's flash_attn through SDPA, KV
+  heads must be `repeat_interleave`'d explicitly.
+- [ ] **VAE normalization symmetry** (REVIEW item 20) — if upstream
+  bundles `z = z*std + mean` in `decode()` and FastVideo expects
+  pre-denormalized, the test must compensate explicitly.
+
+#### Example script review (`examples/inference/basic/basic_<family>*.py`)
+
+The bar here is high — the example is the user's entry point, not the
+porter's debugging script.
+
+- [ ] **User-story-shaped docstring** (REVIEW item 31) — at least one
+  `User story (<persona>):` block, then a "How it works" / "Picking
+  the dial" / "Tunable knobs" section. Not a code-narration docstring.
+- [ ] **5-15 LOC of constants + one `generate_video()` call**
+  (REVIEW item 35). If the example carries a 25-line `_load_reference`
+  / shape-norm / resample helper, that glue belongs *in the
+  pipeline*, not the example.
+- [ ] **Pipeline accepts file paths** for any media-input kwarg
+  (`init_audio`, `inpaint_audio`, `image_path`). The example just
+  passes the path; pipeline does decode + resample internally.
+- [ ] **No `torchaudio.load` on container formats** (mp4, m4a) —
+  routes through `torchcodec` → CUDA NVRTC. PyAV (already a
+  FastVideo dep, used by `_mux_audio`) handles all formats.
+- [ ] **Tunable-knob defaults** match the model's published sweet
+  spot, not the upstream library's generic defaults.
+
+#### Local repro doc review (optional but recommended)
+
+If the family ships a `tests/local_tests/<family>.md`:
+
+- [ ] **Setup section** covers HF gated access, optional inference
+  deps, upstream clone instructions, model cache pre-warm.
+- [ ] **Per-test table** explains what each test compares against
+  with expected drift numbers.
+- [ ] **Troubleshooting section** covers gated-skip behavior, batch-
+  vs-single-run flag interactions (cuDNN benchmark!), first-call
+  cache download blowup.
+
+### 4. Cross-check the "first-class component" rule
+
+Run `grep -rn "from diffusers import\|from transformers import"
+fastvideo/pipelines/basic/<family>/ fastvideo/models/dits/<family>*
+fastvideo/models/vaes/<family>* fastvideo/models/encoders/<family>*`.
+
+The **only** acceptable hits are `from transformers import
+<TokenizerFast>` (data-utility, no weights) and `from transformers
+import T5EncoderModel` *only if* (a) it's loaded via HF and (b) the
+T5 weights are absent from the model's checkpoint by design (e.g.
+SA conditioner).
+
+Anything else — `StableAudioDiTModel`, `AutoencoderKL`,
+`UnetXxx`, `T2VPipeline` — is a REVIEW item 30 violation. Block
+the PR with a pointer to that item.
+
+### 5. Run the parity tests yourself if budget permits
+
+The porter said it passes. Confirm:
+
+```bash
+# DiT/VAE/encoder component parity
+pytest tests/local_tests/<bucket>/test_<family>_*.py -v -s
+
+# Pipeline parity (the gate per add-model step 13(a))
+pytest tests/local_tests/pipelines/test_<family>_pipeline_parity.py -v -s
+
+# Smoke (no GPU)
+pytest tests/local_tests/pipelines/test_<family>_pipeline_smoke.py -v
+```
+
+If any *parity* test SKIPs rather than PASSes on your machine and
+you have the prerequisites set up, that means the porter never
+actually verified parity (REVIEW item 16). Block.
+
+### 6. Read REVIEW.md for any new failure modes the porter didn't address
+
+The PR may also amend REVIEW.md with newly-discovered failure modes.
+Skim those — they're typically the most accurate signal of what to
+look for in *this specific* port. If the porter added a REVIEW item
+and the linked code in the PR doesn't actually mitigate that item,
+that's a contradiction — flag it.
+
+### 7. Write the verdict
+
+Structure your review comment as three buckets:
+
+- **Block (must fix before merge)** — REVIEW item 30 violations,
+  skipped parity tests, missing required Files-table rows, dead
+  `batch.extra` writes that hide caller bugs, footguns like `0 or
+  default`.
+- **Nit (would be better)** — naming convention drift, narrative
+  comments per REVIEW item 36, missing user-story docstrings,
+  hot-path allocations that don't change correctness.
+- **Follow-up (not this PR)** — extracting test boilerplate to a
+  shared helper, future variant pipelines, performance benchmarks.
+
+Always link each item back to the `add-model/REVIEW.md` item number
+when applicable; that's the institutional memory and lets the porter
+fix the issue with full context.
+
+## Pitfall map (REVIEW.md items by where they show up in a diff)
+
+Use this when you've spotted something off and want to find the
+documented failure mode it corresponds to.
+
+| If the diff has... | Suspect REVIEW item(s) |
+|---|---|
+| `from diffusers import <ModelClass>` in `fastvideo/...` | **30** (hard ban) |
+| Raw `nn.Linear` in DiT projections | 10, 11 (only acceptable for MoE-packed weights with comment) |
+| Custom RoPE that's *not* `_apply_rotary_emb` | Probably fine if the convention differs (e.g. halves-swap), but require a one-line WHY comment + a parity-test row |
+| `parity_test.py` that calls `pytest.skip` unconditionally | **16** (silent no-op trap) — block |
+| Component parity tests missing | **16** — block |
+| New pipeline file with no `register_configs` call in registry.py | Files-table row 11 missing |
+| `_required_config_modules` that doesn't match `model_index.json` keys | Silent loading degradation; block |
+| `text_encoder_configs=tuple()` but other text-encoder tuples non-empty | Length-equality validator will fail at runtime |
+| ArchConfig has `num_inference_steps` / `guidance_scale` / `flow_shift` | **15a** — pipeline-level fields leaking into ArchConfig; block |
+| `<family>vae.py` for an arch-shared VAE (e.g. Oobleck) | **29** — should be `<arch>.py` (e.g. `oobleck.py`) |
+| Pipeline imports `from transformers import T5EncoderModel` | OK only if the conditioner intentionally hides T5 from `named_parameters()` and the checkpoint omits T5 keys; document the exception |
+| `torch.backends.*` set in a stage's `forward()` | **33** — should be one-shot in `load_modules` |
+| `init_X = 0` silently treated as default via `or` | **33-adjacent** — Python truthy footgun; flag |
+| Per-step `torch.zeros_like` / `randn_like` in sampler callback | **33** — pre-allocate; can also fix accuracy regressions |
+| Magic `_DOWNSAMPLING_RATIO = 2048` in stage code | **33** — derive from the VAE config |
+| `examples/.../basic_<family>*.py` >30 LOC | **35** — decode/resample/shape-norm belongs in the pipeline |
+| Example uses `torchaudio.load(...)` on mp4 | **35** — torchcodec → NVRTC dep chain; use PyAV |
+| Example docstring narrates code | **31** — needs `User story (<persona>):` block |
+| Pipeline class docstring narrates upstream provenance | **36** — strip "Vendored from X" / "Mirrors upstream Y" |
+| Comment says "previously this was X, we moved it because Y" | **36** — strip; goes in commit message, not code |
+| Diff adds `tests/local_tests/helpers/<family>_upstream.py` that's a no-op | **32** — delete the no-op shim + its call sites |
+| Diff adds `init_audio` / `inpaint_audio` / similar to `SamplingParam` | OK; verify they default to `None` and other families ignore |
+| Multiple pipeline classes that share weights/components | **34** — should be one class with kwargs-driven modes |
+| Config under `fastvideo/configs/models/encoders/` for a VAE | **24** — wrong bucket; should be under `vaes/` with `VAEConfig` base |
+| New pipeline that depends on a not-yet-ported component (placeholder import) | **30** — hold the pipeline back until the component is ported |
+
+## Outputs
+
+A structured review comment on the PR with:
+
+1. **Verdict** — `approve` / `request-changes` / `comment-only`.
+2. **Block list** — REVIEW-item-linked issues that must be fixed.
+3. **Nit list** — style / convention drift.
+4. **Follow-up list** — items the porter should know about but
+   shouldn't fix in this PR.
+5. **Parity numbers you confirmed** — diff_max / diff_mean / drift /
+   element-wise bound, per parity test you ran. Lets the next reviewer
+   skip re-running.
+
+## Example review comment skeleton
+
+```
+## Review summary
+
+**Verdict:** request-changes
+
+I ran the smoke + DiT-component parity tests and read the full diff.
+The native ports look right; two REVIEW-30 violations in the pipeline
+file need resolving before merge, plus a few smaller nits.
+
+### Block (must fix)
+
+- `fastvideo/pipelines/basic/<family>/<family>_pipeline.py:NN` — `from
+  diffusers import <ModelClass>` violates REVIEW item 30 (hard ban).
+  Either ship the first-class port now or hold this pipeline back
+  until the component is ported.
+- `tests/local_tests/pipelines/test_<family>_pipeline_parity.py` skips
+  on my machine even with HF token set (the upstream-clone path check
+  fails). REVIEW item 16: a parity test that always skips is worse
+  than no test. Update the path resolution or document it in
+  `tests/local_tests/<family>.md`.
+
+### Nit
+
+- `examples/inference/basic/basic_<family>.py` docstring is code-narration;
+  REVIEW item 31 wants a `User story (<persona>):` block.
+- `<family>_pipeline.py:NN` has a per-call `torch.backends.cuda.matmul.allow_tf32 = False`;
+  REVIEW item 33 says move to one-shot in `load_modules`.
+
+### Follow-up
+
+- HF-token boilerplate is duplicated across 4 parity files. REVIEW
+  item 1-tier work, but doesn't need to land in this PR.
+
+### Parity numbers I confirmed locally
+
+| Test | diff_max | diff_mean | drift |
+|---|---|---|---|
+| DiT component parity | 0.0 | 0.0 | bit-identical |
+| VAE decode parity | 0.0 | 0.0 | bit-identical |
+| Pipeline parity (T2V, 25 steps) | 0.012 | 0.0009 | 0.31% |
+```
+
+## References
+
+- `.agents/skills/add-model/SKILL.md` — the canonical procedure;
+  every "should be there" claim in this skill is a row in that
+  skill's Files-table.
+- `.agents/skills/add-model/REVIEW.md` — 36 documented failure modes;
+  the Pitfall map above indexes them by where they show up in a diff.
+- `.agents/skills/seed-ssim-references/SKILL.md` — for SSIM regression
+  add-on PRs (separate from this skill's scope).

.agents/skills/search-related-work/SKILL.md

@@ -0,0 +1,82 @@
+---
+name: search-related-work
+description: Query the related work index for relevant papers, repos, or comparisons
+---
+
+# Search Related Work
+
+## Purpose
+Search through `.agents/memory/related-work/` to find indexed papers, repos,
+or blog posts relevant to a query. Use this when you need to understand how
+other work compares to FastVideo's approach, or when looking for techniques
+to adopt.
+
+## Prerequisites
+- The related work index has entries (`.agents/memory/related-work/*.md`).
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `query` | Yes | Natural language query |
+| `tags` | No | Filter by tags (e.g., `[distillation, evaluation]`) |
+| `type` | No | Filter by type (`paper`, `repo`, `blog`) |
+
+## Steps
+
+### 1. Search the index
+
+Use grep-based search through `.agents/memory/related-work/`:
+
+```bash
+# Search by content
+grep -rl "<query>" .agents/memory/related-work/
+
+# Search by tags (in frontmatter)
+grep -l "tags:.*<tag>" .agents/memory/related-work/*.md
+```
+
+### 2. Rank results
+
+For each matching file:
+1. Read the file.
+2. Score relevance to the query based on:
+   - Title match
+   - Tag match
+   - Content match (summary, differences, insights)
+3. Return top results.
+
+### 3. Format output
+
+```markdown
+## Related Work Search: "<query>"
+
+### 1. <Title> (relevance: high)
+- **Source**: <URL>
+- **Tags**: <tags>
+- **Key insight**: <most relevant excerpt>
+- **File**: `.agents/memory/related-work/<slug>.md`
+
+### 2. <Title> (relevance: medium)
+...
+```
+
+## Outputs
+- Ranked list of relevant related work entries with excerpts.
+
+## Example Usage
+
+```
+Search for work related to video quality evaluation metrics:
+
+  query: "video generation quality evaluation metrics"
+  tags: [evaluation]
+```
+
+## References
+- `.agents/memory/related-work/README.md` — index schema
+
+## Changelog
+| Date | Change |
+|------|--------|
+| 2026-03-02 | Initial version |

.agents/skills/seed-ssim-references/SKILL.md

@@ -0,0 +1,250 @@
+---
+name: seed-ssim-references
+description: Seed HF reference videos for a single newly-added SSIM test. Runs the test on Modal L40S, downloads the generated mp4s via `modal volume get`, pauses for the user to eyeball quality, then uploads only that test's files to `FastVideo/ssim-reference-videos`. Use when a new `fastvideo/tests/ssim/test_*_similarity.py` has just been added and has no references on HF yet.
+---
+
+# Seed SSIM Reference Videos
+
+## Purpose
+
+A brand-new SSIM test in `fastvideo/tests/ssim/` fails forever until its
+reference videos exist on the HF dataset (`FastVideo/ssim-reference-videos`).
+This skill:
+
+1. Runs the test on Modal's L40S pool to generate the videos.
+2. Downloads them to the local repo via `modal volume get`.
+3. Pauses so the user can eyeball the mp4s and confirm quality.
+4. Uploads only the new test's files to HF, with a guard that refuses to
+   overwrite anything already present.
+
+The skill is run **manually**, once per new test. Before invoking it, the user
+has already sanity-tested the new test locally — it launches `VideoGenerator`
+and writes an mp4 without crashing. The skill does not re-test locally; it
+goes straight to Modal L40S (which is what CI uses).
+
+## When to use
+
+- A new `test_*_similarity.py` file has been added in `fastvideo/tests/ssim/`
+  and the HF dataset has no `reference_videos/default/L40S_reference_videos/<model_id>/`
+  subtree for it yet.
+
+## When not to use
+
+- Regular CI runs — once refs exist, `pytest fastvideo/tests/ssim/` downloads
+  them automatically.
+- Re-seeding an existing test. That requires `--force` on the upload step, and
+  is out of scope here; treat as a separate, deliberate operation.
+
+## Inputs
+
+The skill has **one required input**: the path to the new SSIM test file.
+Prompt the user for it if they didn't supply it.
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `test_file` | Yes | e.g. `fastvideo/tests/ssim/test_ltx2_similarity.py`. The skill's first action is to ask for this if missing. |
+
+Everything else is fixed:
+
+- Modal runner GPU: **L40S** (hardcoded in `fastvideo/tests/modal/ssim_test.py`).
+- Device folder: `L40S_reference_videos`.
+- Quality tier: `default` (the tier CI runs). The `full_quality` tier is not
+  seeded by this skill.
+- HF repo: `FastVideo/ssim-reference-videos` (dataset).
+- Multi-model test files: all model ids in `*_MODEL_TO_PARAMS` are seeded
+  together; the Modal run produces one mp4 per (model, prompt, backend) and
+  the upload scopes by `--model-id`, looping if there is more than one.
+
+## Prerequisites
+
+The user has confirmed:
+
+- `modal` CLI authenticated.
+- `HF_API_KEY` (or `HUGGINGFACE_HUB_TOKEN` / `HF_TOKEN`) exported with write
+  access to `FastVideo/ssim-reference-videos`.
+- The test file runs locally end-to-end (generates an mp4; SSIM assertion
+  failure due to missing reference is expected and fine).
+
+Fail fast if the token env var is missing.
+
+## Steps
+
+### 1. Ask for the test file
+
+If the user didn't name one, ask: *"Which SSIM test file do you want to seed
+references for? (e.g. `fastvideo/tests/ssim/test_ltx2_similarity.py`)"*.
+
+Validate:
+
+- Path exists and matches `fastvideo/tests/ssim/test_*_similarity.py`.
+- File defines a `*_MODEL_TO_PARAMS` dict — grep it to extract the set of
+  model ids. Those ids drive step 5.
+
+If either check fails, stop and tell the user what's wrong.
+
+### 2. Run the test on Modal L40S
+
+Pick a subdir name so repeated runs don't collide:
+
+```bash
+SHORT_COMMIT=$(git rev-parse --short=12 HEAD)
+TIMESTAMP=$(date -u +%Y%m%d_%H%M%S)
+SUBDIR="${TIMESTAMP}_${SHORT_COMMIT}"
+```
+
+Then launch the Modal run:
+
+```bash
+modal run fastvideo/tests/modal/ssim_test.py \
+    --git-repo="$(git config --get remote.origin.url)" \
+    --git-commit="$(git rev-parse HEAD)" \
+    --hf-api-key="$HF_API_KEY" \
+    --test-files="<test_file>" \
+    --sync-generated-to-volume \
+    --generated-volume-subdir="$SUBDIR" \
+    --skip-reference-download \
+    --no-fail-fast
+```
+
+Flag rationale:
+- `--skip-reference-download`: no refs exist yet, so conftest must not try to
+  pull them.
+- `--no-fail-fast`: lets the test finish generation before `_assert_similarity`
+  raises `FileNotFoundError: Reference video folder does not exist`. The
+  expected failure is what we want — the mp4 has already been written.
+- `--sync-generated-to-volume` + `--generated-volume-subdir`: copies the
+  generated mp4s to the `hf-model-weights` Modal volume under
+  `ssim_generated_videos/default/<SUBDIR>/generated_videos/` so we can pull
+  them locally.
+
+The Modal run will end with a nonzero exit (expected) and print a
+`modal volume get hf-model-weights ssim_generated_videos/default/<SUBDIR>/generated_videos ./generated_videos_modal/default`
+command. Capture that `<SUBDIR>` — you need it for step 3.
+
+### 3. Download generated videos locally
+
+```bash
+modal volume get --force hf-model-weights \
+    ssim_generated_videos/default/"$SUBDIR"/generated_videos \
+    ./generated_videos_modal/default
+```
+
+`--force` is required when the parent `./generated_videos_modal/default`
+already exists; without it, `modal volume get` errors with `[Errno 21] Is a
+directory`. Safe to pass on the first run too.
+
+After this, the mp4s live at
+`./generated_videos_modal/default/generated_videos/L40S_reference_videos/<model_id>/<backend>/<prompt>.mp4`.
+The extra `generated_videos/` level comes from the volume layout in
+`_sync_generated_videos_to_volume` (`ssim_test.py`) — the command copies
+`<repo>/fastvideo/tests/ssim/generated_videos/<tier>` to
+`ssim_generated_videos/<tier>/<SUBDIR>/generated_videos/`, and `modal volume
+get` preserves that trailing `generated_videos/` segment.
+
+### 4. PAUSE — user reviews quality
+
+Print the list of downloaded mp4s and their paths, then stop. Tell the user:
+
+> "Generated videos downloaded to `./generated_videos_modal/default/generated_videos/L40S_reference_videos/`. Please open them and confirm the quality looks correct. Reply **`upload`** to continue, or anything else to abort."
+
+Do not proceed until the user explicitly says `upload`. If they abort, leave
+everything on disk so they can inspect further — no cleanup.
+
+### 5. Copy into the local reference layout
+
+Scoped copy — only the new test's mp4s. Loop over each `<model_id>` extracted
+in step 1:
+
+```bash
+python fastvideo/tests/ssim/reference_videos_cli.py copy-local \
+    --quality-tier default \
+    --device-folder L40S_reference_videos \
+    --generated-dir ./generated_videos_modal/default/generated_videos/L40S_reference_videos
+```
+
+(The `--generated-dir` points at the device-folder root inside the
+downloaded tree; `copy-local` walks all `<model>/<backend>/*.mp4`
+underneath it. Since the Modal run was scoped to a single test file via
+`--test-files`, only that test's model(s) are present — so the copy is
+implicitly per-test.)
+
+Result: `fastvideo/tests/ssim/reference_videos/default/L40S_reference_videos/<model_id>/<backend>/<prompt>.mp4`.
+
+### 6. Upload to HF — scoped per model_id, with overwrite guard
+
+For each `<model_id>`:
+
+```bash
+python fastvideo/tests/ssim/reference_videos_cli.py upload \
+    --quality-tier default \
+    --device-folder L40S_reference_videos \
+    --model-id "<model_id>"
+```
+
+The upload command:
+
+- Uploads **only** `reference_videos/default/L40S_reference_videos/<model_id>/`.
+- **Refuses** if any file already exists at that path on HF (this is the
+  guard — seeding a new test should never clobber existing refs). To override,
+  the user must re-run with `--force`. If the guard fires, stop and report
+  exactly which files exist; do not silently `--force`.
+
+Reads the HF token from `HF_API_KEY` / `HUGGINGFACE_HUB_TOKEN` / `HF_TOKEN`.
+
+### 7. Report success
+
+List what was uploaded (paths in repo) and remind the user to push any
+related code changes. Do **not** auto-verify by re-running Modal — the user
+can run `pytest fastvideo/tests/ssim/<test_file>` later to confirm end-to-end;
+it will auto-download the refs they just uploaded.
+
+## Failure modes and how to handle them
+
+- **`HF_API_KEY` unset.** Stop before step 2. The Modal run needs it (passed
+  via `--hf-api-key`), and step 6 needs it for upload.
+- **Modal run fails before generation.** No mp4s on the volume — nothing to
+  download. Fix the test locally (`pytest fastvideo/tests/ssim/<test_file>`)
+  and retry from step 2.
+- **`./generated_videos_modal/default/L40S_reference_videos/` missing after
+  `modal volume get`.** The run didn't produce videos (most likely the test
+  crashed before writing, or `REQUIRED_GPUS` exceeded the partition capacity
+  — see Modal logs).
+- **Upload guard fires (files already exist).** The test name / model id
+  collides with something already on HF. Verify the user actually wants to
+  replace existing refs; if so, re-run the upload with `--force`. If not,
+  rename the model id in `*_MODEL_TO_PARAMS` and re-seed.
+- **Quality looks wrong in step 4.** Abort. The mp4s stay on disk for
+  inspection. The fix is usually in the test's params (resolution, steps,
+  seed) — edit the test, then re-run the skill.
+
+## Design notes (for future skill maintainers)
+
+- The skill deliberately runs on Modal, **not** locally, because the CI
+  runner is L40S. Seeding from a different GPU SKU produces refs that CI's
+  L40S runs can't match (SSIM drifts across SKUs).
+- The skill is default-tier only. `full_quality` refs are seeded by a
+  separate, deliberate operation — they double runtime and aren't what CI
+  gates on.
+- The overwrite guard in `reference_videos_cli.py upload` is default-on
+  specifically because this skill exists. Re-seeding is a distinct operation
+  that requires explicit `--force`.
+
+## References
+
+- `fastvideo/tests/modal/ssim_test.py` — Modal orchestrator; see
+  `--sync-generated-to-volume`, `--generated-volume-subdir`,
+  `--skip-reference-download`, `--no-fail-fast`.
+- `fastvideo/tests/ssim/reference_videos_cli.py` — `copy-local`, `upload`
+  (with `--model-id`, `--force`), `download`, `ensure` subcommands.
+- `fastvideo/tests/ssim/README.md` — reference layout, HF repo conventions.
+- `fastvideo/tests/ssim/inference_similarity_utils.py` —
+  `run_text_to_video_similarity_test` + `_build_init_kwargs`: what each test
+  config passes to `VideoGenerator.from_pretrained`.
+
+## Changelog
+
+| Date | Change |
+|------|--------|
+| 2026-04-17 | Initial version (Modal sync-to-volume flow). |
+| 2026-04-21 | Rewrite: single-test scope, explicit user-review pause, per-`model_id` upload, HF overwrite guard. Dropped `scripts/seed_ssim.sh`. |
+| 2026-04-21 | Post-first-run fixes: `modal volume get` needs `--force` when parent exists; download tree has an extra `generated_videos/` level so `--generated-dir` must reflect it. |

.agents/skills/summarize-run/SKILL.md

@@ -0,0 +1,137 @@
+---
+name: summarize-run
+description: Extract a W&B run summary into a structured experiment report
+---
+
+# Summarize Run
+
+## Purpose
+After a training run completes (or at any checkpoint), extract key metrics from
+the W&B run summary and produce a structured markdown report. Supports both
+online (W&B API) and offline (local `wandb-summary.json`) modes.
+
+## Prerequisites
+- Run has completed or reached a checkpoint with a saved summary.
+- For online: `WANDB_API_KEY` set in environment.
+- For offline: access to `<output_dir>/tracker/wandb/latest-run/files/wandb-summary.json`.
+
+## Inputs
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `run_id` | Yes* | W&B run ID for online access |
+| `output_dir` | Yes* | Local output dir for offline access |
+| `reference_run` | No | Path to reference `wandb-summary.json` for comparison |
+| `experiment_name` | No | Name for the journal entry (default: from W&B) |
+
+\* One of `run_id` or `output_dir` is required.
+
+## Steps
+
+### 1. Load run summary
+
+**Online**:
+
+```python
+import wandb
+api = wandb.Api()
+run = api.run("<run_id>")
+summary = dict(run.summary)
+config = dict(run.config)
+```
+
+**Offline** (existing codebase pattern from `fastvideo/tests/training/`):
+
+```python
+import json
+summary_path = f"{output_dir}/tracker/wandb/latest-run/files/wandb-summary.json"
+with open(summary_path) as f:
+    summary = json.load(f)
+```
+
+### 2. Extract key fields
+
+| Field | Source | Description |
+|-------|--------|-------------|
+| `train_loss` | `summary["train_loss"]` | Final training loss |
+| `avg_step_time` | `summary["avg_step_time"]` | Average seconds per step |
+| `step_time` | `summary["step_time"]` | Last step time |
+| `grad_norm` | `summary["grad_norm"]` | Final gradient norm |
+| `learning_rate` | `summary["learning_rate"]` | Final LR |
+| `_step` | `summary["_step"]` | Total steps completed |
+| `_runtime` | `summary["_runtime"]` | Total wall-clock seconds |
+| `validation_videos_*` | `summary[key]` | Validation video artifacts |
+
+### 3. Compare against reference (optional)
+
+Follow the pattern in `fastvideo/tests/training/Vanilla/test_training_loss.py`:
+
+```python
+# Fields to compare
+compare_fields = ["train_loss", "grad_norm", "avg_step_time"]
+tolerance = 0.05  # 5% relative tolerance
+
+for field in compare_fields:
+    ref_val = reference_summary[field]
+    cur_val = summary[field]
+    diff_pct = abs(cur_val - ref_val) / abs(ref_val) * 100
+    status = "✅" if diff_pct < tolerance * 100 else "⚠️"
+    print(f"{status} {field}: {cur_val:.4f} (ref: {ref_val:.4f}, diff: {diff_pct:.1f}%)")
+```
+
+### 4. Generate report
+
+```markdown
+# Run Summary: <experiment_name>
+
+| Metric | Value | Reference | Diff |
+|--------|-------|-----------|------|
+| Train Loss | 0.0788 | 0.0800 | -1.5% ✅ |
+| Avg Step Time | 2.81s | 2.80s | +0.4% ✅ |
+| Grad Norm | 0.408 | 0.410 | -0.5% ✅ |
+| Total Steps | 500 | — | — |
+| Wall Time | 23m 30s | — | — |
+
+## Configuration
+- Model: Wan-AI/Wan2.1-T2V-1.3B-Diffusers
+- Learning Rate: 1e-6
+- Batch Size: 1
+- GPUs: 8 × (SP=1, TP=1)
+- Mixed Precision: bf16
+
+## Validation Videos
+<list of validation video paths if available>
+
+## Notes
+<any observations or anomalies>
+```
+
+### 5. Update experiment journal
+
+Append or update the experiment's entry in `.agents/memory/experiment-journal/README.md`
+with the final metrics and status.
+
+## Outputs
+- Structured markdown report.
+- Updated experiment journal entry.
+
+## Example Usage
+
+```
+Summarize the run in output directory "outputs/wan_finetune":
+
+  output_dir: outputs/wan_finetune
+  reference_run: fastvideo/tests/training/Vanilla/a40_reference_wandb_summary.json
+  experiment_name: wan-t2v-finetune-lr1e6
+```
+
+## References
+- `fastvideo/tests/training/Vanilla/test_training_loss.py` — reference comparison pattern
+- `fastvideo/tests/training/Vanilla/a40_reference_wandb_summary.json` — example summary
+- `fastvideo/tests/training/lora/test_lora_training.py` — LoRA summary comparison
+- `fastvideo/training/trackers.py` — tracker summary generation
+
+## Changelog
+| Date | Change |
+|------|--------|
+| 2026-03-02 | Initial version |

.agents/workflows/evaluation-development.md

@@ -0,0 +1,54 @@
+---
+description: How to develop, validate, and register a new evaluation metric
+---
+
+# Evaluation Development SOP
+
+Standard procedure for adding new video quality evaluation metrics to the
+FastVideo agent toolkit.
+
+## When to Use
+
+- You need a metric that doesn't exist in `.agents/memory/evaluation-registry/README.md`.
+- An existing metric needs significant changes to its methodology.
+- You're exploring a new evaluation approach.
+
+## Steps
+
+### 1. Research
+
+- Search `.agents/memory/related-work/` for existing evaluation approaches.
+- Check the `evaluation_registry.md` for current metrics and their limitations.
+- Review literature: FVD, CLIP-Score, human preference, etc.
+
+### 2. Prototype
+
+- Write a standalone script in `.agents/exploration/<metric-name>.md`.
+- Keep it simple: one script, minimal dependencies.
+- Test on a few known-good and known-bad video samples.
+
+### 3. Validate
+
+- **Known-good test**: Metric should score high on reference-quality videos.
+- **Known-bad test**: Metric should score low on degraded/unrelated videos.
+- **Sensitivity test**: Small quality differences should produce meaningful
+  score differences.
+- Document thresholds and their justification.
+
+### 4. Register
+
+Update `.agents/memory/evaluation-registry/README.md`:
+- Add the metric with status `Active`.
+- Document location, thresholds, and trust level.
+
+### 5. Integrate
+
+Update `.agents/skills/evaluate-video-quality.md`:
+- Add the new metric as a section.
+- Include code examples and interpretation guide.
+
+### 6. Document
+
+- Move the exploration log content into the skill.
+- Clean up the exploration file or mark it as `promoted`.
+- If anything went wrong during development, create a lesson.

.agents/workflows/experiment-journaling.md

@@ -0,0 +1,47 @@
+---
+description: When and how to log experiments in the experiment journal
+---
+
+# Experiment Journaling SOP
+
+Ensures every experiment is properly recorded with context and outcomes.
+
+## When to Log
+
+**Always.** Every experiment — even quick tests — should be journaled.
+
+## Steps
+
+### 1. Before Launch — Create Draft Entry
+
+Use the `log-experiment` skill with `status: running`:
+- Include hypothesis and config.
+- Leave metrics, duration, and insight blank.
+
+### 2. After 30-Minute Check — Update with Initial Metrics
+
+Update the entry with:
+- Current loss and its trajectory direction.
+- Step time.
+- Number of validation videos generated.
+- Preliminary go/no-go assessment.
+
+### 3. On Completion — Fill Final Entry
+
+Update the entry with `status: completed`:
+- Final loss, grad norm, avg step time.
+- Total duration and steps.
+- Checkpoint path.
+- Key insight.
+
+### 4. On Failure — Document Failure Mode
+
+Update the entry with `status: failed`:
+- What went wrong (OOM, NaN, crash, etc.).
+- At what step the failure occurred.
+- Create a lesson in `.agents/lessons/` for non-trivial failures.
+
+### 5. Cross-Reference
+
+- Link related lessons: `**Related lessons**: .agents/lessons/<filename>.md`
+- Link related experiments: if this is a follow-up, reference the prior entry.

.agents/workflows/experiment-lifecycle.md

@@ -0,0 +1,87 @@
+---
+description: End-to-end experiment lifecycle from hypothesis to lessons learned
+---
+
+# Experiment Lifecycle SOP
+
+Standard operating procedure for running ML training experiments on
+FastVideo-WorldModel. Every experiment should follow this flow.
+
+## Overview
+
+```
+Plan → Launch → Monitor → Summarize → Journal → Reflect
+```
+
+## Steps
+
+### 1. Plan the Experiment
+
+Before launching:
+- [ ] Define a clear **hypothesis** (what you expect to learn).
+- [ ] Select the **model** and **pipeline** type (finetune, distill, lora, etc.).
+- [ ] Prepare the **dataset** (preprocessed into parquet format).
+- [ ] Review existing experiments in `.agents/memory/experiment-journal/README.md` for related work.
+- [ ] Check `.agents/lessons/` for known pitfalls with this configuration.
+- [ ] Document the plan in the experiment journal as a draft entry.
+
+### 2. Launch the Experiment
+
+Use the `launch-experiment` skill:
+- Provide: pipeline, model, data_path, num_gpus, and any hyperparameter overrides.
+- The skill generates the `torchrun` command and creates a journal entry.
+- Verify the command looks correct before executing.
+
+Reference: `.agents/skills/launch-experiment.md`
+
+### 3. Monitor the Experiment
+
+Use the `monitor-experiment` skill:
+- Provide the W&B run ID (or output_dir for offline).
+- Monitor alerts: loss spikes, NaN gradients, step time regressions.
+- At the **30-minute mark**: perform the quality check.
+  - Is loss decreasing?
+  - Are validation videos reasonable?
+  - Is step time consistent?
+- **Decision point**: Continue or abort based on the 30-min check.
+
+Reference: `.agents/skills/monitor-experiment.md`
+
+### 4. Summarize the Run
+
+After completion (or at any checkpoint), use the `summarize-run` skill:
+- Extract final metrics from W&B summary.
+- Compare against reference runs if available.
+- Generate a structured report.
+
+Reference: `.agents/skills/summarize-run.md`
+
+### 5. Update the Experiment Journal
+
+Use the `log-experiment` skill to update the journal entry:
+- Fill in final metrics, duration, checkpoint paths.
+- Record the key insight learned.
+- Set status to `completed`, `failed`, or `abandoned`.
+
+Reference: `.agents/skills/log-experiment.md`
+
+### 6. Reflect and Capture Lessons
+
+After every experiment:
+- **What went right?** → Note in the journal insight field.
+- **What went wrong?** → Create a lesson in `.agents/lessons/`:
+  - Use the template in `.agents/lessons/README.md`.
+  - Cross-reference the experiment journal entry.
+- **What was surprising?** → Consider creating an exploration log if this
+  warrants further investigation.
+
+Reference: `.agents/workflows/lesson-capture.md`
+
+## Validation Criteria
+
+This SOP is validated when an agent can:
+1. Follow steps 1–6 end-to-end for a minimal training run
+   (e.g., `examples/training/finetune/wan_t2v_1.3B/crush_smol/finetune_t2v.sh`
+   with `--max_train_steps 5`).
+2. Produce a complete experiment journal entry.
+3. Generate a run summary report.

.agents/workflows/lesson-capture.md

@@ -0,0 +1,71 @@
+---
+description: Post-experiment reflection to capture lessons learned
+---
+
+# Lesson Capture SOP
+
+Systematic procedure for turning experiment outcomes into persistent knowledge.
+
+## When to Use
+
+After **every** completed or failed experiment. Even successful experiments
+can yield lessons (e.g., "LR 5e-5 works better than 1e-5 for LoRA").
+
+## Steps
+
+### 1. Review the Experiment
+
+Read the experiment journal entry. Ask:
+- Did anything go wrong?
+- Was anything surprising?
+- Did anything take longer than expected?
+- Was a workaround needed?
+
+### 2. Decide: Lesson or Not?
+
+| Situation | Action |
+|-----------|--------|
+| Something broke | Create a lesson (category: `infrastructure` or `data`) |
+| Hyperparameter choice mattered | Create a lesson (category: `hyperparameter`) |
+| Porting issue found | Create a lesson (category: `porting`) |
+| Evaluation metric was misleading | Create a lesson (category: `evaluation`) |
+| Everything went smoothly | No lesson needed, but note in the journal insight |
+
+### 3. Create the Lesson File
+
+In `.agents/lessons/`, create `<YYYY-MM-DD>_<short-slug>.md`:
+
+```markdown
+---
+date: <ISO-8601>
+experiment: <journal entry reference>
+category: hyperparameter | data | infrastructure | evaluation | porting
+severity: critical | important | minor
+---
+
+# <Short Descriptive Title>
+
+## What Happened
+<description>
+
+## Root Cause
+<analysis>
+
+## Fix / Workaround
+<resolution>
+
+## Prevention
+<how to avoid in future>
+```
+
+### 4. Cross-Reference
+
+- Update the experiment journal entry with a link to the lesson file.
+- If a similar lesson already exists, add a reference or update it.
+
+### 5. Periodic Pattern Review
+
+Every ~10 lessons, scan for patterns:
+- Multiple lessons in the same category → consider a new skill or SOP.
+- Repeated mistakes → strengthen the relevant SOP with a checklist item.
+- Infrastructure issues → propose a codebase fix.

.agents/workflows/sync-dashboard.md

@@ -0,0 +1,67 @@
+---
+description: Synchronize the STATUS.md dashboard by scanning .agents/ directories
+---
+
+# Sync Dashboard
+
+Updates `.agents/STATUS.md` by scanning the skills, workflows, memory, lessons,
+and exploration directories to reflect what actually exists on disk.
+
+## When to Use
+
+- After adding, removing, or renaming any file in `.agents/`.
+- Periodically (e.g., at end of each conversation session).
+- When the dashboard feels out of date.
+
+## Steps
+
+### 1. Scan directories
+
+List all files in each directory:
+
+```bash
+echo "=== Skills ==="
+ls -1 .agents/skills/*.md 2>/dev/null | grep -v SKILL_TEMPLATE
+
+echo "=== Workflows ==="
+ls -1 .agents/workflows/*.md 2>/dev/null
+
+echo "=== Memory ==="
+ls -1 .agents/memory/*.md 2>/dev/null
+ls -1 .agents/memory/related-work/*.md 2>/dev/null | grep -v README
+
+echo "=== Lessons ==="
+ls -1 .agents/lessons/*.md 2>/dev/null | grep -v README
+
+echo "=== Exploration ==="
+ls -1 .agents/exploration/*.md 2>/dev/null | grep -v README
+```
+
+### 2. Compare with STATUS.md
+
+For each file found:
+- If it's in STATUS.md → leave it (preserve status/trust/tested fields).
+- If it's NOT in STATUS.md → add it with status `🔴 Stub`, trust `None`, tested `❌`.
+
+For each entry in STATUS.md:
+- If the file no longer exists → mark it as `❌ Removed` or delete the row.
+
+### 3. Update counts
+
+Recalculate the summary table at the top:
+- Count files per category.
+- Count by status (Ready, Draft, Stub).
+
+### 4. Update timestamp
+
+Set `_Last synced: <current date>_` at the top of STATUS.md.
+
+### 5. Review
+
+Read through the updated STATUS.md for accuracy. Flag anything that looks wrong.
+
+## Notes
+
+- Do NOT change trust levels during sync — those are set manually after testing.
+- Do NOT change status during sync — status changes require actual validation.
+- This workflow only handles structural sync (file existence), not content review.

.buildkite/performance-benchmarks/tests/wan-t2v-1.3b.json

@@ -0,0 +1,46 @@
+{
+  "benchmark_id": "wan-t2v-1.3b-2gpu",
+  "description": "Wan2.1 T2V 1.3B inference performance",
+  "model": {
+    "model_path": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
+    "model_short_name": "Wan2.1-T2V-1.3B"
+  },
+  "init_kwargs": {
+    "num_gpus": 2,
+    "flow_shift": 7.0,
+    "sp_size": 2,
+    "tp_size": 1,
+    "vae_sp": true,
+    "vae_tiling": true,
+    "text_encoder_precisions": ["fp32"]
+  },
+  "generation_kwargs": {
+    "height": 480,
+    "width": 832,
+    "num_frames": 45,
+    "num_inference_steps": 4,
+    "guidance_scale": 3,
+    "embedded_cfg_scale": 6,
+    "seed": 1024,
+    "fps": 24,
+    "neg_prompt": "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+  },
+  "test_prompts": [
+    "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."
+  ],
+  "run_config": {
+    "num_warmup_runs": 1,
+    "num_measurement_runs": 3,
+    "required_gpus": 2
+  },
+  "thresholds": {
+    "L40S": {
+      "max_generation_time_s": 34.0,
+      "max_peak_memory_mb": 11000.0
+    },
+    "default": {
+      "max_generation_time_s": 120.0,
+      "max_peak_memory_mb": 30000.0
+    }
+  }
+}

.buildkite/pipeline.yml

@@ -2,20 +2,218 @@ env:
   IMAGE_VERSION: "py3.12-latest"
   BUILDKITE_CLEAN_CHECKOUT: true
 
+notify:
+  - github_commit_status:
+      context: "fastcheck-passed"
+    if: build.env("TEST_SCOPE") == "fastcheck" || build.env("TEST_SCOPE") == null
+  - github_commit_status:
+      context: "full-suite-passed"
+    if: build.env("TEST_SCOPE") == "full"
+  - github_commit_status:
+      context: "direct-test-completed"
+    if: build.env("TEST_SCOPE") == "direct"
+
 steps:
-  - label: "pre-commit"
-    command: ".buildkite/scripts/pre_commit.sh"
-    agents:
-      queue: "default"
+  # ============================================================
+  # Direct test: triggered by /test <name> slash command.
+  # Labels match fastcheck/full-suite counterparts so the GitHub
+  # check status overwrites the original failed check.
+  # Only ONE step executes per build (gated by TEST_TYPE).
+  # ============================================================
 
-  - wait
+  # --- Fastcheck-scope direct tests ---
+    - label: ":microscope: Encoder Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "encoder"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":microscope: VAE Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "vae"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":microscope: Transformer Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "transformer"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":microscope: Kernel Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "kernel_tests"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":microscope: Unit Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "unit_test"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
 
-  - label: "Trigger Tests"
-    plugins:
-      - monorepo-diff#v1.4.0:
-          diff: 'git fetch origin "$BUILDKITE_PULL_REQUEST_BASE_BRANCH" && git diff --name-only origin/"$BUILDKITE_PULL_REQUEST_BASE_BRANCH"...HEAD'
-          watch:
-            - path: 
+  # --- Full-suite-scope direct tests ---
+    - label: ":bar_chart: SSIM Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "ssim"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+          - exit_status: 1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: LoRA Inference Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "inference_lora"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: Training Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "training"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: Distillation DMD Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "distillation_dmd"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: Self-Forcing Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "self_forcing"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: LoRA Training Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "training_lora"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+          - exit_status: 1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: Training Tests VSA"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "training_vsa"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+          - exit_status: 1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: Inference Tests VMoBA"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "inference_vmoba"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: Performance Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "performance"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+    - label: ":test_tube: API Server Tests"
+      if: build.env("TEST_SCOPE") == "direct" && build.env("TEST_TYPE") == "api_server"
+      command: "timeout 90m .buildkite/scripts/pr_test.sh"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      agents:
+        queue: "default"
+
+  # ============================================================
+  # Fastcheck: Runs on every PR (~10-15 min parallel)
+  # Core component validation: encoders, VAEs, transformers,
+  # CUDA kernels, and unit tests.
+  # ============================================================
+    - label: "Trigger Fastcheck"
+      if: build.env("TEST_SCOPE") == "fastcheck" || build.env("TEST_SCOPE") == null
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      plugins:
+        - monorepo-diff#v1.4.0:
+            diff: 'git fetch origin "${BUILDKITE_PULL_REQUEST_BASE_BRANCH:-main}" && git diff --name-only "origin/${BUILDKITE_PULL_REQUEST_BASE_BRANCH:-main}...HEAD"'
+            watch:
+            - path:
                 - "fastvideo/models/encoders/**"
                 - "fastvideo/models/loader/**"
                 - "fastvideo/tests/encoders/**"
@@ -23,7 +221,7 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 20m .buildkite/scripts/pr_test.sh"
-                label: "Encoder Tests"
+                label: ":microscope: Encoder Tests"
                 env:
                   - TEST_TYPE=encoder
                 agents:
@@ -36,7 +234,7 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 20m .buildkite/scripts/pr_test.sh"
-                label: "VAE Tests"
+                label: ":microscope: VAE Tests"
                 env:
                   - TEST_TYPE=vae
                 agents:
@@ -51,20 +249,68 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Transformer Tests"
+                label: ":microscope: Transformer Tests"
                 env:
                   - TEST_TYPE=transformer
                 agents:
                   queue: "default"
-            - path: 
+            - path:
+                - "fastvideo-kernel/**"
+                - "pyproject.toml"
+                - "docker/Dockerfile.python3.12"
+              config:
+                command: "timeout 15m .buildkite/scripts/pr_test.sh"
+                label: ":microscope: Kernel Tests"
+                env:
+                  - TEST_TYPE=kernel_tests
+                agents:
+                  queue: "default"
+            - path:
+                - "fastvideo/**"
+                - ".buildkite/**"
+                - ".github/**"
+                - "pyproject.toml"
+                - "docker/Dockerfile.python3.12"
+              config:
+                command: "timeout 15m .buildkite/scripts/pr_test.sh"
+                label: ":microscope: Unit Tests"
+                env:
+                  - TEST_TYPE=unit_test
+                agents:
+                  queue: "default"
+
+  # ============================================================
+  # Full Suite: Runs when TEST_SCOPE=full
+  # Triggered by adding the 'ready' label (via ci-trigger-full-suite.yml)
+  # or on-demand via /test full slash command.
+  # Includes integration tests, SSIM regression, training pipelines,
+  # and performance benchmarks.
+  # ============================================================
+    - label: "Trigger Full Suite"
+      if: build.env("TEST_SCOPE") == "full"
+      retry:
+        automatic:
+          - exit_status: 128
+            limit: 3
+          - exit_status: -1
+            limit: 2
+      plugins:
+        - monorepo-diff#v1.4.0:
+            diff: 'git fetch origin "${BUILDKITE_PULL_REQUEST_BASE_BRANCH:-main}" && git diff --name-only "origin/${BUILDKITE_PULL_REQUEST_BASE_BRANCH:-main}...HEAD"'
+            watch:
+            - path:
                 - "fastvideo/**/*.py"
                 - "pyproject.toml"
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 90m .buildkite/scripts/pr_test.sh"
-                label: "SSIM Tests"
+                label: ":bar_chart: SSIM Tests"
                 env:
                   - TEST_TYPE=ssim
+                retry:
+                  automatic:
+                    - exit_status: 1
+                      limit: 2
                 agents:
                   queue: "default"
             - path:
@@ -77,7 +323,7 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 20m .buildkite/scripts/pr_test.sh"
-                label: "LoRA Inference Tests"
+                label: ":test_tube: LoRA Inference Tests"
                 env:
                   - TEST_TYPE=inference_lora
                 agents:
@@ -88,7 +334,7 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Training Tests"
+                label: ":test_tube: Training Tests"
                 env:
                   - TEST_TYPE=training
                 agents:
@@ -99,7 +345,7 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Distillation DMDTests"
+                label: ":test_tube: Distillation DMD Tests"
                 env:
                   - TEST_TYPE=distillation_dmd
                 agents:
@@ -111,7 +357,7 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Self-Forcing Tests"
+                label: ":test_tube: Self-Forcing Tests"
                 env:
                   - TEST_TYPE=self_forcing
                 agents:
@@ -122,9 +368,13 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "LoRA Training Tests"
+                label: ":test_tube: LoRA Training Tests"
                 env:
                   - TEST_TYPE=training_lora
+                retry:
+                  automatic:
+                    - exit_status: 1
+                      limit: 2
                 agents:
                   queue: "default"
             - path:
@@ -134,32 +384,15 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Training Tests VSA"
+                label: ":test_tube: Training Tests VSA"
                 env:
                   - TEST_TYPE=training_vsa
+                retry:
+                  automatic:
+                    - exit_status: 1
+                      limit: 2
                 agents:
                   queue: "default"
-            - path:
-                - "fastvideo/**"
-                - "fastvideo-kernel/**"
-                - "pyproject.toml"
-                - "docker/Dockerfile.python3.12"
-              config:
-                command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Inference Tests STA"
-                env:
-                  - TEST_TYPE=inference_sta
-                agents:
-                  queue: "default"
-            - path:
-                - "fastvideo-kernel/**"
-                - "pyproject.toml"
-                - "docker/Dockerfile.python3.12"
-              config:
-                command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Kernel Tests"
-                env:
-                  - TEST_TYPE=kernel_tests
             - path:
                 - "fastvideo-kernel/**"
                 - "fastvideo/attention/backends/vmoba.py"
@@ -167,30 +400,39 @@ steps:
                 - "docker/Dockerfile.python3.12"
               config:
                 command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Inference Tests VMoBA"
-                env:  
+                label: ":test_tube: Inference Tests VMoBA"
+                env:
                   - TEST_TYPE=inference_vmoba
                 agents:
                   queue: "default"
             - path:
-                - "fastvideo/**"
+                - "fastvideo/models/dits/**"
+                - "fastvideo/pipelines/**"
+                - "fastvideo/attention/**"
+                - "fastvideo/layers/**"
+                - "fastvideo/worker/**"
+                - "fastvideo/entrypoints/**"
+                - "fastvideo/tests/performance/**"
+                - ".buildkite/performance-benchmarks/**"
                 - "pyproject.toml"
                 - "docker/Dockerfile.python3.12"
               config:
-                command: "timeout 15m .buildkite/scripts/pr_test.sh"
-                label: "Unit Tests"
+                command: "timeout 30m .buildkite/scripts/pr_test.sh"
+                label: ":test_tube: Performance Tests"
                 env:
-                  - TEST_TYPE=unit_test
+                  - TEST_TYPE=performance
+                agents:
+                  queue: "default"
+            - path:
+                - "fastvideo/entrypoints/openai/**"
+                - "fastvideo/entrypoints/cli/serve.py"
+                - "fastvideo/tests/entrypoints/test_openai_api_integration.py"
+                - "pyproject.toml"
+                - "docker/Dockerfile.python3.12"
+              config:
+                command: "timeout 30m .buildkite/scripts/pr_test.sh"
+                label: ":test_tube: API Server Tests"
+                env:
+                  - TEST_TYPE=api_server
                 agents:
                   queue: "default"
-            # - path:
-            #     - "scripts/lora_extraction/**"
-            #     - "pyproject.toml"
-            #     - "docker/Dockerfile.python3.12"
-            #   config:
-            #     command: "timeout 90m .buildkite/scripts/pr_test.sh"
-            #     label: "LoRA Extraction Tests"
-            #     env:
-            #       - TEST_TYPE=lora_extraction
-            #     agents:
-            #       queue: "default"

.buildkite/scripts/pr_test.sh

@@ -51,6 +51,7 @@ else
 fi
 
 MODAL_TEST_FILE="fastvideo/tests/modal/pr_test.py"
+MODAL_SSIM_TEST_FILE="fastvideo/tests/modal/ssim_test.py"
 
 if [ -z "${TEST_TYPE:-}" ]; then
     log "Error: TEST_TYPE environment variable is not set"
@@ -58,7 +59,11 @@ if [ -z "${TEST_TYPE:-}" ]; then
 fi
 log "Test type: $TEST_TYPE"
 
-MODAL_ENV="BUILDKITE_REPO=$BUILDKITE_REPO BUILDKITE_COMMIT=$BUILDKITE_COMMIT BUILDKITE_PULL_REQUEST=$BUILDKITE_PULL_REQUEST IMAGE_VERSION=$IMAGE_VERSION"
+EFFECTIVE_PR=${BUILDKITE_PULL_REQUEST:-false}
+if [ "$EFFECTIVE_PR" = "false" ] && [ -n "${PR_NUMBER:-}" ]; then
+    EFFECTIVE_PR=$PR_NUMBER
+fi
+MODAL_ENV="BUILDKITE_REPO=$BUILDKITE_REPO BUILDKITE_COMMIT=$BUILDKITE_COMMIT BUILDKITE_PULL_REQUEST=$EFFECTIVE_PR IMAGE_VERSION=$IMAGE_VERSION"
 
 case "$TEST_TYPE" in
     "encoder")
@@ -75,7 +80,7 @@ case "$TEST_TYPE" in
         ;;
     "ssim")
         log "Running SSIM tests..."
-        MODAL_COMMAND="$MODAL_ENV HF_API_KEY=$HF_API_KEY python3 -m modal run $MODAL_TEST_FILE::run_ssim_tests"
+        MODAL_COMMAND="$MODAL_ENV HF_API_KEY=$HF_API_KEY python3 -m modal run $MODAL_SSIM_TEST_FILE::run_ssim_tests"
         ;;
     "training")
         log "Running training tests..."
@@ -89,10 +94,6 @@ case "$TEST_TYPE" in
         log "Running training VSA tests..."
         MODAL_COMMAND="$MODAL_ENV WANDB_API_KEY=$WANDB_API_KEY python3 -m modal run $MODAL_TEST_FILE::run_training_tests_VSA"
         ;;
-    "inference_sta")
-        log "Running inference STA tests..."
-        MODAL_COMMAND="$MODAL_ENV python3 -m modal run $MODAL_TEST_FILE::run_inference_tests_STA"
-        ;;
     "kernel_tests")
         log "Running kernel tests..."
         MODAL_COMMAND="$MODAL_ENV python3 -m modal run $MODAL_TEST_FILE::run_kernel_tests"
@@ -122,6 +123,14 @@ case "$TEST_TYPE" in
         log "Running LoRA extraction tests..."
         MODAL_COMMAND="$MODAL_ENV HF_API_KEY=$HF_API_KEY python3 -m modal run $MODAL_TEST_FILE::run_lora_extraction_tests"
         ;;
+    "performance")
+        log "Running performance tests..."
+        MODAL_COMMAND="$MODAL_ENV HF_API_KEY=$HF_API_KEY python3 -m modal run $MODAL_TEST_FILE::run_performance_tests"
+        ;;
+    "api_server")
+        log "Running API server integration tests..."
+        MODAL_COMMAND="$MODAL_ENV HF_API_KEY=$HF_API_KEY python3 -m modal run $MODAL_TEST_FILE::run_api_server_tests"
+        ;;
     *)
         log "Error: Unknown test type: $TEST_TYPE"
         exit 1

.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,62 @@
+<!--
+PR TITLE: Must start with a type tag, e.g.:
+  [feat] Add new model       [bugfix] Fix VAE tiling      [refactor] Restructure pipeline
+  [perf] Optimize kernel     [ci] Update tests             [docs] Add guide
+  [misc] Cleanup configs     [new-model] Port Flux2
+
+MERGE WORKFLOW:
+  1. Ensure pre-commit passes and you have at least 1 approval
+  2. Comment /merge (or add the "ready" label) to enter the Merge Queue
+  3. Full Test Suite runs automatically on a staging branch → auto-merge on success
+
+ON-DEMAND TESTING (write access required):
+  /test full       — Full Test Suite        /test ssim        — SSIM regression
+  /test training   — Training pipeline      /test encoder     — Encoder tests
+  /test transformer — Transformer tests     /test vae         — VAE tests
+  /test kernel     — CUDA kernel tests      /test unit        — Unit tests
+  See docs/contributing/pull_requests.md for all 17 test commands
+-->
+
+## Purpose
+
+<!-- What does this PR do? Link the related issue if applicable. -->
+
+Fixes #
+
+## Changes
+
+<!-- Describe your changes concisely. What approach did you take? -->
+
+-
+
+## Test Plan
+
+<!-- How did you verify your changes? Paste exact commands and output. -->
+
+```bash
+# Commands you ran
+```
+
+## Test Results
+
+<!-- Paste test output, before/after comparisons, or SSIM scores for model changes. -->
+
+<details>
+<summary>Test output</summary>
+
+```
+# Paste output here
+```
+
+</details>
+
+## Checklist
+
+- [ ] I ran `pre-commit run --all-files` and fixed all issues
+- [ ] I added or updated tests for my changes
+- [ ] I updated documentation if needed
+- [ ] I considered GPU memory impact of my changes
+
+**For model/pipeline changes, also check:**
+- [ ] I verified SSIM regression tests pass
+- [ ] I updated the support matrix if adding a new model

.github/mergify.yml

@@ -0,0 +1,316 @@
+merge_protections:
+  - name: PR merge requirements
+    if:
+      - base = main
+    success_conditions:
+      - "title~=(?i)^\\[(feat|feature|bugfix|fix|refactor|perf|ci|doc|docs|misc|chore|kernel|new.?model)\\]"
+      - "#approved-reviews-by>=1"
+      - check-success~=pre-commit
+      - check-success=fastcheck-passed
+      - check-success=full-suite-passed
+
+pull_request_rules:
+
+  # ============================================================
+  # Type labels (from PR title prefix)
+  # ============================================================
+
+  - name: "label type: feat"
+    conditions:
+      - "title~=(?i)^\\[(feat|feature)\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: feat"]
+
+  - name: "label type: bugfix"
+    conditions:
+      - "title~=(?i)^\\[(bug)?fix\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: bugfix"]
+
+  - name: "label type: refactor"
+    conditions:
+      - "title~=(?i)^\\[refactor\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: refactor"]
+
+  - name: "label type: perf"
+    conditions:
+      - "title~=(?i)^\\[perf\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: perf"]
+
+  - name: "label type: ci"
+    conditions:
+      - "title~=(?i)^\\[ci\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: ci"]
+
+  - name: "label type: docs"
+    conditions:
+      - "title~=(?i)^\\[(doc|docs)\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: docs"]
+
+  - name: "label type: misc"
+    conditions:
+      - "title~=(?i)^\\[(misc|chore)\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: misc"]
+
+  - name: "label type: new-model"
+    conditions:
+      - "title~=(?i)^\\[new.?model\\]"
+      - -closed
+    actions:
+      label:
+        add: ["type: new-model"]
+
+  # ============================================================
+  # Scope labels (from changed files)
+  # ============================================================
+
+  - name: "label scope: training"
+    conditions:
+      - or:
+          - files~=^fastvideo/train/
+          - files~=^fastvideo/training/
+          - files~=^fastvideo/distillation/
+          - files~=^examples/train/
+          - files~=^examples/training/
+          - files~=^examples/distill/
+      - -closed
+    actions:
+      label:
+        add: ["scope: training"]
+
+  - name: "label scope: inference"
+    conditions:
+      - or:
+          - files~=^fastvideo/pipelines/basic/
+          - files~=^fastvideo/pipelines/stages/
+          - files~=^fastvideo/pipelines/samplers/
+          - files~=^fastvideo/entrypoints/
+          - files~=^fastvideo/worker/
+          - files~=^fastvideo/api/sampling_param
+          - files~=^fastvideo/configs/pipelines/
+          - files~=^examples/inference/
+      - -closed
+    actions:
+      label:
+        add: ["scope: inference"]
+
+  - name: "label scope: attention"
+    conditions:
+      - files~=^fastvideo/attention/
+      - -closed
+    actions:
+      label:
+        add: ["scope: attention"]
+
+  - name: "label scope: kernel"
+    conditions:
+      - or:
+          - files~=^fastvideo-kernel/
+          - files~=^csrc/
+      - -closed
+    actions:
+      label:
+        add: ["scope: kernel"]
+
+  - name: "label scope: data"
+    conditions:
+      - or:
+          - files~=^fastvideo/dataset/
+          - files~=^fastvideo/pipelines/preprocess/
+          - files~=^examples/preprocessing/
+      - -closed
+    actions:
+      label:
+        add: ["scope: data"]
+
+  - name: "label scope: infra"
+    conditions:
+      - or:
+          - files~=^\.github/
+          - files~=^\.buildkite/
+          - files~=^fastvideo/tests/
+          - files~=^docker/
+      - -closed
+    actions:
+      label:
+        add: ["scope: infra"]
+
+  - name: "label scope: distributed"
+    conditions:
+      - files~=^fastvideo/distributed/
+      - -closed
+    actions:
+      label:
+        add: ["scope: distributed"]
+
+  - name: "label scope: docs"
+    conditions:
+      - files~=^docs/
+      - -closed
+    actions:
+      label:
+        add: ["scope: docs"]
+
+  - name: "label scope: ui"
+    conditions:
+      - files~=^ui/
+      - -closed
+    actions:
+      label:
+        add: ["scope: ui"]
+
+  - name: "label scope: model"
+    conditions:
+      - or:
+          - files~=^fastvideo/models/
+          - files~=^fastvideo/layers/
+          - files~=^fastvideo/configs/models/
+      - -closed
+    actions:
+      label:
+        add: ["scope: model"]
+
+  # ============================================================
+  # Pre-commit failure help comment
+  # ============================================================
+
+  - name: comment on pre-commit failure
+    conditions:
+      - check-failure~=pre-commit
+      - -closed
+    actions:
+      comment:
+        message: |
+          ## Pre-commit checks failed
+
+          Hi @{{author}}, the pre-commit checks have failed. To fix them locally:
+
+          ```bash
+          # Install pre-commit if you haven't already
+          uv pip install pre-commit
+          pre-commit install
+
+          # Run all checks and auto-fix what's possible
+          pre-commit run --all-files
+          ```
+
+          Common fixes:
+          - **yapf**: `yapf -i <file>` (formatting)
+          - **ruff**: `ruff check --fix <file>` (linting)
+          - **codespell**: `codespell --write-changes <file>` (spelling)
+
+          After fixing, commit and push the changes. The checks will re-run automatically.
+
+          For future commits, `pre-commit` will run automatically on changed files before each commit.
+
+
+  # ============================================================
+  # Merge conflict detection
+  # ============================================================
+
+  - name: label conflicting PRs
+    conditions:
+      - conflict
+      - -closed
+      - label!=stale
+    actions:
+      label:
+        add: [needs-rebase]
+      comment:
+        message: |
+          This PR has merge conflicts with the base branch. Please rebase:
+
+          ```bash
+          git fetch origin main
+          git rebase origin/main
+          # Resolve any conflicts, then:
+          git push --force-with-lease
+          ```
+
+  - name: remove conflict label when resolved
+    conditions:
+      - -conflict
+      - -closed
+      - label=needs-rebase
+    actions:
+      label:
+        remove: [needs-rebase]
+
+  # ============================================================
+  # Auto-merge and auto-rebase
+  # ============================================================
+
+  - name: auto-merge when ready and all checks pass
+    conditions:
+      - label=ready
+      - "title~=(?i)^\\[(feat|feature|bugfix|fix|refactor|perf|ci|doc|docs|misc|chore|kernel|new.?model)\\]"
+      - "#approved-reviews-by>=1"
+      - check-success~=pre-commit
+      - check-success=fastcheck-passed
+      - check-success=full-suite-passed
+      - -conflict
+      - -closed
+      - -draft
+    actions:
+      merge:
+        method: squash
+
+  - name: auto-update when ready
+    conditions:
+      - label=ready
+      - "#approved-reviews-by>=1"
+      - -conflict
+      - -closed
+      - -draft
+    actions:
+      update: {}
+
+  # ============================================================
+  # PR title format help
+  # ============================================================
+
+  - name: comment on invalid PR title format
+    conditions:
+      - -closed
+      - -draft
+      - "-title~=(?i)^\\[(feat|feature|bugfix|fix|refactor|perf|ci|doc|docs|misc|chore|kernel|new.?model)\\]"
+    actions:
+      comment:
+        message: |
+          ## ⚠️ PR title format required
+
+          Your PR title must start with a type tag in brackets. Examples:
+          - `[feat] Add new model support`
+          - `[bugfix] Fix VAE tiling corruption`
+          - `[refactor] Restructure training pipeline`
+          - `[perf] Optimize attention kernel`
+          - `[ci] Update test infrastructure`
+          - `[docs] Add inference guide`
+          - `[misc] Clean up configs`
+          - `[new-model] Port Flux2 to FastVideo`
+
+          Valid tags: `feat`, `feature`, `bugfix`, `fix`, `refactor`, `perf`, `ci`, `doc`, `docs`, `misc`, `chore`, `kernel`, `new-model`
+
+          Please update your PR title and the merge protection check will pass automatically.
+
+merge_protections_settings:
+  reporting_method: check-runs

.github/scripts/runpod_api.py

@@ -1,249 +0,0 @@
-import argparse
-import json
-import os
-import subprocess
-import sys
-import time
-
-import requests
-
-
-def parse_arguments():
-    """Parse command line arguments"""
-    parser = argparse.ArgumentParser(description='Run tests on RunPod GPU')
-    parser.add_argument('--gpu-type', type=str, help='GPU type to use')
-    parser.add_argument('--gpu-count',
-                        type=int,
-                        help='Number of GPUs to use',
-                        default=1)
-    parser.add_argument('--test-command', type=str, help='Test command to run')
-    parser.add_argument('--disk-size',
-                        type=int,
-                        default=20,
-                        help='Container disk size in GB (default: 20)')
-    parser.add_argument('--volume-size',
-                        type=int,
-                        default=20,
-                        help='Persistent volume size in GB (default: 20)')
-    parser.add_argument(
-        '--image',
-        type=str,
-        required=True,
-        help='Docker image to use')
-    return parser.parse_args()
-
-
-args = parse_arguments()
-API_KEY = os.environ['RUNPOD_API_KEY']
-RUN_ID = os.environ['GITHUB_RUN_ID']
-JOB_ID = os.environ['JOB_ID']
-PODS_API = "https://rest.runpod.io/v1/pods"
-HEADERS = {
-    "Content-Type": "application/json",
-    "Authorization": f"Bearer {API_KEY}"
-}
-
-
-def create_pod():
-    """Create a RunPod instance"""
-    # Ensure image name is lowercase (Docker requirement)
-    image_name = args.image.lower()
-    print(f"Using specified image: {image_name}")
-
-    docker_start_cmd = [
-        "bash", 
-        "-c", 
-        "apt update;DEBIAN_FRONTEND=noninteractive apt-get install openssh-server -y;mkdir -p ~/.ssh;cd $_;chmod 700 ~/.ssh;echo \"$PUBLIC_KEY\" >> authorized_keys;chmod 700 authorized_keys;service ssh start;sleep infinity"
-    ]
-    
-    print(f"Creating RunPod instance with GPU: {args.gpu_type}...")
-    payload = {
-        "name": f"fastvideo-{JOB_ID}-{RUN_ID}",
-        "containerDiskInGb": args.disk_size,
-        "volumeInGb": args.volume_size,
-        "gpuTypeIds": [args.gpu_type],
-        "gpuCount": args.gpu_count,
-        "imageName": image_name,
-        "allowedCudaVersions": ["12.4"],
-        "dockerStartCmd": docker_start_cmd
-    }
-
-    response = requests.post(PODS_API, headers=HEADERS, json=payload)
-    response_data = response.json()
-    print(f"Response: {json.dumps(response_data, indent=2)}")
-
-    return response_data["id"]
-
-
-def wait_for_pod(pod_id):
-    """Wait for pod to be in RUNNING state and fully ready with SSH access"""
-    print("Waiting for RunPod to be ready...")
-
-    # First wait for RUNNING status
-    max_attempts = 10
-    attempts = 0
-    while attempts < max_attempts:
-        response = requests.get(f"{PODS_API}/{pod_id}", headers=HEADERS)
-        pod_data = response.json()
-        status = pod_data["desiredStatus"]
-
-        if status == "RUNNING":
-            print("RunPod is running! Now waiting for ports to be assigned...")
-            break
-
-        print(
-            f"Current status: {status}, waiting... (attempt {attempts+1}/{max_attempts})"
-        )
-        time.sleep(2)
-        attempts += 1
-
-    if attempts >= max_attempts:
-        raise TimeoutError(
-            "Timed out waiting for RunPod to reach RUNNING state")
-
-    # Wait for ports to be assigned
-    max_attempts = 50
-    attempts = 0
-    while attempts < max_attempts:
-        response = requests.get(f"{PODS_API}/{pod_id}", headers=HEADERS)
-        pod_data = response.json()
-        port_mappings = pod_data.get("portMappings")
-
-        if (port_mappings is not None and "22" in port_mappings
-                and pod_data.get("publicIp", "") != ""):
-            print("RunPod is ready with SSH access!")
-            print(f"SSH IP: {pod_data['publicIp']}")
-            print(f"SSH Port: {port_mappings['22']}")
-            break
-
-        print(
-            f"Waiting for SSH port and public IP to be available... (attempt {attempts+1}/{max_attempts})"
-        )
-        time.sleep(20)
-        attempts += 1
-
-    if attempts >= max_attempts:
-        raise TimeoutError("Timed out waiting for RunPod SSH access")
-
-
-def execute_command(pod_id):
-    """Execute command on the pod via SSH using system SSH client"""
-    print(f"Running command: {args.test_command}")
-
-    response = requests.get(f"{PODS_API}/{pod_id}", headers=HEADERS)
-    pod_data = response.json()
-    ssh_ip = pod_data["publicIp"]
-    ssh_port = pod_data["portMappings"]["22"]
-
-    # Copy the repository to the pod using scp
-    repo_dir = os.path.abspath(os.getcwd())
-    repo_name = os.path.basename(repo_dir)
-
-    print(f"Copying repository from {repo_dir} to RunPod...")
-
-    tar_command = [
-        "tar", "-czf", "/tmp/repo.tar.gz", "-C",
-        os.path.dirname(repo_dir), repo_name
-    ]
-    subprocess.run(tar_command, check=True)
-
-    # Copy the tarball to the pod
-    scp_command = [
-        "scp", "-o", "StrictHostKeyChecking=no", "-o",
-        "UserKnownHostsFile=/dev/null", "-o", "ServerAliveInterval=60", "-o",
-        "ServerAliveCountMax=10", "-P",
-        str(ssh_port), "/tmp/repo.tar.gz", f"root@{ssh_ip}:/tmp/"
-    ]
-    subprocess.run(scp_command, check=True)
-
-    # For custom image, we can use the pre-configured environment
-    setup_steps = [
-        "tar -xzf /tmp/repo.tar.gz --no-same-owner -C /workspace/",
-        f"cd /workspace/{repo_name}",
-        "source $HOME/.local/bin/env && source /opt/venv/bin/activate",
-        args.test_command
-    ]
-    
-    remote_command = " && ".join(setup_steps)
-
-    ssh_command = [
-        "ssh", "-o", "StrictHostKeyChecking=no", "-o",
-        "UserKnownHostsFile=/dev/null", "-o", "ServerAliveInterval=60", "-o",
-        "ServerAliveCountMax=10", "-p",
-        str(ssh_port), f"root@{ssh_ip}", remote_command
-    ]
-
-    print(f"Connecting to {ssh_ip}:{ssh_port}...")
-
-    try:
-        process = subprocess.Popen(ssh_command,
-                                   stdout=subprocess.PIPE,
-                                   stderr=subprocess.STDOUT,
-                                   universal_newlines=True,
-                                   bufsize=0)
-
-        stdout_lines = []
-
-        print("Command output:")
-
-        for line in iter(process.stdout.readline, ''):
-            print(line.strip())
-            stdout_lines.append(line)
-
-        process.wait()
-
-        return_code = process.returncode
-        success = return_code == 0
-
-        stdout_str = "".join(stdout_lines)
-
-        if success:
-            print("Command executed successfully")
-        else:
-            print(f"Command failed with exit code {return_code}")
-
-        result = {
-            "success": success,
-            "return_code": return_code,
-            "stdout": stdout_str,
-            "stderr": ""
-        }
-        return result
-
-    except Exception as e:
-        print(f"Error executing SSH command: {str(e)}")
-        result = {"success": False, "error": str(e), "stdout": "", "stderr": ""}
-        return result
-
-
-def terminate_pod(pod_id):
-    """Terminate the pod"""
-    print("Terminating RunPod...")
-    requests.delete(f"{PODS_API}/{pod_id}", headers=HEADERS)
-    print(f"Terminated pod {pod_id}")
-
-
-def main():
-    pod_id = None
-    try:
-        pod_id = create_pod()
-        wait_for_pod(pod_id)
-        result = execute_command(pod_id)
-
-        if result.get("error") is not None:
-            print(f"Error executing command: {result['error']}")
-            sys.exit(1)
-
-        if not result.get("success", False):
-            print(
-                "Tests failed - check the output above for details on which tests failed"
-            )
-            sys.exit(1)
-
-    finally:
-        if pod_id:
-            terminate_pod(pod_id)
-
-
-if __name__ == "__main__":
-    main()

.github/scripts/runpod_cleanup.py

@@ -1,90 +0,0 @@
-import json
-import os
-import sys
-import uuid
-
-import requests
-
-API_KEY = os.environ['RUNPOD_API_KEY']
-RUN_ID = os.environ.get('GITHUB_RUN_ID', str(uuid.uuid4()))
-PODS_API = "https://rest.runpod.io/v1/pods"
-HEADERS = {
-    "Content-Type": "application/json",
-    "Authorization": f"Bearer {API_KEY}"
-}
-
-
-def get_job_ids():
-    """Parse job IDs from environment variable"""
-    job_ids_str = os.environ.get('JOB_IDS')
-    try:
-        job_ids = json.loads(job_ids_str)
-        if not isinstance(job_ids, list):
-            print("Error: JOB_IDS is not a list.")
-            sys.exit(1)
-        return job_ids
-    except json.JSONDecodeError as e:
-        print(f"Error parsing JOB_IDS: {e}")
-        sys.exit(1)
-
-
-def cleanup_pods():
-    """Find and terminate RunPod instances"""
-    print(f"Run ID: {RUN_ID}")
-
-    single_job_id = os.environ.get('JOB_ID')
-    
-    if single_job_id:
-        job_ids = [single_job_id]
-        print(f"Job ID: {single_job_id}")
-    else:
-        job_ids = get_job_ids()
-        print(f"Job IDs: {job_ids}")
-
-    # Get all pods associated with RunPod API_KEY
-    try:
-        response = requests.get(PODS_API, headers=HEADERS)
-        response.raise_for_status()
-        pods = response.json()
-    except requests.exceptions.RequestException as e:
-        print(f"Error getting pods: {e}")
-        sys.exit(1)
-
-    # Find and terminate pods created by this workflow run
-    terminated_pods = []
-    for pod in pods:
-        pod_name = pod.get("name", "")
-        pod_id = pod.get("id")
-
-        # Check if this pod was created by one of our jobs
-        if any(f"{job_id}-{RUN_ID}" in pod_name for job_id in job_ids):
-            print(f"Found pod: {pod_id} ({pod_name})")
-            try:
-                print(f"Terminating pod {pod_id}...")
-                term_response = requests.delete(f"{PODS_API}/{pod_id}",
-                                                headers=HEADERS)
-                term_response.raise_for_status()
-                terminated_pods.append(pod_id)
-                print(f"Successfully terminated pod {pod_id}")
-            except requests.exceptions.RequestException as e:
-                print(f"Error terminating pod {pod_id}: {e}")
-                sys.exit(1)
-    
-    if terminated_pods:
-        if single_job_id:
-            print(f"Terminated pod: {terminated_pods[0]}")
-        else:
-            print(f"Terminated {len(terminated_pods)} pods: {terminated_pods}")
-    else:
-        if single_job_id:
-            print(f"No pod found matching pattern: {single_job_id}-{RUN_ID}")
-        else:
-            print("No pods found to terminate.")
-
-
-def main():
-    cleanup_pods()
-
-
-if __name__ == "__main__":
-    main()

.github/workflows/ci-aggregate-status.yml

@@ -0,0 +1,80 @@
+name: Aggregate Test Status
+
+on:
+  status:
+
+permissions:
+  statuses: write
+
+jobs:
+  aggregate:
+    if: >-
+      github.event.context == 'direct-test-completed'
+      && github.event.state == 'success'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check and update aggregate status
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            const sha = context.payload.sha;
+
+            const { data } = await github.rest.repos.getCombinedStatusForRef({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              ref: sha,
+              per_page: 100,
+            });
+
+            const bkStatuses = data.statuses.filter(
+              s => s.context.startsWith('buildkite/ci/')
+            );
+
+            const FASTCHECK_PREFIX = 'buildkite/ci/microscope-';
+            const FULL_SUITE_PREFIXES = [
+              'buildkite/ci/test-tube-',
+              'buildkite/ci/bar-chart-',
+            ];
+
+            const fastcheck = bkStatuses.filter(
+              s => s.context.startsWith(FASTCHECK_PREFIX)
+            );
+            const fullSuite = bkStatuses.filter(
+              s => FULL_SUITE_PREFIXES.some(p => s.context.startsWith(p))
+            );
+
+            if (
+              fastcheck.length > 0
+              && fastcheck.every(s => s.state === 'success')
+            ) {
+              core.info(
+                `All ${fastcheck.length} fastcheck tests passed — updating fastcheck-passed`
+              );
+              await github.rest.repos.createCommitStatus({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                sha,
+                state: 'success',
+                context: 'fastcheck-passed',
+                description:
+                  `All ${fastcheck.length} fastcheck tests passed`,
+              });
+            }
+
+            if (
+              fullSuite.length > 0
+              && fullSuite.every(s => s.state === 'success')
+            ) {
+              core.info(
+                `All ${fullSuite.length} full suite tests passed — updating full-suite-passed`
+              );
+              await github.rest.repos.createCommitStatus({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                sha,
+                state: 'success',
+                context: 'full-suite-passed',
+                description:
+                  `All ${fullSuite.length} full suite tests passed`,
+              });
+            }

.github/workflows/ci-precommit.yml

@@ -0,0 +1,32 @@
+name: pre-commit
+
+on:
+  pull_request:
+    branches: [main]
+  workflow_call:
+    inputs:
+      ref:
+        description: 'Git ref to checkout (defaults to github.ref)'
+        required: false
+        type: string
+
+permissions:
+  contents: read
+
+jobs:
+  pre-commit:
+    if: github.event_name == 'workflow_call' || github.event.pull_request.draft != true
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v4
+      with:
+        ref: ${{ inputs.ref || '' }}
+    - uses: actions/setup-python@v5
+      with:
+        python-version: "3.12"
+    - run: echo "::add-matcher::.github/workflows/matchers/actionlint.json"
+    - run: echo "::add-matcher::.github/workflows/matchers/mypy.json"
+    - run: echo "::add-matcher::.github/workflows/matchers/ruff.json"
+    - uses: pre-commit/action@v3.0.1
+      with:
+        extra_args: --all-files --hook-stage manual

.github/workflows/ci-slash-commands.yml

@@ -0,0 +1,271 @@
+name: Slash Commands
+
+on:
+  issue_comment:
+    types: [created]
+
+permissions:
+  contents: read
+  pull-requests: write
+  statuses: write
+
+jobs:
+  handle-merge:
+    if: >-
+      github.event.issue.pull_request != null
+      && startsWith(github.event.comment.body, '/merge')
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check write permission
+        id: perm
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            const { data: perm } = await github.rest.repos.getCollaboratorPermissionLevel({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              username: context.payload.comment.user.login,
+            });
+            const hasWrite = ['admin', 'write'].includes(perm.permission);
+            if (!hasWrite) {
+              core.setFailed(`User ${context.payload.comment.user.login} lacks write permission (has: ${perm.permission}).`);
+            }
+            core.setOutput('has_write', String(hasWrite));
+
+      - name: Add ready label and react
+        id: label
+        if: steps.perm.outputs.has_write == 'true'
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            const owner = context.repo.owner;
+            const repo = context.repo.repo;
+            const prNumber = context.payload.issue.number;
+            try { await github.rest.issues.removeLabel({ owner, repo, issue_number: prNumber, name: 'ready' }); } catch {}
+            await github.rest.issues.addLabels({ owner, repo, issue_number: prNumber, labels: ['ready'] });
+            await github.rest.reactions.createForIssueComment({
+              owner, repo,
+              comment_id: context.payload.comment.id,
+              content: 'rocket',
+            });
+            const { data: pr } = await github.rest.pulls.get({ owner, repo, pull_number: prNumber });
+            core.setOutput('pr_sha', pr.head.sha);
+            core.setOutput('pr_branch', pr.head.ref);
+            core.setOutput('pr_number', String(prNumber));
+
+      - name: Trigger Full Suite
+        if: steps.perm.outputs.has_write == 'true'
+        env:
+          BUILDKITE_API_TOKEN: ${{ secrets.BUILDKITE_API_TOKEN }}
+          PR_SHA: ${{ steps.label.outputs.pr_sha }}
+          PR_BRANCH: ${{ steps.label.outputs.pr_branch }}
+          PR_NUMBER: ${{ steps.label.outputs.pr_number }}
+          BK_ORG: ${{ vars.BUILDKITE_ORG_SLUG }}
+          BK_PIPELINE: ${{ vars.BUILDKITE_PIPELINE_SLUG }}
+        run: |
+          curl -sS --fail-with-body -X POST \
+            "https://api.buildkite.com/v2/organizations/${BK_ORG}/pipelines/${BK_PIPELINE}/builds" \
+            -H "Authorization: Bearer $BUILDKITE_API_TOKEN" \
+            -H "Content-Type: application/json" \
+            --data-raw "$(jq -n \
+              --arg commit "$PR_SHA" \
+              --arg branch "$PR_BRANCH" \
+              --arg message "Full Suite for PR #${PR_NUMBER} (via /merge)" \
+              --argjson pr_id "$PR_NUMBER" \
+              '{
+                commit: $commit,
+                branch: $branch,
+                message: $message,
+                ignore_pipeline_branch_filters: true,
+                pull_request_id: $pr_id,
+                pull_request_base_branch: "main",
+                env: {
+                  TEST_SCOPE: "full",
+                  FULL_SUITE: "true",
+                  PR_NUMBER: ($pr_id | tostring)
+                }
+              }')"
+
+  parse-command:
+    if: >-
+      github.event.issue.pull_request != null
+      && startsWith(github.event.comment.body, '/test')
+    runs-on: ubuntu-latest
+    outputs:
+      test_type: ${{ steps.parse.outputs.test_type }}
+      test_scope: ${{ steps.parse.outputs.test_scope }}
+      full_suite: ${{ steps.parse.outputs.full_suite }}
+      pr_sha: ${{ steps.pr.outputs.sha }}
+      pr_branch: ${{ steps.pr.outputs.branch }}
+      has_write: ${{ steps.perm.outputs.has_write }}
+    steps:
+      - name: Check write permission
+        id: perm
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            const { data: perm } = await github.rest.repos.getCollaboratorPermissionLevel({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              username: context.payload.comment.user.login,
+            });
+            const hasWrite = ['admin', 'write'].includes(perm.permission);
+            core.setOutput('has_write', String(hasWrite));
+            if (!hasWrite) {
+              core.info(`User ${context.payload.comment.user.login} lacks write permission — ignoring.`);
+            }
+
+      - name: Parse /test command
+        id: parse
+        if: steps.perm.outputs.has_write == 'true'
+        shell: bash
+        env:
+          COMMENT: ${{ github.event.comment.body }}
+        run: |
+          set -euo pipefail
+          TEST_NAME=$(echo "$COMMENT" | grep -oP '(?<=/test\s)\S+' | head -1 || true)
+
+          VALID="encoder vae transformer kernel unit ssim training lora-inference lora-training distillation self-forcing vsa vmoba performance api full fastcheck pre-commit"
+          if [ -z "$TEST_NAME" ] || ! echo "$VALID" | grep -qw "$TEST_NAME"; then
+            echo "Unknown test: '$TEST_NAME'. Valid: $VALID"
+            exit 1
+          fi
+
+          declare -A MAP=(
+            [encoder]=encoder [vae]=vae [transformer]=transformer
+            [kernel]=kernel_tests [unit]=unit_test
+            [ssim]=ssim [training]=training
+            [lora-inference]=inference_lora [lora-training]=training_lora
+            [distillation]=distillation_dmd [self-forcing]=self_forcing
+            [vsa]=training_vsa [vmoba]=inference_vmoba
+            [performance]=performance [api]=api_server
+          )
+
+          if [ "$TEST_NAME" = "full" ]; then
+            {
+              echo "test_type=all"
+              echo "test_scope=full"
+              echo "full_suite=true"
+            } >> "$GITHUB_OUTPUT"
+          elif [ "$TEST_NAME" = "fastcheck" ]; then
+            {
+              echo "test_type=fastcheck"
+              echo "test_scope=fastcheck"
+              echo "full_suite=false"
+            } >> "$GITHUB_OUTPUT"
+          elif [ "$TEST_NAME" = "pre-commit" ]; then
+            {
+              echo "test_type="
+              echo "test_scope=precommit"
+              echo "full_suite=false"
+            } >> "$GITHUB_OUTPUT"
+          else
+            {
+              echo "test_type=${MAP[$TEST_NAME]}"
+              echo "test_scope=direct"
+              echo "full_suite=false"
+            } >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Get PR details
+        id: pr
+        if: steps.perm.outputs.has_write == 'true'
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            const { data: pr } = await github.rest.pulls.get({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              pull_number: context.payload.issue.number,
+            });
+            core.setOutput('sha', pr.head.sha);
+            core.setOutput('branch', pr.head.ref);
+
+      - name: React to comment
+        if: steps.perm.outputs.has_write == 'true'
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            await github.rest.reactions.createForIssueComment({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              comment_id: context.payload.comment.id,
+              content: 'rocket',
+            });
+
+  pre-commit:
+    needs: parse-command
+    if: >-
+      needs.parse-command.outputs.has_write == 'true'
+      && needs.parse-command.outputs.test_scope == 'precommit'
+    uses: ./.github/workflows/ci-precommit.yml
+    with:
+      ref: refs/pull/${{ github.event.issue.number }}/merge
+
+  post-precommit-status:
+    needs: [parse-command, pre-commit]
+    if: always() && needs.parse-command.outputs.test_scope == 'precommit'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        env:
+          PR_SHA: ${{ needs.parse-command.outputs.pr_sha }}
+          RESULT: ${{ needs.pre-commit.result }}
+        with:
+          script: |
+            const state = process.env.RESULT === 'success' ? 'success' : 'failure';
+            await github.rest.repos.createCommitStatus({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              sha: process.env.PR_SHA,
+              state,
+              context: 'pre-commit',
+              description: `Triggered via /test pre-commit (${state})`,
+            });
+
+  trigger-buildkite:
+    needs: parse-command
+    if: >-
+      needs.parse-command.outputs.has_write == 'true'
+      && needs.parse-command.outputs.test_type != ''
+    runs-on: ubuntu-latest
+    steps:
+      - name: Trigger Buildkite
+        env:
+          BUILDKITE_API_TOKEN: ${{ secrets.BUILDKITE_API_TOKEN }}
+          PR_SHA: ${{ needs.parse-command.outputs.pr_sha }}
+          PR_BRANCH: ${{ needs.parse-command.outputs.pr_branch }}
+          PR_NUMBER: ${{ github.event.issue.number }}
+          TEST_SCOPE: ${{ needs.parse-command.outputs.test_scope }}
+          FULL_SUITE: ${{ needs.parse-command.outputs.full_suite }}
+          TEST_TYPE: ${{ needs.parse-command.outputs.test_type }}
+          BK_ORG: ${{ vars.BUILDKITE_ORG_SLUG }}
+          BK_PIPELINE: ${{ vars.BUILDKITE_PIPELINE_SLUG }}
+        run: |
+          curl -sS --fail-with-body -X POST \
+            "https://api.buildkite.com/v2/organizations/${BK_ORG}/pipelines/${BK_PIPELINE}/builds" \
+            -H "Authorization: Bearer $BUILDKITE_API_TOKEN" \
+            -H "Content-Type: application/json" \
+            --data-raw "$(jq -n \
+              --arg commit "$PR_SHA" \
+              --arg branch "$PR_BRANCH" \
+              --arg message "/test ${TEST_TYPE} on PR #${PR_NUMBER}" \
+              --argjson pr_id "$PR_NUMBER" \
+              --arg test_scope "$TEST_SCOPE" \
+              --arg full_suite "$FULL_SUITE" \
+              --arg test_type "$TEST_TYPE" \
+              --arg pr_number "$PR_NUMBER" \
+              '{
+                commit: $commit,
+                branch: $branch,
+                message: $message,
+                ignore_pipeline_branch_filters: true,
+                pull_request_id: $pr_id,
+                pull_request_base_branch: "main",
+                env: {
+                  TEST_SCOPE: $test_scope,
+                  FULL_SUITE: $full_suite,
+                  TEST_TYPE: $test_type,
+                  PR_NUMBER: $pr_number
+                }
+              }')"

.github/workflows/ci-trigger-full-suite.yml

@@ -0,0 +1,83 @@
+name: Trigger Full Suite
+
+on:
+  pull_request_target:
+    types: [labeled, synchronize]
+
+permissions:
+  contents: read
+  pull-requests: read
+
+concurrency:
+  group: full-suite-${{ github.event.pull_request.number }}
+  cancel-in-progress: false
+
+jobs:
+  trigger:
+    if: >-
+      (github.event.action == 'labeled' && github.event.label.name == 'ready')
+      || github.event.action == 'synchronize'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check ready label
+        id: check
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            const { data: pr } = await github.rest.pulls.get({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              pull_number: context.payload.pull_request.number,
+            });
+            const hasReady = pr.labels.some(l => l.name === 'ready');
+            core.setOutput('has_ready', String(hasReady));
+            if (!hasReady) core.info('No ready label — skipping Full Suite trigger.');
+
+      - name: Cancel previous Buildkite builds
+        if: steps.check.outputs.has_ready == 'true'
+        env:
+          BUILDKITE_API_TOKEN: ${{ secrets.BUILDKITE_API_TOKEN }}
+          PR_BRANCH: ${{ github.event.pull_request.head.ref }}
+        run: |
+          # Find running builds for this branch with TEST_SCOPE=full and cancel them
+          builds=$(curl -sS -H "Authorization: Bearer $BUILDKITE_API_TOKEN" \
+            "https://api.buildkite.com/v2/organizations/${{ vars.BUILDKITE_ORG_SLUG }}/pipelines/${{ vars.BUILDKITE_PIPELINE_SLUG }}/builds?branch=${PR_BRANCH}&state=running,scheduled" \
+            | jq -r '.[] | select(try (.env.TEST_SCOPE == "full") catch false) | .number')
+          for build_num in $builds; do
+            echo "Cancelling Buildkite build #$build_num"
+            curl -sS -X PUT -H "Authorization: Bearer $BUILDKITE_API_TOKEN" \
+              "https://api.buildkite.com/v2/organizations/${{ vars.BUILDKITE_ORG_SLUG }}/pipelines/${{ vars.BUILDKITE_PIPELINE_SLUG }}/builds/${build_num}/cancel"
+          done
+
+      - name: Trigger Buildkite Full Suite
+        if: steps.check.outputs.has_ready == 'true'
+        env:
+          BUILDKITE_API_TOKEN: ${{ secrets.BUILDKITE_API_TOKEN }}
+          PR_SHA: ${{ github.event.pull_request.head.sha }}
+          PR_BRANCH: ${{ github.event.pull_request.head.ref }}
+          PR_NUMBER: ${{ github.event.pull_request.number }}
+          BK_ORG: ${{ vars.BUILDKITE_ORG_SLUG }}
+          BK_PIPELINE: ${{ vars.BUILDKITE_PIPELINE_SLUG }}
+        run: |
+          curl -sS --fail-with-body -X POST \
+            "https://api.buildkite.com/v2/organizations/${BK_ORG}/pipelines/${BK_PIPELINE}/builds" \
+            -H "Authorization: Bearer $BUILDKITE_API_TOKEN" \
+            -H "Content-Type: application/json" \
+            --data-raw "$(jq -n \
+              --arg commit "$PR_SHA" \
+              --arg branch "$PR_BRANCH" \
+              --arg message "Full Suite for PR #${PR_NUMBER}" \
+              --argjson pr_id "$PR_NUMBER" \
+              '{
+                commit: $commit,
+                branch: $branch,
+                message: $message,
+                ignore_pipeline_branch_filters: true,
+                pull_request_id: $pr_id,
+                pull_request_base_branch: "main",
+                env: {
+                  TEST_SCOPE: "full",
+                  FULL_SUITE: "true",
+                  PR_NUMBER: ($pr_id | tostring)
+                }
+              }')"

.github/workflows/community-issue-labeler.yml

@@ -0,0 +1,65 @@
+name: Auto-Label Issues
+
+on:
+  issues:
+    types: [opened, edited]
+
+permissions:
+  issues: write
+
+jobs:
+  label-issues:
+    if: github.repository == 'hao-ai-lab/FastVideo'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Label by keywords
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        with:
+          script: |
+            const title = context.payload.issue.title.toLowerCase();
+            const body = (context.payload.issue.body || '').toLowerCase();
+            const text = title + ' ' + body;
+            const labels = [];
+
+            const rules = [
+              // scope labels (shared with PR labeling via Mergify)
+              // Mapping: label → repo directories
+              //   scope: training    → fastvideo/train/, fastvideo/training/, fastvideo/distillation/
+              //   scope: inference   → fastvideo/pipelines/, fastvideo/entrypoints/, fastvideo/worker/
+              //   scope: attention   → fastvideo/attention/
+              //   scope: kernel      → fastvideo-kernel/, csrc/
+              //   scope: model       → fastvideo/models/, fastvideo/layers/, fastvideo/configs/models/
+              //   scope: data        → fastvideo/dataset/, fastvideo/pipelines/preprocess/
+              //   scope: distributed → fastvideo/distributed/
+              //   scope: docs        → docs/
+              { keywords: ['training', 'finetune', 'fine-tune', 'lora', 'fsdp', 'distill'], label: 'scope: training' },
+              { keywords: ['inference', 'generate', 'pipeline', 'slow', 'latency'], label: 'scope: inference' },
+              { keywords: ['attention', 'vsa', 'flash', 'sta', 'vmoba', 'sparse attn'], label: 'scope: attention' },
+              { keywords: ['kernel', 'csrc', 'cuda kernel', 'thunderkittens'], label: 'scope: kernel' },
+              { keywords: ['wan', 'hunyuan', 'mochi', 'ltx', 'cogvideo', 'flux', 'sd3', 'cosmos'], label: 'scope: model' },
+              { keywords: ['dataset', 'dataloader', 'preprocessing', 'preprocess'], label: 'scope: data' },
+              { keywords: ['distributed', 'sequence parallel', 'fsdp', 'tensor parallel', 'multi-node', 'multi-gpu'], label: 'scope: distributed' },
+              { keywords: ['docs', 'documentation', 'tutorial', 'example'], label: 'scope: docs' },
+              // issue-only labels (cross-module, no single repo directory)
+              { keywords: ['install', 'setup', 'pip', 'cuda', 'uv ', 'import error', 'modulenotfound'], label: 'installation' },
+              { keywords: ['memory', 'oom', 'out of memory', 'gpu memory', 'vram'], label: 'performance' },
+              { keywords: ['windows', 'macos', 'mac os', 'apple', 'mps', 'rocm', 'amd', 'npu'], label: 'platform' },
+            ];
+
+            for (const rule of rules) {
+              if (rule.keywords.some(kw => text.includes(kw))) {
+                labels.push(rule.label);
+              }
+            }
+
+            if (labels.length > 0) {
+              await github.rest.issues.addLabels({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                issue_number: context.payload.issue.number,
+                labels: labels,
+              });
+              console.log(`Added labels: ${labels.join(', ')}`);
+            } else {
+              console.log('No keyword matches found');
+            }

.github/workflows/community-stale.yml

@@ -0,0 +1,51 @@
+name: Close Stale Issues and PRs
+
+on:
+  schedule:
+    # Daily at 1:30 AM UTC
+    - cron: '30 1 * * *'
+
+jobs:
+  stale:
+    if: github.repository == 'hao-ai-lab/FastVideo'
+    permissions:
+      issues: write
+      pull-requests: write
+      actions: write
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/stale@997185467fa4f803885201cee163a9f38240193d # v10.1.1
+        with:
+          operations-per-run: 500
+
+          exempt-draft-pr: true
+          exempt-issue-labels: 'keep-open,pinned,security,Bug,RFC'
+          exempt-pr-labels: 'keep-open,pinned'
+
+          labels-to-add-when-unstale: 'unstale'
+          labels-to-remove-when-stale: 'unstale'
+
+          days-before-issue-stale: 90
+          days-before-issue-close: 30
+          stale-issue-label: 'stale'
+          stale-issue-message: >
+            This issue has been automatically marked as stale because it has not
+            had any activity within 90 days. It will be automatically closed if
+            no further activity occurs within 30 days. Leave a comment if you
+            feel this issue should remain open. Thank you!
+          close-issue-message: >
+            This issue has been automatically closed due to inactivity. Please
+            feel free to reopen if you feel it is still relevant. Thank you!
+
+          days-before-pr-stale: 60
+          days-before-pr-close: 14
+          stale-pr-label: 'stale'
+          stale-pr-message: >
+            This pull request has been automatically marked as stale because it
+            has not had any activity within 60 days. It will be automatically
+            closed if no further activity occurs within 14 days. Leave a comment
+            if you feel this pull request should remain open. Thank you!
+          close-pr-message: >
+            This pull request has been automatically closed due to inactivity.
+            Please feel free to reopen if you intend to continue working on it.
+            Thank you!

.github/workflows/community-welcome.yml

@@ -0,0 +1,56 @@
+name: Welcome First-Time Contributors
+
+on:
+  issues:
+    types: [opened]
+  pull_request_target:
+    types: [opened]
+
+permissions:
+  issues: write
+  pull-requests: write
+
+jobs:
+  welcome:
+    if: github.repository == 'hao-ai-lab/FastVideo'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/first-interaction@34f15e814fe48ac9312ccf29db4e74fa767cbab7 # v1.3.0
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+          issue-message: |
+            Welcome to FastVideo! Thanks for opening your first issue.
+
+            To help us investigate, please include:
+            - **FastVideo version**: `pip show fastvideo`
+            - **GPU**: `nvidia-smi` output (GPU model, driver, CUDA version)
+            - **Python version**: `python --version`
+            - **OS**: e.g., Ubuntu 22.04
+
+            If this is a bug, a minimal reproduction script helps us fix it faster.
+
+            Useful links:
+            - [Documentation](https://hao-ai-lab.github.io/FastVideo)
+            - [Contributing Guide](https://hao-ai-lab.github.io/FastVideo/contributing/overview/)
+            - [Slack](https://join.slack.com/t/fastvideo/shared_invite/zt-3f4lao1uq-u~Ipx6Lt4J27AlD2y~IdLQ)
+          pr-message: |
+            Welcome to FastVideo! Thanks for your first pull request.
+
+            **How our CI works:**
+
+            PRs run a two-tier CI system:
+            1. **Pre-commit** — formatting (yapf), linting (ruff), type checking (mypy). Runs immediately on every PR.
+            2. **Fastcheck** — core GPU tests (encoders, VAEs, transformers, kernels, unit tests). Runs automatically via Buildkite on relevant file changes (~10-15 min).
+            3. **Full Suite** — integration tests, training pipelines, SSIM regression. Runs only when a reviewer adds the `ready` label.
+
+            **Before your PR is reviewed:**
+            - [ ] `pre-commit run --all-files` passes locally
+            - [ ] You've added or updated tests for your changes
+            - [ ] The PR description explains what and why
+
+            If pre-commit fails, a bot comment will explain how to fix it. Fastcheck and Full Suite results appear in the Checks section below.
+
+            **Useful links:**
+            - [Contributing Guide](https://hao-ai-lab.github.io/FastVideo/contributing/overview/)
+            - [Development Roadmap](https://github.com/hao-ai-lab/FastVideo/issues/899)
+            - [Slack](https://join.slack.com/t/fastvideo/shared_invite/zt-3f4lao1uq-u~Ipx6Lt4J27AlD2y~IdLQ)

.github/workflows/infra-build-image.yml

@@ -32,7 +32,7 @@ permissions:
 jobs:
   build-python-3-10:
     if: ${{ github.event.inputs.python_3_10 == 'true' }}
-    uses: ./.github/workflows/build-image-template.yml
+    uses: ./.github/workflows/_template-build-image.yml
     with:
       python_version: '3.10'
       dockerfile_path: docker/Dockerfile.python3.10
@@ -41,7 +41,7 @@ jobs:
 
   build-python-3-11:
     if: ${{ github.event.inputs.python_3_11 == 'true' }}
-    uses: ./.github/workflows/build-image-template.yml
+    uses: ./.github/workflows/_template-build-image.yml
     with:
       python_version: '3.11'
       dockerfile_path: docker/Dockerfile.python3.11
@@ -50,7 +50,7 @@ jobs:
 
   build-python-3-12:
     if: ${{ github.event.inputs.python_3_12 == 'true' }}
-    uses: ./.github/workflows/build-image-template.yml
+    uses: ./.github/workflows/_template-build-image.yml
     with:
       python_version: '3.12'
       dockerfile_path: docker/Dockerfile.python3.12
@@ -59,9 +59,9 @@ jobs:
 
   build-python-3-12-cuda-12-9:
     if: ${{ github.event.inputs.python_3_12_cuda_12_9 == 'true' }}
-    uses: ./.github/workflows/build-image-template.yml
+    uses: ./.github/workflows/_template-build-image.yml
     with:
       python_version: '3.12'
       dockerfile_path: docker/Dockerfile.python3.12.cuda12.9.1
       tag_suffix: py3.12-cuda12.9.1
-    secrets: inherit
+    secrets: inherit

.github/workflows/infra-docs.yml

@@ -7,14 +7,14 @@ on:
       - 'docs/**'
       - 'mkdocs.yml'
       - 'requirements-mkdocs.txt'
-      - '.github/workflows/docs.yml'
+      - '.github/workflows/infra-docs.yml'
   pull_request:
     branches: [ main ]
     paths:
       - 'docs/**'
       - 'mkdocs.yml'
       - 'requirements-mkdocs.txt'
-      - '.github/workflows/docs.yml'
+      - '.github/workflows/infra-docs.yml'
 
 permissions:
   contents: read
@@ -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/matchers/ruff.json

@@ -0,0 +1,17 @@
+{
+  "problemMatcher": [
+    {
+      "owner": "ruff",
+      "pattern": [
+        {
+          "regexp": "^(.+):(\\d+):(\\d+): (\\w+) (.+)$",
+          "file": 1,
+          "line": 2,
+          "column": 3,
+          "code": 4,
+          "message": 5
+        }
+      ]
+    }
+  ]
+}

.github/workflows/pr-test.yml

@@ -1,401 +0,0 @@
-name: PR Test
-
-on:
-  push:
-    branches: [main]
-    paths:
-      - "fastvideo/**/*.py"
-      - ".github/workflows/pr-test.yml"
-  pull_request:
-    branches: [main]
-    types: [opened, ready_for_review, synchronize, reopened]
-    paths:
-      - "fastvideo/**/*.py"
-      - ".github/workflows/pr-test.yml"
-      - "pyproject.toml"
-      - "docker/Dockerfile.python3.12"
-      - "csrc/**"
-  workflow_dispatch:
-    inputs:
-      run_encoder_test:
-        description: "Run encoder-test"
-        required: false
-        default: false
-        type: boolean
-      run_vae_test:
-        description: "Run vae-test"
-        required: false
-        default: false
-        type: boolean
-      run_transformer_test:
-        description: "Run transformer-test"
-        required: false
-        default: false
-        type: boolean
-      run_ssim_test:
-        description: "Run ssim-test"
-        required: false
-        default: false
-        type: boolean
-      run_training_test:
-        description: "Run training-test"
-        required: false
-        default: false
-        type: boolean
-      run_training_test_VSA:
-        description: "Run training-test-VSA"
-        required: false
-        default: false
-        type: boolean
-      run_inference_test_STA:
-        description: "Run inference-test-STA"
-        required: false
-        default: false
-        type: boolean
-      run_precision_test_STA:
-        description: "Run precision-test-STA"
-        required: false
-        default: false
-        type: boolean
-      run_precision_test_VSA:
-        description: "Run precision-test-VSA"
-        required: false
-        default: false
-        type: boolean
-      run_unit_test:
-        description: "Run unit-test"
-        required: false
-        default: false
-        type: boolean
-
-env:
-  PYTHONUNBUFFERED: "1"
-
-
-concurrency:
-  group: pr-test-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  pre-commit:
-    uses: ./.github/workflows/pre-commit.yml
-
-  change-filter:
-    runs-on: ubuntu-latest
-    needs: pre-commit
-    if: ${{ github.event.pull_request.draft == false || github.event_name == 'workflow_dispatch' }}
-    outputs:
-      encoder-test: ${{ steps.filter.outputs.encoder-test }}
-      vae-test: ${{ steps.filter.outputs.vae-test }}
-      transformer-test: ${{ steps.filter.outputs.transformer-test }}
-      training-test: ${{ steps.filter.outputs.training-test }}
-      training-test-VSA: ${{ steps.filter.outputs.training-test-VSA }}
-      inference-test-STA: ${{ steps.filter.outputs.inference-test-STA }}
-      precision-test-STA: ${{ steps.filter.outputs.precision-test-STA }}
-      precision-test-VSA: ${{ steps.filter.outputs.precision-test-VSA }}
-      unit-test: ${{ steps.filter.outputs.unit-test }}
-    steps:
-      - uses: actions/checkout@v4
-      - uses: dorny/paths-filter@v3
-        id: filter
-        with:
-          filters: |
-            # Define reusable path patterns
-            common-paths: &common-paths
-              - 'pyproject.toml'
-              - 'docker/Dockerfile.python3.10'
-              - 'docker/Dockerfile.python3.11'
-              - 'docker/Dockerfile.python3.12'
-            sta-kernel-paths: &sta-kernel-paths
-              - 'csrc/attn/sliding_tile_attn/**'
-              - 'csrc/attn/sliding_tile_attn/tk/**'
-              - 'csrc/attn/sliding_tile_attn/setup.py'
-              - 'csrc/attn/sliding_tile_attn/config_sta.py'
-              - 'csrc/attn/sliding_tile_attn/st_attn.cpp'
-            vsa-kernel-paths: &vsa-kernel-paths
-              - 'csrc/attn/video_sparse_attn/**'
-              - 'csrc/attn/video_sparse_attn/tk/**'
-              - 'csrc/attn/video_sparse_attn/setup.py'
-              - 'csrc/attn/video_sparse_attn/config_vsa.py'
-              - 'csrc/attn/video_sparse_attn/vsa.cpp'
-            vsa-paths: &vsa-paths
-              - 'fastvideo/**'
-              - *common-paths
-              - *vsa-kernel-paths
-            
-            # Actual tests
-            encoder-test:
-              - 'fastvideo/models/encoders/**'
-              - 'fastvideo/models/loader/**'
-              - 'fastvideo/tests/encoders/**'
-              - *common-paths
-            vae-test:
-              - 'fastvideo/models/vaes/**'
-              - 'fastvideo/models/loader/**'
-              - 'fastvideo/tests/vaes/**'
-              - *common-paths
-            transformer-test:
-              - 'fastvideo/models/dits/**'
-              - 'fastvideo/models/loader/**'
-              - 'fastvideo/tests/transformers/**'
-              - 'fastvideo/layers/**'
-              - 'fastvideo/attention/**'
-              - *common-paths
-            training-test:
-              - 'fastvideo/**'
-              - *common-paths
-            training-test-VSA:
-              - 'fastvideo/**'
-              - *common-paths
-              - *vsa-kernel-paths
-            inference-test-STA:
-              - 'fastvideo/**'
-              - *common-paths
-              - *sta-kernel-paths
-            precision-test-STA:
-              - *common-paths
-              - *sta-kernel-paths
-            precision-test-VSA:
-              - *common-paths
-              - *vsa-kernel-paths
-            unit-test:
-              - 'fastvideo/**'
-              - *common-paths
-
-  encoder-test:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.encoder-test == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_encoder_test == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "encoder-test"
-      gpu_type: "NVIDIA A40"
-      gpu_count: 1
-      volume_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "uv pip install -e .[test] && pytest ./fastvideo/tests/encoders -s"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-
-  vae-test:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.vae-test == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_vae_test == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "vae-test"
-      gpu_type: "NVIDIA A40"
-      gpu_count: 1
-      volume_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "uv pip install -e .[test] && pytest ./fastvideo/tests/vaes -s"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-
-  transformer-test:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.transformer-test == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_transformer_test == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "transformer-test"
-      gpu_type: "NVIDIA L40S"
-      gpu_count: 1
-      volume_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "uv pip install -e .[test] && pytest ./fastvideo/tests/transformers -s"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-      
-  ssim-test:
-    needs: change-filter
-    if: >-
-      github.event_name != 'workflow_dispatch' || (github.event_name == 'workflow_dispatch' && github.event.inputs.run_ssim_test == 'true')
-    strategy:
-      fail-fast: false
-      matrix:
-        python-version: [
-          # {version: "3.10", tag: "latest"},
-          # {version: "3.11", tag: "py3.11-latest"},
-          {version: "3.12", tag: "py3.12-latest"}
-        ]
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "ssim-test-py${{ matrix.python-version.version }}"
-      gpu_type: "NVIDIA A40"
-      gpu_count: 2
-      volume_size: 200
-      disk_size: 200
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:${{ matrix.python-version.tag }}"
-      test_command: "uv pip install -e .[test] && pytest ./fastvideo/tests/ssim -vs"
-      timeout_minutes: 60
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-
-  training-test:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.training-test == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_training_test == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "training-test"
-      gpu_type: "NVIDIA A40"
-      gpu_count: 4
-      volume_size: 100
-      disk_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "wandb login $WANDB_API_KEY && uv pip install -e .[test] && pytest ./fastvideo/tests/training/Vanilla -srP"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-      WANDB_API_KEY: ${{ secrets.WANDB_API_KEY }}
-
-  training-test-VSA:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.training-test-VSA == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_training_test_VSA == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "training-test-VSA"
-      gpu_type: "NVIDIA H100 NVL"
-      gpu_count: 2
-      volume_size: 100
-      disk_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "wandb login $WANDB_API_KEY && uv pip install -e .[test] && pytest ./fastvideo/tests/training/VSA -srP"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-      WANDB_API_KEY: ${{ secrets.WANDB_API_KEY }}
-
-  inference-test-STA:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.inference-test-STA == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_inference_test_STA == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "inference-test-STA"
-      gpu_type: "NVIDIA H100 NVL"
-      gpu_count: 2
-      volume_size: 100
-      disk_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "uv pip install -e .[test] && pytest ./fastvideo/tests/inference/STA -srP"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-
-  precision-test-STA:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.precision-test-STA == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_precision_test_STA == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "precision-test-STA"
-      gpu_type: "NVIDIA H100 NVL"
-      gpu_count: 1
-      volume_size: 100
-      disk_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "uv pip install -e .[test] && python csrc/attn/tests/test_sta.py"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-
-  precision-test-VSA:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.precision-test-VSA == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_precision_test_VSA == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "precision-test-VSA"
-      gpu_type: "NVIDIA H100 NVL"
-      gpu_count: 1
-      volume_size: 100
-      disk_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "uv pip install -e .[test] && python csrc/attn/tests/test_vsa.py"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-
-  unit-test:
-    needs: change-filter
-    if: >-
-      (github.event_name != 'workflow_dispatch' && needs.change-filter.outputs.unit-test == 'true') || 
-      (github.event_name == 'workflow_dispatch' && github.event.inputs.run_unit_test == 'true')
-    uses: ./.github/workflows/runpod-test.yml
-    with:
-      job_id: "unit-test"
-      gpu_type: "NVIDIA L40S"
-      gpu_count: 1
-      volume_size: 100
-      disk_size: 100
-      image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-      test_command: "uv pip install -e .[test] && pytest ./fastvideo/dataset/ -vs && pytest ./fastvideo/workflow/ -vs && pytest ./fastvideo/entrypoints/ -vs"
-      timeout_minutes: 30
-    secrets:
-      RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-      RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-
-  # nightly-test:
-  #   if: >-
-  #     (github.event_name == 'workflow_dispatch' && github.event.inputs.run_nightly_test == 'true')
-  #   uses: ./.github/workflows/runpod-test.yml
-  #   with:
-  #     job_id: "nightly-test"
-  #     gpu_type: "NVIDIA A40"
-  #     gpu_count: 4
-  #     volume_size: 100
-  #     disk_size: 100
-  #     image: "ghcr.io/${{ github.repository }}/fastvideo-dev:py3.12-latest"
-  #     test_command: "wandb login $WANDB_API_KEY && uv pip install -e .[test] && pytest ./fastvideo/tests/nightly/test_e2e_overfit_single_sample.py -vs"
-  #     timeout_minutes: 30
-  #   secrets:
-  #     RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-  #     RUNPOD_PRIVATE_KEY: ${{ secrets.RUNPOD_PRIVATE_KEY }}
-  #     WANDB_API_KEY: ${{ secrets.WANDB_API_KEY }}
-
-  runpod-cleanup:
-    # Add other jobs to this list as you create them
-    needs: [encoder-test, vae-test, transformer-test, ssim-test, training-test, training-test-VSA, inference-test-STA, precision-test-STA, precision-test-VSA] 
-    if: ${{ always() && ((github.event_name != 'workflow_dispatch' && github.event.pull_request.draft == false) || github.event_name == 'workflow_dispatch') }}
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-
-      - name: Set up Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: "3.10"
-
-      - name: Install dependencies
-        run: pip install requests
-
-      - name: Cleanup all RunPod instances
-        env:
-          JOB_IDS: '["encoder-test", "vae-test", "transformer-test", "ssim-test-py3.10", "ssim-test-py3.11", "ssim-test-py3.12", "training-test", "training-test-VSA", "inference-test-STA", "precision-test-STA", "precision-test-VSA"]'
-          RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-          GITHUB_RUN_ID: ${{ github.run_id }}
-        run: python .github/scripts/runpod_cleanup.py

.github/workflows/pre-commit.yml

@@ -1,18 +0,0 @@
-name: pre-commit
-
-on:
-    workflow_call:
-
-jobs:
-  pre-commit:
-    runs-on: ubuntu-latest
-    steps:
-    - uses: actions/checkout@v4
-    - uses: actions/setup-python@v5
-      with:
-        python-version: "3.12"
-    - run: echo "::add-matcher::.github/workflows/matchers/actionlint.json"
-    - run: echo "::add-matcher::.github/workflows/matchers/mypy.json"
-    - uses: pre-commit/action@v3.0.1
-      with:
-        extra_args: --all-files --hook-stage manual

.github/workflows/publish-kernel.yml

@@ -36,11 +36,11 @@ jobs:
           
           if [ "$NEW_VERSION" != "$OLD_VERSION" ]; then
             echo "Version changed from $OLD_VERSION to $NEW_VERSION"
-            echo "changed=true" >> $GITHUB_OUTPUT
-            echo "new-version=$NEW_VERSION" >> $GITHUB_OUTPUT
+            echo "changed=true" >> "$GITHUB_OUTPUT"
+            echo "new-version=$NEW_VERSION" >> "$GITHUB_OUTPUT"
           else
             echo "Version did not change"
-            echo "changed=false" >> $GITHUB_OUTPUT
+            echo "changed=false" >> "$GITHUB_OUTPUT"
           fi
 
   build_wheels:

.github/workflows/runpod-test.yml

@@ -1,94 +0,0 @@
-name: RunPod Test
-
-on:
-  workflow_call:
-    inputs:
-      job_id:
-        required: true
-        type: string
-        description: "Unique identifier for this test job"
-      gpu_type:
-        required: true
-        type: string
-        description: "GPU type to use (e.g. NVIDIA A40, NVIDIA L40S)"
-      gpu_count:
-        required: true
-        type: number
-        description: "Number of GPUs to use"
-      volume_size:
-        required: false
-        type: number
-        default: 20
-        description: "Volume size in GB"
-      disk_size:
-        required: false
-        type: number
-        default: 20
-        description: "Disk size in GB"
-      image:
-        required: true
-        type: string
-        description: "Docker image to use"
-      test_command:
-        required: true
-        type: string
-        description: "Command to run tests"
-      timeout_minutes:
-        required: false
-        type: number
-        default: 30
-        description: "Timeout in minutes"
-    secrets:
-      RUNPOD_API_KEY:
-        required: true
-      RUNPOD_PRIVATE_KEY:
-        required: true
-      WANDB_API_KEY:
-        required: false
-
-jobs:
-  run-test:
-    runs-on: ubuntu-latest
-    environment: runpod-runners
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-
-      - name: Set up Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: "3.12"
-
-      - name: Set up SSH key
-        run: |
-          mkdir -p ~/.ssh
-          echo "${{ secrets.RUNPOD_PRIVATE_KEY }}" > ~/.ssh/id_rsa
-          chmod 600 ~/.ssh/id_rsa
-          ssh-keygen -y -f ~/.ssh/id_rsa > ~/.ssh/id_rsa.pub
-
-      - name: Install dependencies
-        run: pip install requests
-
-      - name: Run tests on RunPod
-        env:
-          JOB_ID: ${{ inputs.job_id }}
-          RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-          GITHUB_RUN_ID: ${{ github.run_id }}
-          WANDB_API_KEY: ${{ secrets.WANDB_API_KEY }}
-        timeout-minutes: ${{ inputs.timeout_minutes }}
-        run: >-
-          python .github/scripts/runpod_api.py 
-          --gpu-type "${{ inputs.gpu_type }}"
-          --gpu-count ${{ inputs.gpu_count }}
-          --volume-size ${{ inputs.volume_size }}
-          --disk-size ${{ inputs.disk_size }}
-          --image "${{ inputs.image }}"
-          --test-command "${{ inputs.test_command }}"
-
-      - name: Terminate RunPod Instances
-        if: ${{ always() }}
-        env:
-          RUNPOD_API_KEY: ${{ secrets.RUNPOD_API_KEY }}
-          GITHUB_RUN_ID: ${{ github.run_id }}
-          JOB_ID: ${{ inputs.job_id }}
-        run: python .github/scripts/runpod_cleanup.py 

.github/workflows/sta-publish.yml

@@ -1,249 +0,0 @@
-name: Publish Sliding Tile Attention Kernel to PyPI on Version Change
-
-on:
-  push:
-    branches:
-      - main
-    paths:
-      - "csrc/attn/sliding_tile_attn/setup.py"
-  workflow_dispatch:
-
-jobs:
-  check-version-change:
-    runs-on: ubuntu-latest
-    outputs:
-      version-changed: ${{ steps.check-version.outputs.changed }}
-      new-version: ${{ steps.check-version.outputs.new-version }}
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 2
-
-      - name: Check if version changed
-        id: check-version
-        run: |
-          cd csrc/attn/sliding_tile_attn
-          # Get current commit's version
-          NEW_VERSION=$(grep -oP 'VERSION\s*=\s*"\K[^"]+' setup.py)
-          echo "New version: $NEW_VERSION"
-          
-          # Get previous version from git history
-          OLD_VERSION=$(git show HEAD~1:./setup.py | grep -oP 'VERSION\s*=\s*"\K[^"]+' || echo "0.0.0")
-          echo "Old version: $OLD_VERSION"
-          
-          if [ "$NEW_VERSION" != "$OLD_VERSION" ]; then
-            echo "Version changed from $OLD_VERSION to $NEW_VERSION"
-            echo "changed=true" >> $GITHUB_OUTPUT
-            echo "new-version=$NEW_VERSION" >> $GITHUB_OUTPUT
-          else
-            echo "Version did not change"
-            echo "changed=false" >> $GITHUB_OUTPUT
-          fi
-
-  build_wheels:
-    name: Build Wheel
-    needs: check-version-change
-    if: ${{ needs.check-version-change.outputs.version-changed == 'true' || github.event_name == 'workflow_dispatch' }}
-    runs-on: ${{ matrix.os }}
-
-    strategy:
-      fail-fast: false
-      matrix:
-          # Using ubuntu-20.04 instead of 22.04 for more compatibility (glibc). Ideally we'd use the
-          # manylinux docker image, but I haven't figured out how to install CUDA on manylinux.
-          os: [ubuntu-22.04]
-          python-version: ['3.10', '3.11', '3.12', '3.13']
-          torch-version: ['2.5.1', '2.6.0']
-          cuda-version: ['12.4.1', '12.5.1', '12.6.3']
-
-    steps:
-      - name: Free up disk space
-        run: |
-          echo "Initial disk space:"
-          df -h
-          
-          # Remove large directories
-          sudo rm -rf /usr/share/dotnet
-          sudo rm -rf /usr/local/lib/android
-          sudo rm -rf /opt/ghc
-          sudo rm -rf /usr/local/share/boost
-          sudo rm -rf /usr/share/swift
-          sudo rm -rf /usr/local/lib/node_modules
-          sudo rm -rf /usr/local/share/powershell
-          sudo rm -rf /usr/share/rust
-          sudo rm -rf /usr/local/.ghcup
-          
-          # Remove cached files
-          sudo rm -rf /var/lib/apt/lists/*
-          sudo rm -rf /var/cache/apt/archives/*
-          
-          echo "Disk space after cleanup:"
-          df -h
-
-      - name: Checkout
-        uses: actions/checkout@v4
-
-      - name: Set up Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: ${{ matrix.python-version }}
-
-      - name: Install CUDA ${{ matrix.cuda-version }}
-        uses: Jimver/cuda-toolkit@v0.2.21
-        id: cuda-toolkit
-        with:
-          cuda: ${{ matrix.cuda-version }}
-          linux-local-args: '["--toolkit"]'
-          method: 'network'
-
-      - name: Install dependencies (GCC, Clang, CUDA Paths, Git)
-        run: |
-          sudo apt update
-          sudo apt install -y git patchelf gcc-11 g++-11 clang-11
-          sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-11 100 --slave /usr/bin/g++ g++ /usr/bin/g++-11
-
-          # Allow Git to Access Safe Directory
-          git config --global --add safe.directory /__w/FastVideo/FastVideo
-
-          # Set CUDA environment variables
-          export CUDA_HOME=/usr/local/cuda-${{ matrix.cuda-version }}
-          export PATH=${CUDA_HOME}/bin:${PATH}
-          export LD_LIBRARY_PATH=${CUDA_HOME}/lib64:$LD_LIBRARY_PATH
-          
-          # Verify installation
-          gcc --version
-          g++ --version
-          clang-11 --version
-          nvcc --version
-
-      - name: Install PyTorch ${{ matrix.torch-version }}+cu${{ matrix.cuda-version }}
-        run: |
-          pip install --upgrade pip
-          # With python 3.13 and torch 2.5.1, unless we update typing-extensions, we get error
-          # AttributeError: attribute '__default__' of 'typing.ParamSpec' objects is not writable
-          pip install typing-extensions==4.12.2
-          # We want to figure out the CUDA version to download pytorch
-          # e.g. we can have system CUDA version being 11.7 but if torch==1.12 then we need to download the wheel from cu116
-          # see https://github.com/pytorch/pytorch/blob/main/RELEASE.md#release-compatibility-matrix
-          export TORCH_CUDA_VERSION=124
-          pip install --no-cache-dir torch==${{ matrix.torch-version }} --index-url https://download.pytorch.org/whl/cu${TORCH_CUDA_VERSION}
-          nvcc --version
-          python --version
-          python -c "import torch; print('PyTorch:', torch.__version__)"
-          python -c "import torch; print('CUDA:', torch.version.cuda)"
-          python -c "from torch.utils import cpp_extension; print (cpp_extension.CUDA_HOME)"
-
-      - name: Build wheel
-        run: |
-          export PYTHONPATH=$GITHUB_WORKSPACE:$PYTHONPATH
-          
-          # We want setuptools >= 49.6.0 otherwise we can't compile the extension if system CUDA version is 11.7 and pytorch cuda version is 11.6
-          # https://github.com/pytorch/pytorch/blob/664058fa83f1d8eede5d66418abff6e20bd76ca8/torch/utils/cpp_extension.py#L810
-          # However this still fails so I'm using a newer version of setuptools
-          pip install setuptools
-          pip install ninja packaging wheel
-
-          cd csrc/attn/sliding_tile_attn  # Move into the correct folder
-          git submodule update --init --recursive # Ensure ThunderKittens submodule is initialized
-          python setup.py bdist_wheel --dist-dir=dist
-
-      - name: Rename wheel file
-        run: |
-          cd csrc/attn/sliding_tile_attn
-
-          CUDA_SHORT_VERSION=$(echo ${{ matrix.cuda-version }} | cut -d. -f1,2 | sed 's/\.//g')
-          TORCH_SHORT_VERSION=$(echo ${{ matrix.torch-version }} | cut -d. -f1,2)
-          # Get the correct version format
-          tmpname=cu${CUDA_SHORT_VERSION}torch${TORCH_SHORT_VERSION}
-          wheel_name=$(ls dist/*whl | xargs -n 1 basename | sed "s/-/+$tmpname-/2")
-          # Rename with version information
-          ls dist/*whl |xargs -I {} mv {} dist/${wheel_name}
-          echo "wheel_name=${wheel_name}" >> $GITHUB_ENV
-
-      - name: Upload wheel artifact
-        uses: actions/upload-artifact@v4
-        with:
-          name: ${{ env.wheel_name }}
-          path: csrc/attn/sliding_tile_attn/dist/*.whl
-          retention-days: 90
-
-  publish_package:
-    name: Publish package
-    needs: [build_wheels, check-version-change]
-    if: ${{ needs.check-version-change.outputs.version-changed == 'true' || github.event_name == 'workflow_dispatch' }}
-    runs-on: ubuntu-22.04
-    permissions:
-      id-token: write  # Needed for OIDC Trusted Publishing
-
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/setup-python@v5
-        with:
-          python-version: '3.10'
-      
-      - name: Install CUDA 12.4.1
-        uses: Jimver/cuda-toolkit@v0.2.21
-        id: cuda-toolkit
-        with:
-          cuda: 12.4.1
-          linux-local-args: '["--toolkit"]'
-          method: 'network'
-          sub-packages: '["nvcc"]'
-
-      - name: Install dependencies (GCC, Clang, CUDA Paths, Git)
-        run: |
-          sudo apt update
-          sudo apt install -y git patchelf gcc-11 g++-11 clang-11
-          sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-11 100 --slave /usr/bin/g++ g++ /usr/bin/g++-11
-
-          # Allow Git to Access Safe Directory
-          git config --global --add safe.directory /__w/FastVideo/FastVideo
-
-          # Set CUDA environment variables
-          export CUDA_HOME=/usr/local/cuda-12.4.1
-          export PATH=${CUDA_HOME}/bin:${PATH}
-          export LD_LIBRARY_PATH=${CUDA_HOME}/lib64:$LD_LIBRARY_PATH
-          
-          # Verify installation
-          gcc --version
-          g++ --version
-          clang-11 --version
-          nvcc --version
-
-      - name: Install PyTorch 2.5.1+cu12.4.1
-        run: |
-          pip install --upgrade pip
-          # With python 3.13 and torch 2.5.1, unless we update typing-extensions, we get error
-          # AttributeError: attribute '__default__' of 'typing.ParamSpec' objects is not writable
-          pip install typing-extensions==4.12.2
-          # We want to figure out the CUDA version to download pytorch
-          # e.g. we can have system CUDA version being 11.7 but if torch==1.12 then we need to download the wheel from cu116
-          # see https://github.com/pytorch/pytorch/blob/main/RELEASE.md#release-compatibility-matrix
-          export TORCH_CUDA_VERSION=124
-          pip install --no-cache-dir torch==2.5.1 --index-url https://download.pytorch.org/whl/cu${TORCH_CUDA_VERSION}
-          nvcc --version
-          python --version
-          python -c "import torch; print('PyTorch:', torch.__version__)"
-          python -c "import torch; print('CUDA:', torch.version.cuda)"
-          python -c "from torch.utils import cpp_extension; print (cpp_extension.CUDA_HOME)"
-
-      - name: Build source distribution
-        run: |
-          export PYTHONPATH=$GITHUB_WORKSPACE:$PYTHONPATH
-          
-          # We want setuptools >= 49.6.0 otherwise we can't compile the extension if system CUDA version is 11.7 and pytorch cuda version is 11.6
-          # https://github.com/pytorch/pytorch/blob/664058fa83f1d8eede5d66418abff6e20bd76ca8/torch/utils/cpp_extension.py#L810
-          # However this still fails so I'm using a newer version of setuptools
-          pip install setuptools
-          pip install ninja packaging wheel
-
-          cd csrc/attn/sliding_tile_attn  # Move into the correct folder
-          git submodule update --init --recursive # Ensure ThunderKittens submodule is initialized
-          python setup.py sdist --dist-dir=dist
-
-      - name: Publish release distributions to PyPI
-        uses: pypa/gh-action-pypi-publish@release/v1
-        with:
-          packages-dir: csrc/attn/sliding_tile_attn/dist/

.github/workflows/test.yml

@@ -1,31 +0,0 @@
-name: Run Tests
-
-on:
-  push:
-    branches: [ main ]
-  pull_request:
-    branches: [ main ]
-
-jobs:
-  test:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Check out repository
-        uses: actions/checkout@v4
-
-      - name: Set up Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: '3.12'  # or any version you need
-
-      - name: Install dependencies
-        run: |
-          python -m pip install --upgrade pip setuptools wheel
-          pip install torch
-          pip install packaging ninja
-          pip install -e .
-          pip install pytest
-
-      - name: Run Pytest
-        run: |
-          pytest --ignore csrc/attn/test

.github/workflows/vsa-publish.yml

@@ -1,257 +0,0 @@
-name: Publish Video Sparse Attention Kernel to PyPI on Version Change
-
-on:
-  push:
-    branches:
-      - main
-    paths:
-      - "csrc/attn/video_sparse_attn/setup.py"
-  workflow_dispatch:
-
-jobs:
-  check-version-change:
-    runs-on: ubuntu-latest
-    outputs:
-      version-changed: ${{ steps.check-version.outputs.changed }}
-      new-version: ${{ steps.check-version.outputs.new-version }}
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-        with:
-          fetch-depth: 2
-
-      - name: Check if version changed
-        id: check-version
-        run: |
-          cd csrc/attn/video_sparse_attn
-          # Get current commit's version
-          NEW_VERSION=$(grep -oP 'VERSION\s*=\s*"\K[^"]+' setup.py)
-          echo "New version: $NEW_VERSION"
-          
-          # Get previous version from git history
-          OLD_VERSION=$(git show HEAD~1:./setup.py | grep -oP 'VERSION\s*=\s*"\K[^"]+' || echo "0.0.0")
-          echo "Old version: $OLD_VERSION"
-          
-          if [ "$NEW_VERSION" != "$OLD_VERSION" ]; then
-            echo "Version changed from $OLD_VERSION to $NEW_VERSION"
-            echo "changed=true" >> $GITHUB_OUTPUT
-            echo "new-version=$NEW_VERSION" >> $GITHUB_OUTPUT
-          else
-            echo "Version did not change"
-            echo "changed=false" >> $GITHUB_OUTPUT
-          fi
-
-  build_wheels:
-    name: Build Wheel
-    needs: check-version-change
-    if: ${{ needs.check-version-change.outputs.version-changed == 'true' || github.event_name == 'workflow_dispatch' }}
-    runs-on: ${{ matrix.os }}
-
-    strategy:
-      fail-fast: false
-      matrix:
-          # Using ubuntu-20.04 instead of 22.04 for more compatibility (glibc). Ideally we'd use the
-          # manylinux docker image, but I haven't figured out how to install CUDA on manylinux.
-          os: [ubuntu-22.04]
-          python-version: ['3.10', '3.11', '3.12', '3.13']
-          # For version reference https://pytorch.org/get-started/previous-versions/
-          torch-cuda:
-            - torch-version: '2.5.1'
-              cuda-version: '12.4.1'
-              torch-cuda-short: 'cu124'
-            - torch-version: '2.6.0'
-              cuda-version: '12.6.3'
-              torch-cuda-short: 'cu126'
-            - torch-version: '2.7.1'
-              cuda-version: '12.8.0'
-              torch-cuda-short: 'cu128'
-
-    steps:
-      - name: Free up disk space
-        run: |
-          echo "Initial disk space:"
-          df -h
-          
-          # Remove large directories
-          sudo rm -rf /usr/share/dotnet
-          sudo rm -rf /usr/local/lib/android
-          sudo rm -rf /opt/ghc
-          sudo rm -rf /usr/local/share/boost
-          sudo rm -rf /usr/share/swift
-          sudo rm -rf /usr/local/lib/node_modules
-          sudo rm -rf /usr/local/share/powershell
-          sudo rm -rf /usr/share/rust
-          sudo rm -rf /usr/local/.ghcup
-          
-          # Remove cached files
-          sudo rm -rf /var/lib/apt/lists/*
-          sudo rm -rf /var/cache/apt/archives/*
-          
-          echo "Disk space after cleanup:"
-          df -h
-
-      - name: Checkout
-        uses: actions/checkout@v4
-
-      - name: Set up Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: ${{ matrix.python-version }}
-
-      - name: Install CUDA ${{ matrix.torch-cuda.cuda-version }}
-        uses: Jimver/cuda-toolkit@v0.2.21
-        id: cuda-toolkit
-        with:
-          cuda: ${{ matrix.torch-cuda.cuda-version }}
-          linux-local-args: '["--toolkit"]'
-          method: 'network'
-
-      - name: Install dependencies (GCC, Clang, CUDA Paths, Git)
-        run: |
-          sudo apt update
-          sudo apt install -y git patchelf gcc-11 g++-11 clang-11
-          sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-11 100 --slave /usr/bin/g++ g++ /usr/bin/g++-11
-
-          # Allow Git to Access Safe Directory
-          git config --global --add safe.directory /__w/FastVideo/FastVideo
-
-          # Set CUDA environment variables
-          export CUDA_HOME=/usr/local/cuda-${{ matrix.torch-cuda.cuda-version }}
-          export PATH=${CUDA_HOME}/bin:${PATH}
-          export LD_LIBRARY_PATH=${CUDA_HOME}/lib64:$LD_LIBRARY_PATH
-          
-          # Verify installation
-          gcc --version
-          g++ --version
-          clang-11 --version
-          nvcc --version
-
-      - name: Install PyTorch ${{ matrix.torch-cuda.torch-version }}+cu${{ matrix.torch-cuda.cuda-version }}
-        run: |
-          pip install --upgrade pip
-          # With python 3.13 and torch 2.5.1, unless we update typing-extensions, we get error
-          # AttributeError: attribute '__default__' of 'typing.ParamSpec' objects is not writable
-          pip install typing-extensions==4.12.2
-          # We want to figure out the CUDA version to download pytorch
-          # e.g. we can have system CUDA version being 11.7 but if torch==1.12 then we need to download the wheel from cu116
-          # see https://github.com/pytorch/pytorch/blob/main/RELEASE.md#release-compatibility-matrix
-          pip install --no-cache-dir torch==${{ matrix.torch-cuda.torch-version }} --index-url https://download.pytorch.org/whl/${{matrix.torch-cuda.torch-cuda-short}}
-          nvcc --version
-          python --version
-          python -c "import torch; print('PyTorch:', torch.__version__)"
-          python -c "import torch; print('CUDA:', torch.version.cuda)"
-          python -c "from torch.utils import cpp_extension; print (cpp_extension.CUDA_HOME)"
-
-      - name: Build wheel
-        run: |
-          export PYTHONPATH=$GITHUB_WORKSPACE:$PYTHONPATH
-          
-          # We want setuptools >= 49.6.0 otherwise we can't compile the extension if system CUDA version is 11.7 and pytorch cuda version is 11.6
-          # https://github.com/pytorch/pytorch/blob/664058fa83f1d8eede5d66418abff6e20bd76ca8/torch/utils/cpp_extension.py#L810
-          # However this still fails so I'm using a newer version of setuptools
-          pip install setuptools
-          pip install ninja packaging wheel
-
-          cd csrc/attn/video_sparse_attn  # Move into the correct folder
-          git submodule update --init --recursive # Ensure ThunderKittens submodule is initialized
-          python setup.py bdist_wheel --dist-dir=dist
-
-      - name: Rename wheel file
-        run: |
-          cd csrc/attn/video_sparse_attn
-
-          CUDA_SHORT_VERSION=$(echo ${{ matrix.torch-cuda.cuda-version }} | cut -d. -f1,2 | sed 's/\.//g')
-          TORCH_SHORT_VERSION=$(echo ${{ matrix.torch-cuda.torch-version }} | cut -d. -f1,2)
-          # Get the correct version format
-          tmpname=cu${CUDA_SHORT_VERSION}torch${TORCH_SHORT_VERSION}
-          wheel_name=$(ls dist/*whl | xargs -n 1 basename | sed "s/-/+$tmpname-/2")
-          # Rename with version information
-          ls dist/*whl |xargs -I {} mv {} dist/${wheel_name}
-          echo "wheel_name=${wheel_name}" >> $GITHUB_ENV
-
-      - name: Upload wheel artifact
-        uses: actions/upload-artifact@v4
-        with:
-          name: ${{ env.wheel_name }}
-          path: csrc/attn/video_sparse_attn/dist/*.whl
-          retention-days: 90
-
-  publish_package:
-    name: Publish package
-    needs: [build_wheels, check-version-change]
-    if: ${{ needs.check-version-change.outputs.version-changed == 'true' || github.event_name == 'workflow_dispatch' }}
-    runs-on: ubuntu-22.04
-    permissions:
-      id-token: write  # Needed for OIDC Trusted Publishing
-
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/setup-python@v5
-        with:
-          python-version: '3.10'
-
-      - name: Install CUDA 12.4.1
-        uses: Jimver/cuda-toolkit@v0.2.21
-        id: cuda-toolkit
-        with:
-          cuda: 12.4.1
-          linux-local-args: '["--toolkit"]'
-          method: 'network'
-          sub-packages: '["nvcc"]'
-
-      - name: Install dependencies (GCC, Clang, CUDA Paths, Git)
-        run: |
-          sudo apt update
-          sudo apt install -y git patchelf gcc-11 g++-11 clang-11
-          sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-11 100 --slave /usr/bin/g++ g++ /usr/bin/g++-11
-
-          # Allow Git to Access Safe Directory
-          git config --global --add safe.directory /__w/FastVideo/FastVideo
-
-          # Set CUDA environment variables
-          export CUDA_HOME=/usr/local/cuda-12.4.1
-          export PATH=${CUDA_HOME}/bin:${PATH}
-          export LD_LIBRARY_PATH=${CUDA_HOME}/lib64:$LD_LIBRARY_PATH
-          
-          # Verify installation
-          gcc --version
-          g++ --version
-          clang-11 --version
-          nvcc --version
-
-      - name: Install PyTorch 2.5.1+cu12.4.1
-        run: |
-          pip install --upgrade pip
-          # With python 3.13 and torch 2.5.1, unless we update typing-extensions, we get error
-          # AttributeError: attribute '__default__' of 'typing.ParamSpec' objects is not writable
-          pip install typing-extensions==4.12.2
-          # We want to figure out the CUDA version to download pytorch
-          # e.g. we can have system CUDA version being 11.7 but if torch==1.12 then we need to download the wheel from cu116
-          # see https://github.com/pytorch/pytorch/blob/main/RELEASE.md#release-compatibility-matrix
-          export TORCH_CUDA_VERSION=124
-          pip install --no-cache-dir torch==2.5.1 --index-url https://download.pytorch.org/whl/cu${TORCH_CUDA_VERSION}
-          nvcc --version
-          python --version
-          python -c "import torch; print('PyTorch:', torch.__version__)"
-          python -c "import torch; print('CUDA:', torch.version.cuda)"
-          python -c "from torch.utils import cpp_extension; print (cpp_extension.CUDA_HOME)"
-
-      - name: Build source distribution
-        run: |
-          export PYTHONPATH=$GITHUB_WORKSPACE:$PYTHONPATH
-          
-          # We want setuptools >= 49.6.0 otherwise we can't compile the extension if system CUDA version is 11.7 and pytorch cuda version is 11.6
-          # https://github.com/pytorch/pytorch/blob/664058fa83f1d8eede5d66418abff6e20bd76ca8/torch/utils/cpp_extension.py#L810
-          # However this still fails so I'm using a newer version of setuptools
-          pip install setuptools
-          pip install ninja packaging wheel
-
-          cd csrc/attn/video_sparse_attn  # Move into the correct folder
-          git submodule update --init --recursive # Ensure ThunderKittens submodule is initialized
-          python setup.py sdist --dist-dir=dist
-
-      - name: Publish release distributions to PyPI
-        uses: pypa/gh-action-pypi-publish@release/v1
-        with:
-          packages-dir: csrc/attn/video_sparse_attn/dist/

.gitignore

@@ -33,11 +33,12 @@ env
 **.txt
 *.log
 weights/
-official_weights/
-converted_weights/
+logs/
 
 # SSIM test outputs
 fastvideo/tests/ssim/generated_videos/
+**/.cache/**
+
 
 # Distribution / packaging
 build/
@@ -83,3 +84,11 @@ docs/distillation/examples/
 
 dmd_t2v_output/
 preprocess_output_text/
+
+# Next.js / Node artifacts under ui/: see ui/.gitignore
+
+.claude/
+.codex/
+.sisyphus/
+openspec/
+fastvideo/tests/ssim/reference_videos/**

.nvimlog

@@ -0,0 +1 @@
+WRN 2026-03-26T13:46:33.469 ?.19646    server_start:193: Failed to start server: operation not permitted: /var/folders/z_/h_6myyk14d1b7z87z3vy4mjh0000gn/T/nvim.dsynkd/iSe0el/nvim.19646.0

.pre-commit-config.yaml

@@ -18,10 +18,8 @@ exclude: |
     fastvideo/train\.py|
     fastvideo/utils/.*|
     examples/.*|
-    .github/workflows/fastvideo-publish.yml|
-    .github/workflows/sta-publish.yml|
-    .github/workflows/vsa-publish.yml|
-    .github/workflows/build-image-template.yml|
+    .github/workflows/publish-fastvideo.yml|
+    .github/workflows/_template-build-image.yml|
     docs/source/inference/support_matrix.md
   )
 repos:
@@ -60,7 +58,7 @@ repos:
   hooks:
   - id: mypy
     args: [--python-version, '3.10', --follow-imports, "skip", "--disable-error-code", "union-attr", "--disable-error-code", "override" ]
-    additional_dependencies: [types-cachetools, types-setuptools, types-PyYAML, types-requests]
+    additional_dependencies: [types-aiofiles, types-cachetools, types-setuptools, types-PyYAML, types-requests]
 - repo: local
   hooks:
   - id: check-filenames

.python-version

@@ -0,0 +1 @@
+3.12

AGENTS.md

@@ -40,3 +40,17 @@
   - test evidence (`pytest`/SSIM outputs or rationale if skipped),
   - linked issue/PR context,
   - screenshots or sample outputs for UI/demo/docs changes.
+
+## Agent Infrastructure
+
+This repository is agent-friendly. Before doing any work, read:
+
+1. `.agents/onboarding/README.md` — full onboarding guide with step-by-step instructions.
+2. `.agents/memory/codebase-map/README.md` — structural index of the entire repository.
+3. `.agents/skills/` — available agent skills (check if one exists before writing code).
+4. `.agents/workflows/` — SOPs for common procedures (experiment lifecycle, evaluation, etc.).
+5. `.agents/lessons/` — known pitfalls and their documented fixes.
+
+If you are exploring a new procedure that has no existing SOP, document your
+progress in `.agents/exploration/` and flag it for review at the end of your
+session.

README.md

@@ -6,16 +6,17 @@
      | <a href="https://hao-ai-lab.github.io/FastVideo"><b>Documentation</b></a> | <a href="https://hao-ai-lab.github.io/FastVideo/inference/inference_quick_start/"><b> Quick Start</b></a> | <a href="https://github.com/hao-ai-lab/FastVideo/discussions/982"  target="_blank"><b>Weekly Dev Meeting</b></a>  | 🟣💬 <a href="https://join.slack.com/t/fastvideo/shared_invite/zt-3f4lao1uq-u~Ipx6Lt4J27AlD2y~IdLQ" target="_blank"> <b>Slack</b> </a> |  🟣💬 <a href="https://github.com/hao-ai-lab/FastVideo/discussions/1097" target="_blank"> <b> WeChat </b> </a> |
 </p>
 
-**FastVideo is a unified post-training and inference framework for accelerated video generation.**
+**FastVideo is a unified post-training and real-time inference framework for accelerated video generation.**
 
 ## NEWS
-
-- `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)
+- `2026/03/17`: Release Live demo: [Into the Dreamverse: Vibe Directing in FastVideo](https://dreamverse.fastvideo.org/), check out the [Blog](https://haoailab.com/blogs/dreamverse/).
+- `2026/03/13`: Release Live demo: [Create a 5s 1080p Video in 4.5s with FastVideo on a Single GPU](https://1080p.fastvideo.org/), check out the [Blog](https://haoailab.com/blogs/fastvideo_realtime_1080p/).
+- `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/).
 
 ### More News
 
-- `2025/06/14`: Release finetuning and inference code for [VSA](https://arxiv.org/pdf/2505.13389)
+- `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/).
 
@@ -40,19 +41,19 @@ FastVideo has the following features:
 - 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.
+  - See this [page](https://hao-ai-lab.github.io/FastVideo/inference/support_matrix/) for full list of supported models, hardware assumptions, and optimization compatibility.
 
 ## Getting Started
 
-We recommend using an environment manager such as `Conda` to create a clean environment:
+We recommend using [uv](https://docs.astral.sh/uv/) to create a clean environment. If you previously used Conda, switching to uv generally gives faster and more stable installs.
 
 ```bash
-# Create and activate a new conda environment
-conda create -n fastvideo python=3.12
-conda activate fastvideo
+# Create and activate a new uv environment
+uv venv --python 3.12 --seed
+source .venv/bin/activate
 
 # Install FastVideo
-pip install fastvideo
+uv pip install fastvideo
 ```
 
 Please see our [docs](https://hao-ai-lab.github.io/FastVideo/getting_started/installation/) for more detailed installation instructions.
@@ -72,7 +73,7 @@ See below for recipes and datasets:
 
 ### 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:
+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/attention/vsa/#installation). Create a file called `example.py` with the following code:
 
 ```python
 import os
@@ -93,7 +94,6 @@ def main():
     # 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
     )
@@ -122,6 +122,7 @@ For a more detailed guide, please see our [inference quick start](https://hao-ai
 - [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.
+- [HY-WorldPlay](https://github.com/Tencent-Hunyuan/HY-WorldPlay): An action-conditioned world model model trained using FastVideo framework.
 - [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.

__init__.py

@@ -1,15 +1,10 @@
 try:
-    from .comfyui.video_generator.nodes import (NODE_CLASS_MAPPINGS,
-                                                NODE_DISPLAY_NAME_MAPPINGS)
+    from .comfyui.video_generator.nodes import (NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS)
     WEB_DIRECTORY = "./web"
-    __all__ = [
-        'NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS', 'WEB_DIRECTORY'
-    ]
+    __all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS', 'WEB_DIRECTORY']
 except ImportError:
     # ComfyUI environment not available, skip comfyui imports
     NODE_CLASS_MAPPINGS = {}
     NODE_DISPLAY_NAME_MAPPINGS = {}
     WEB_DIRECTORY = "./web"
-    __all__ = [
-        'NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS', 'WEB_DIRECTORY'
-    ]
+    __all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS', 'WEB_DIRECTORY']

benchmarks/fvd/__init__.py

@@ -13,8 +13,7 @@ from .fvd import (
     compute_statistics,
     FVDConfig,
 )
-from .feature_extractors import (BaseFeatureExtractor, I3DFeatureExtractor,
-                                 load_extractor)
+from .feature_extractors import (BaseFeatureExtractor, I3DFeatureExtractor, load_extractor)
 from .video_utils import (
     load_video_auto,
     sample_clips_from_video,

benchmarks/fvd/cli.py

@@ -5,18 +5,11 @@ from .fvd import compute_fvd_with_config, FVDConfig
 
 
 def main() -> int:
-    parser = argparse.ArgumentParser(
-        description='Compute Fréchet Video Distance (FVD)')
+    parser = argparse.ArgumentParser(description='Compute Fréchet Video Distance (FVD)')
 
     # Required arguments
-    parser.add_argument('--real-path',
-                        type=str,
-                        required=True,
-                        help='Path to real videos')
-    parser.add_argument('--gen-path',
-                        type=str,
-                        required=True,
-                        help='Path to generated videos')
+    parser.add_argument('--real-path', type=str, required=True, help='Path to real videos')
+    parser.add_argument('--gen-path', type=str, required=True, help='Path to generated videos')
 
     # Extractor selection
     parser.add_argument('--extractor',
@@ -26,42 +19,19 @@ def main() -> int:
                         help='Feature extractor model to use (default: i3d)')
 
     # Standard args
-    parser.add_argument('--seed',
-                        type=int,
-                        default=None,
-                        help='Random seed for reproducibility')
+    parser.add_argument('--seed', type=int, default=None, help='Random seed for reproducibility')
     parser.add_argument('--protocol',
                         type=str,
                         default=None,
                         choices=['fvd2048_16f', 'fvd2048_128f', 'quick_test'],
                         help='Use standard protocol (overrides other settings)')
-    parser.add_argument('--num-videos',
-                        type=int,
-                        default=2048,
-                        help='Number of videos to use')
-    parser.add_argument('--num-frames',
-                        type=int,
-                        default=16,
-                        help='Number of frames per clip')
-    parser.add_argument('--clip-strategy',
-                        type=str,
-                        default='beginning',
-                        help='Clip sampling strategy')
-    parser.add_argument('--batch-size',
-                        type=int,
-                        default=32,
-                        help='Batch size for feature extraction')
-    parser.add_argument('--device',
-                        type=str,
-                        default='cuda',
-                        help='Device to use (cuda or cpu)')
-    parser.add_argument('--cache-real-features',
-                        type=str,
-                        default=None,
-                        help='Path to cache real video features')
-    parser.add_argument('--quiet',
-                        action='store_true',
-                        help='Suppress progress output')
+    parser.add_argument('--num-videos', type=int, default=2048, help='Number of videos to use')
+    parser.add_argument('--num-frames', type=int, default=16, help='Number of frames per clip')
+    parser.add_argument('--clip-strategy', type=str, default='beginning', help='Clip sampling strategy')
+    parser.add_argument('--batch-size', type=int, default=32, help='Batch size for feature extraction')
+    parser.add_argument('--device', type=str, default='cuda', help='Device to use (cuda or cpu)')
+    parser.add_argument('--cache-real-features', type=str, default=None, help='Path to cache real video features')
+    parser.add_argument('--quiet', action='store_true', help='Suppress progress output')
 
     args = parser.parse_args()
 

benchmarks/fvd/feature_extractors.py

@@ -22,8 +22,7 @@ class BaseFeatureExtractor(ABC, nn.Module):
 
     def __init__(self, device: str = 'cuda'):
         super().__init__()
-        self.device = torch.device(
-            device if torch.cuda.is_available() else 'cpu')
+        self.device = torch.device(device if torch.cuda.is_available() else 'cpu')
 
     @property
     @abstractmethod
@@ -53,10 +52,7 @@ class BaseFeatureExtractor(ABC, nn.Module):
         pass
 
     @torch.no_grad()
-    def extract_features(self,
-                         videos: torch.Tensor,
-                         batch_size: int = 32,
-                         verbose: bool = True) -> torch.Tensor:
+    def extract_features(self, videos: torch.Tensor, batch_size: int = 32, verbose: bool = True) -> torch.Tensor:
         """
         Extract features for a large tensor of videos by batching.
         """
@@ -65,9 +61,7 @@ class BaseFeatureExtractor(ABC, nn.Module):
 
         iterator = range(0, N, batch_size)
         if verbose:
-            iterator = tqdm(
-                iterator,
-                desc=f"Extracting features ({self.__class__.__name__})")
+            iterator = tqdm(iterator, desc=f"Extracting features ({self.__class__.__name__})")
 
         for i in iterator:
             batch = videos[i:i + batch_size].to(self.device)
@@ -95,9 +89,7 @@ class I3DFeatureExtractor(BaseFeatureExtractor):
 
     def _load_model(self) -> torch.nn.Module:
         try:
-            model_path = hf_hub_download(repo_id=self.REPO_ID,
-                                         filename=self.MODEL_FILENAME,
-                                         cache_dir=self.cache_dir)
+            model_path = hf_hub_download(repo_id=self.REPO_ID, filename=self.MODEL_FILENAME, cache_dir=self.cache_dir)
             return torch.jit.load(model_path, map_location=self.device)
         except Exception as e:
             raise RuntimeError(f"Failed to load I3D model: {e}") from e
@@ -119,10 +111,7 @@ class I3DFeatureExtractor(BaseFeatureExtractor):
         # Resize to 224x224
         if H != 224 or W != 224:
             videos = videos.reshape(B * T, C, H, W)
-            videos = F.interpolate(videos,
-                                   size=(224, 224),
-                                   mode='bilinear',
-                                   align_corners=False)
+            videos = F.interpolate(videos, size=(224, 224), mode='bilinear', align_corners=False)
             videos = videos.reshape(B, T, C, 224, 224)
 
         # [B, T, C, H, W] -> [B, C, T, H, W]
@@ -131,21 +120,15 @@ class I3DFeatureExtractor(BaseFeatureExtractor):
     def extract_features_batch(self, videos: torch.Tensor) -> torch.Tensor:
         batch = self.preprocess(videos)
         # TorchScript I3D returns raw logits when return_features=True
-        return self.model(batch,
-                          rescale=False,
-                          resize=False,
-                          return_features=True)
+        return self.model(batch, rescale=False, resize=False, return_features=True)
 
 
 # 2. CLIP Extractor (Semantic/Content Quality)
 class CLIPFeatureExtractor(BaseFeatureExtractor):
 
-    def __init__(self,
-                 device: str = 'cuda',
-                 model_name: str = "openai/clip-vit-base-patch32"):
+    def __init__(self, device: str = 'cuda', model_name: str = "openai/clip-vit-base-patch32"):
         if not TRANSFORMERS_AVAILABLE:
-            raise ImportError(
-                "Please install transformers: pip install transformers")
+            raise ImportError("Please install transformers: pip install transformers")
         super().__init__(device)
         self.processor = CLIPProcessor.from_pretrained(model_name)
         self.model = CLIPModel.from_pretrained(model_name).to(self.device)
@@ -172,9 +155,7 @@ class CLIPFeatureExtractor(BaseFeatureExtractor):
         images = videos.view(B * T, C, H, W)
 
         # HF Processor
-        inputs = self.processor(images=images,
-                                return_tensors="pt",
-                                padding=True)
+        inputs = self.processor(images=images, return_tensors="pt", padding=True)
         inputs = {k: v.to(self.device) for k, v in inputs.items()}
 
         # Extract features [B*T, Dim]
@@ -188,24 +169,15 @@ class CLIPFeatureExtractor(BaseFeatureExtractor):
 # 3. VideoMAE Extractor (Structure/Motion Quality)
 class VideoMAEFeatureExtractor(BaseFeatureExtractor):
 
-    def __init__(self,
-                 device: str = 'cuda',
-                 model_name: str = "MCG-NJU/videomae-base"):
+    def __init__(self, device: str = 'cuda', model_name: str = "MCG-NJU/videomae-base"):
         if not TRANSFORMERS_AVAILABLE:
-            raise ImportError(
-                "Please install transformers: pip install transformers")
+            raise ImportError("Please install transformers: pip install transformers")
         super().__init__(device)
         self.model = VideoMAEModel.from_pretrained(model_name).to(self.device)
         self.model.eval()
 
-        self.register_buffer(
-            'mean',
-            torch.tensor([0.485, 0.456, 0.406],
-                         device=self.device).view(1, 1, 3, 1, 1))
-        self.register_buffer(
-            'std',
-            torch.tensor([0.229, 0.224, 0.225],
-                         device=self.device).view(1, 1, 3, 1, 1))
+        self.register_buffer('mean', torch.tensor([0.485, 0.456, 0.406], device=self.device).view(1, 1, 3, 1, 1))
+        self.register_buffer('std', torch.tensor([0.229, 0.224, 0.225], device=self.device).view(1, 1, 3, 1, 1))
 
     @property
     def feature_dim(self) -> int:
@@ -221,10 +193,7 @@ class VideoMAEFeatureExtractor(BaseFeatureExtractor):
         # 1. Resize to 224x224
         if H != 224 or W != 224:
             videos = videos.view(B * T, C, H, W)
-            videos = F.interpolate(videos,
-                                   size=(224, 224),
-                                   mode='bilinear',
-                                   align_corners=False)
+            videos = F.interpolate(videos, size=(224, 224), mode='bilinear', align_corners=False)
             videos = videos.view(B, T, C, 224, 224)
 
         # 2. Normalize to [0, 1]
@@ -260,5 +229,4 @@ def load_extractor(name: str, device: str = 'cuda') -> BaseFeatureExtractor:
     elif name == 'videomae':
         return VideoMAEFeatureExtractor(device)
     else:
-        raise ValueError(
-            f"Unknown extractor: {name}. Options: i3d, clip, videomae")
+        raise ValueError(f"Unknown extractor: {name}. Options: i3d, clip, videomae")

benchmarks/fvd/fvd.py

@@ -36,8 +36,7 @@ def compute_frechet_distance(mu1: np.ndarray,
 
     if np.iscomplexobj(covmean):
         if not np.allclose(np.diagonal(covmean).imag, 0, atol=1e-3):
-            print(
-                f"Warning: Imaginary component: {np.max(np.abs(covmean.imag))}")
+            print(f"Warning: Imaginary component: {np.max(np.abs(covmean.imag))}")
         covmean = covmean.real
 
     trace_product = np.trace(covmean)
@@ -67,8 +66,7 @@ class FVDConfig:
     temporal_stride: int = 1  # For sliding window clips
 
     # Data processing
-    video_extensions: list[str] = field(
-        default_factory=lambda: ['.mp4', '.avi', '.mov', '.mkv'])
+    video_extensions: list[str] = field(default_factory=lambda: ['.mp4', '.avi', '.mov', '.mkv'])
     support_frame_dirs: bool = True
 
     # Computation
@@ -88,25 +86,17 @@ class FVDConfig:
     @classmethod
     def fvd2048_16f(cls) -> 'FVDConfig':
         """Standard FVD protocol: 2048 videos, 16 frames, beginning clip."""
-        return cls(num_videos=2048,
-                   num_frames_per_clip=16,
-                   clip_strategy='beginning',
-                   use_streaming=True)
+        return cls(num_videos=2048, num_frames_per_clip=16, clip_strategy='beginning', use_streaming=True)
 
     @classmethod
     def fvd2048_128f(cls) -> 'FVDConfig':
         """Long video protocol: 2048 videos, 128 frames."""
-        return cls(num_videos=2048,
-                   num_frames_per_clip=128,
-                   clip_strategy='beginning',
-                   use_streaming=True)
+        return cls(num_videos=2048, num_frames_per_clip=128, clip_strategy='beginning', use_streaming=True)
 
     @classmethod
     def quick_test(cls) -> 'FVDConfig':
         """Quick test config: 100 videos, 16 frames."""
-        return cls(num_videos=100,
-                   num_frames_per_clip=16,
-                   clip_strategy='beginning')
+        return cls(num_videos=100, num_frames_per_clip=16, clip_strategy='beginning')
 
     def to_dict(self) -> dict:
         """Export config to dict for logging"""
@@ -196,14 +186,10 @@ def load_or_compute_features(videos: str | Path | torch.Tensor,
             max_features = config.num_videos * config.num_clips_per_video
 
             if len(features) < max_features:
-                print(
-                    f"WARNING: Cache has {len(features)} features but need {max_features}"
-                )
+                print(f"WARNING: Cache has {len(features)} features but need {max_features}")
                 print("Cached features insufficient - will recompute...")
             elif len(features) > max_features:
-                print(
-                    f"Using {max_features} features from cache (truncated from {len(features)})"
-                )
+                print(f"Using {max_features} features from cache (truncated from {len(features)})")
                 features = features[:max_features]
                 return features
             else:
@@ -215,17 +201,16 @@ def load_or_compute_features(videos: str | Path | torch.Tensor,
     if isinstance(videos, (str | Path)):
         target_size = (224, 224) if config.resize_before_extraction else None
 
-        video_generator = load_video_clips_streaming(
-            videos,
-            num_frames=config.num_frames_per_clip,
-            max_videos=config.num_videos,
-            clip_strategy=config.clip_strategy,
-            frame_stride=config.frame_stride,
-            num_clips_per_video=config.num_clips_per_video,
-            video_extensions=config.video_extensions,
-            support_frame_dirs=config.support_frame_dirs,
-            target_size=target_size,
-            verbose=True)
+        video_generator = load_video_clips_streaming(videos,
+                                                     num_frames=config.num_frames_per_clip,
+                                                     max_videos=config.num_videos,
+                                                     clip_strategy=config.clip_strategy,
+                                                     frame_stride=config.frame_stride,
+                                                     num_clips_per_video=config.num_clips_per_video,
+                                                     video_extensions=config.video_extensions,
+                                                     support_frame_dirs=config.support_frame_dirs,
+                                                     target_size=target_size,
+                                                     verbose=True)
 
         max_clips = config.num_videos * config.num_clips_per_video
         features = extract_features_streaming(video_generator,
@@ -235,17 +220,14 @@ def load_or_compute_features(videos: str | Path | torch.Tensor,
                                               verbose=True)
     else:
         print(f"Extracting features from {len(videos)} video tensors...")
-        features = extractor.extract_features(videos,
-                                              batch_size=config.batch_size,
-                                              verbose=True)
+        features = extractor.extract_features(videos, batch_size=config.batch_size, verbose=True)
         features = features.numpy()
 
     # Validate feature count
     expected_count = config.num_videos * config.num_clips_per_video
     if len(features) < expected_count:
-        raise ValueError(
-            f"ERROR: Only extracted {len(features)} features, but need {expected_count}!\n"
-            f"Found fewer videos than expected. Check your video directory.")
+        raise ValueError(f"ERROR: Only extracted {len(features)} features, but need {expected_count}!\n"
+                         f"Found fewer videos than expected. Check your video directory.")
     elif len(features) > expected_count:
         print(f"Truncating {len(features)} features to {expected_count}")
         features = features[:expected_count]
@@ -312,9 +294,7 @@ def compute_fvd_with_config(real_videos: str | Path | torch.Tensor,
 
     # Initialize Extractor using Factory
     if verbose:
-        print(
-            f"\nInitializing {config.extractor_model.upper()} model on {config.device}..."
-        )
+        print(f"\nInitializing {config.extractor_model.upper()} model on {config.device}...")
 
     extractor = load_extractor(config.extractor_model, device=config.device)
 
@@ -324,12 +304,11 @@ def compute_fvd_with_config(real_videos: str | Path | torch.Tensor,
         print("Extracting REAL video features...")
         print(f"{'='*70}")
 
-    real_features = load_or_compute_features(
-        videos=real_videos,
-        extractor=extractor,
-        config=config,
-        cache_path=config.cache_real_features,
-        cache_name="real_features")
+    real_features = load_or_compute_features(videos=real_videos,
+                                             extractor=extractor,
+                                             config=config,
+                                             cache_path=config.cache_real_features,
+                                             cache_name="real_features")
 
     if verbose:
         print(f"\n{'='*70}")

benchmarks/fvd/i3d_model.py

@@ -19,16 +19,12 @@ class I3DFeatureExtractor(nn.Module):
     REPO_ID = 'flateon/FVD-I3D-torchscript'
     MODEL_FILENAME = 'i3d_torchscript.pt'
 
-    def __init__(self,
-                 device: str = 'cuda',
-                 cache_dir: str | Path | None = None):
+    def __init__(self, device: str = 'cuda', cache_dir: str | Path | None = None):
         super().__init__()
 
         self.device_str = device
         if device == 'cuda' and not torch.cuda.is_available():
-            print(
-                "Warning: CUDA requested but not available – falling back to CPU"
-            )
+            print("Warning: CUDA requested but not available – falling back to CPU")
             self.device = torch.device('cpu')
         else:
             self.device = torch.device(device)
@@ -51,9 +47,7 @@ class I3DFeatureExtractor(nn.Module):
 
         try:
             # Download model from Hugging Face Hub
-            model_path = hf_hub_download(repo_id=self.REPO_ID,
-                                         filename=self.MODEL_FILENAME,
-                                         cache_dir=self.cache_dir)
+            model_path = hf_hub_download(repo_id=self.REPO_ID, filename=self.MODEL_FILENAME, cache_dir=self.cache_dir)
 
             # Load directly to chosen device
             model = torch.jit.load(model_path, map_location=self.device)
@@ -61,10 +55,9 @@ class I3DFeatureExtractor(nn.Module):
             return model
 
         except Exception as e:
-            raise RuntimeError(
-                f"Failed to load I3D model from Hugging Face Hub. Error: {e}\n"
-                f"Ensure you have internet connection and huggingface_hub installed:\n"
-                f"pip install huggingface_hub") from e
+            raise RuntimeError(f"Failed to load I3D model from Hugging Face Hub. Error: {e}\n"
+                               f"Ensure you have internet connection and huggingface_hub installed:\n"
+                               f"pip install huggingface_hub") from e
 
     def preprocess(self, videos: torch.Tensor) -> torch.Tensor:
         """
@@ -88,10 +81,7 @@ class I3DFeatureExtractor(nn.Module):
         # Resize to 224x224 if needed
         if H != 224 or W != 224:
             videos = videos.reshape(B * T, C, H, W)
-            videos = F.interpolate(videos,
-                                   size=(224, 224),
-                                   mode='bilinear',
-                                   align_corners=False)
+            videos = F.interpolate(videos, size=(224, 224), mode='bilinear', align_corners=False)
             videos = videos.reshape(B, T, C, 224, 224)
 
         # Convert to [B, C, T, H, W] format
@@ -100,10 +90,7 @@ class I3DFeatureExtractor(nn.Module):
         return videos
 
     @torch.no_grad()
-    def extract_features(self,
-                         videos: torch.Tensor,
-                         batch_size: int = 32,
-                         verbose: bool = True) -> torch.Tensor:
+    def extract_features(self, videos: torch.Tensor, batch_size: int = 32, verbose: bool = True) -> torch.Tensor:
         """
         Extract I3D features
         
@@ -127,16 +114,11 @@ class I3DFeatureExtractor(nn.Module):
             batch = self.preprocess(batch)  # Now returns [B, C, T, H, W]
 
             # Use the HF model without rescale/resize (we handle it in preprocess)
-            features = self.model(batch,
-                                  rescale=False,
-                                  resize=False,
-                                  return_features=True)
+            features = self.model(batch, rescale=False, resize=False, return_features=True)
 
             all_features.append(features.cpu())
 
         return torch.cat(all_features, dim=0)
 
-    def __call__(self,
-                 videos: torch.Tensor,
-                 batch_size: int = 32) -> torch.Tensor:
+    def __call__(self, videos: torch.Tensor, batch_size: int = 32) -> torch.Tensor:
         return self.extract_features(videos, batch_size=batch_size)

benchmarks/fvd/run_fvd.py

@@ -36,10 +36,7 @@ def main() -> None:
                 cache_real_features=str(script_dir / f'fvd-cache/{model_name}'),
             )
 
-            results = compute_fvd_with_config(real_dir,
-                                              gen_dir,
-                                              cfg,
-                                              verbose=False)
+            results = compute_fvd_with_config(real_dir, gen_dir, cfg, verbose=False)
             print(f"FVD: {results['fvd']}\nModel: {results['model']}")
 
         except Exception as e:

benchmarks/fvd/validate_fvd.py

@@ -59,10 +59,7 @@ def validate_fvd(subset_a: Path, subset_b: Path, num_videos: int):
     print("TEST 1: Identity Test")
     print("=" * 70)
 
-    result1 = compute_fvd_with_config(real_videos=str(subset_a),
-                                      gen_videos=str(subset_a),
-                                      config=config,
-                                      verbose=False)
+    result1 = compute_fvd_with_config(real_videos=str(subset_a), gen_videos=str(subset_a), config=config, verbose=False)
     fvd_identity = result1['fvd']
     print(f"\nIdentity FVD: {fvd_identity:.2f}")
 
@@ -70,10 +67,7 @@ def validate_fvd(subset_a: Path, subset_b: Path, num_videos: int):
     print("TEST 2: Real vs Real")
     print("=" * 70)
 
-    result2 = compute_fvd_with_config(real_videos=str(subset_a),
-                                      gen_videos=str(subset_b),
-                                      config=config,
-                                      verbose=False)
+    result2 = compute_fvd_with_config(real_videos=str(subset_a), gen_videos=str(subset_b), config=config, verbose=False)
     fvd_real = result2['fvd']
     print(f"\nReal vs Real FVD: {fvd_real:.2f}")
 
@@ -87,9 +81,7 @@ def validate_fvd(subset_a: Path, subset_b: Path, num_videos: int):
 def main() -> None:
     bair_dir = Path('benchmarks/data/bair_full_videos')
 
-    subset_a, subset_b, count = split_videos(bair_dir,
-                                             n_per_subset=128,
-                                             seed=42)
+    subset_a, subset_b, count = split_videos(bair_dir, n_per_subset=128, seed=42)
     validate_fvd(subset_a, subset_b, count)
 
 

benchmarks/fvd/video_utils.py

@@ -54,8 +54,7 @@ def _load_video_cv2(video_path: str | Path,
             raise RuntimeError(f"Video has 0 frames: {video_path}")
 
         frames = np.stack(frames)  # [T, H, W, C]
-        frames = torch.from_numpy(frames).permute(0, 3, 1,
-                                                  2).float()  # [T, C, H, W]
+        frames = torch.from_numpy(frames).permute(0, 3, 1, 2).float()  # [T, C, H, W]
         return frames
 
     if total_frames == 0:
@@ -63,14 +62,11 @@ def _load_video_cv2(video_path: str | Path,
 
     # Determine frame indices for sampling
     if total_frames < num_frames:
-        frame_indices = list(range(
-            total_frames)) + [total_frames - 1] * (num_frames - total_frames)
+        frame_indices = list(range(total_frames)) + [total_frames - 1] * (num_frames - total_frames)
     elif sample_strategy == 'uniform':
-        frame_indices = np.linspace(0, total_frames - 1, num_frames,
-                                    dtype=int).tolist()
+        frame_indices = np.linspace(0, total_frames - 1, num_frames, dtype=int).tolist()
     elif sample_strategy == 'random':
-        frame_indices = sorted(
-            np.random.choice(total_frames, num_frames, replace=False))
+        frame_indices = sorted(np.random.choice(total_frames, num_frames, replace=False))
     else:
         raise ValueError(f"Unknown sample_strategy: {sample_strategy}")
 
@@ -94,17 +90,15 @@ def _load_video_cv2(video_path: str | Path,
     cap.release()
 
     frames = np.stack(frames)  # [T, H, W, C]
-    frames = torch.from_numpy(frames).permute(0, 3, 1,
-                                              2).float()  # [T, C, H, W]
+    frames = torch.from_numpy(frames).permute(0, 3, 1, 2).float()  # [T, C, H, W]
 
     return frames
 
 
-def _load_video_from_frames(
-        frame_dir: str | Path,
-        num_frames: int | None = 16,
-        sample_strategy: str = 'uniform',
-        frame_extensions: list[str] | None = None) -> torch.Tensor:
+def _load_video_from_frames(frame_dir: str | Path,
+                            num_frames: int | None = 16,
+                            sample_strategy: str = 'uniform',
+                            frame_extensions: list[str] | None = None) -> torch.Tensor:
     """
     Load video from directory of frame images.
     
@@ -131,9 +125,7 @@ def _load_video_from_frames(
         frame_files.extend(frame_dir.glob(f"*{ext}"))
 
     if len(frame_files) == 0:
-        raise ValueError(
-            f"No frames found in {frame_dir} with extensions {frame_extensions}"
-        )
+        raise ValueError(f"No frames found in {frame_dir} with extensions {frame_extensions}")
 
     frame_files = sorted(frame_files, key=lambda x: x.name)
     total_frames = len(frame_files)
@@ -143,16 +135,11 @@ def _load_video_from_frames(
         frame_indices = list(range(total_frames))
     else:
         if total_frames < num_frames:
-            frame_indices = list(range(total_frames)) + [total_frames - 1] * (
-                num_frames - total_frames)
+            frame_indices = list(range(total_frames)) + [total_frames - 1] * (num_frames - total_frames)
         elif sample_strategy == 'uniform':
-            frame_indices = np.linspace(0,
-                                        total_frames - 1,
-                                        num_frames,
-                                        dtype=int).tolist()
+            frame_indices = np.linspace(0, total_frames - 1, num_frames, dtype=int).tolist()
         elif sample_strategy == 'random':
-            frame_indices = sorted(
-                np.random.choice(total_frames, num_frames, replace=False))
+            frame_indices = sorted(np.random.choice(total_frames, num_frames, replace=False))
         else:
             raise ValueError(f"Unknown sample_strategy: {sample_strategy}")
 
@@ -170,8 +157,7 @@ def _load_video_from_frames(
 
     # Stack and convert to tensor
     frames = np.stack(frames)  # [T, H, W, C]
-    frames = torch.from_numpy(frames).permute(0, 3, 1,
-                                              2).float()  # [T, C, H, W]
+    frames = torch.from_numpy(frames).permute(0, 3, 1, 2).float()  # [T, C, H, W]
 
     return frames
 
@@ -226,13 +212,12 @@ def load_video_auto(video_path: str | Path,
         raise ValueError(f"Unknown video format at {video_path}")
 
 
-def sample_clips_from_video(
-        video: torch.Tensor,
-        num_frames_per_clip: int = 16,
-        num_clips: int = 1,
-        strategy: str | ClipSamplingStrategy = ClipSamplingStrategy.BEGINNING,
-        frame_stride: int = 1,
-        temporal_stride: int = 1) -> list[torch.Tensor]:
+def sample_clips_from_video(video: torch.Tensor,
+                            num_frames_per_clip: int = 16,
+                            num_clips: int = 1,
+                            strategy: str | ClipSamplingStrategy = ClipSamplingStrategy.BEGINNING,
+                            frame_stride: int = 1,
+                            temporal_stride: int = 1) -> list[torch.Tensor]:
     """
     Sample clips from a video with various strategies.
         
@@ -296,10 +281,7 @@ def sample_clips_from_video(
     elif strategy == ClipSamplingStrategy.RANDOM:
         # Sample N random clips
         for _ in range(num_clips):
-            if effective_clip_length == T:
-                start = 0
-            else:
-                start = np.random.randint(0, T - effective_clip_length + 1)
+            start = 0 if effective_clip_length == T else np.random.randint(0, T - effective_clip_length + 1)
             clip = video[start:start + effective_clip_length]
             clips.append(clip)
 
@@ -312,8 +294,7 @@ def sample_clips_from_video(
             clips.append(clip)
         else:
             # Multiple uniformly spaced clips
-            step = (T - effective_clip_length) / (num_clips -
-                                                  1) if num_clips > 1 else 0
+            step = (T - effective_clip_length) / (num_clips - 1) if num_clips > 1 else 0
             for i in range(num_clips):
                 start = int(i * step)
                 start = min(start, T - effective_clip_length)
@@ -422,15 +403,12 @@ def load_video_clips_streaming(directory: str | Path,
     failed_count = 0
     total_clips = 0
 
-    iterator = tqdm(video_paths,
-                    desc="Loading videos") if verbose else video_paths
+    iterator = tqdm(video_paths, desc="Loading videos") if verbose else video_paths
 
     for video_path in iterator:
         try:
             # Load full video
-            video = load_video_auto(video_path,
-                                    num_frames=None,
-                                    sample_strategy='uniform')
+            video = load_video_auto(video_path, num_frames=None, sample_strategy='uniform')
 
             # Sample clips from video
             clips = sample_clips_from_video(video,
@@ -445,18 +423,14 @@ def load_video_clips_streaming(directory: str | Path,
                     T, C, H, W = clip.shape
                     if target_size != (H, W):
                         # Resize to target size
-                        clip = clip.contiguous(
-                        )  # Fix non-contiguous tensors first
-                        clip_flat = clip.view(T * C, H,
-                                              W).unsqueeze(0)  # [1, T*C, H, W]
-                        clip_resized = torch.nn.functional.interpolate(
-                            clip_flat,
-                            size=target_size,
-                            mode='bilinear',
-                            align_corners=False)
-                        clip = clip_resized.squeeze(0).view(
-                            T, C, target_size[0],
-                            target_size[1])  # Back to [T, C, H, W]
+                        clip = clip.contiguous()  # Fix non-contiguous tensors first
+                        clip_flat = clip.view(T * C, H, W).unsqueeze(0)  # [1, T*C, H, W]
+                        clip_resized = torch.nn.functional.interpolate(clip_flat,
+                                                                       size=target_size,
+                                                                       mode='bilinear',
+                                                                       align_corners=False)
+                        clip = clip_resized.squeeze(0).view(T, C, target_size[0],
+                                                            target_size[1])  # Back to [T, C, H, W]
                     resized_clips.append(clip)
                 clips = resized_clips
 
@@ -480,11 +454,7 @@ def load_video_clips_streaming(directory: str | Path,
 
     failure_rate = failed_count / len(video_paths)
     if failure_rate > 0.1:  # More than 10% failed
-        print(
-            f"\nWARNING: {failure_rate:.1%} of videos failed to load ({failed_count}/{len(video_paths)})"
-        )
+        print(f"\nWARNING: {failure_rate:.1%} of videos failed to load ({failed_count}/{len(video_paths)})")
 
     if verbose:
-        print(
-            f"\nSuccessfully loaded {total_clips} clips from {len(video_paths) - failed_count} videos"
-        )
+        print(f"\nSuccessfully loaded {total_clips} clips from {len(video_paths) - failed_count} videos")

collect_env.py

@@ -9,11 +9,12 @@ import datetime
 import locale
 import os
 import re
+import shutil
 import subprocess
 import sys
-# Unlike the rest of the PyTorch this file must be python2 compliant.
-# This script outputs relevant system environment info
-# Run it with `python collect_env.py` or `python -m torch.utils.collect_env`
+# This script outputs relevant system environment info.
+# Run it with: python collect_env.py
+# Requires Python 3.10+ (matches fastvideo); uses shutil.which (Python 3.3+).
 from collections import namedtuple
 
 from fastvideo.envs import environment_variables
@@ -99,10 +100,7 @@ DEFAULT_PIP_PATTERNS = {
 def run(command):
     """Return (return-code, stdout, stderr)."""
     shell = True if type(command) is str else False
-    p = subprocess.Popen(command,
-                         stdout=subprocess.PIPE,
-                         stderr=subprocess.PIPE,
-                         shell=shell)
+    p = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=shell)
     raw_output, raw_err = p.communicate()
     rc = p.returncode
     if get_platform() == 'win32':
@@ -156,8 +154,7 @@ def get_conda_packages(run_lambda, patterns=None):
         return out
 
     return "\n".join(line for line in out.splitlines()
-                     if not line.startswith("#") and any(name in line
-                                                         for name in patterns))
+                     if not line.startswith("#") and any(name in line for name in patterns))
 
 
 def get_gcc_version(run_lambda):
@@ -165,27 +162,24 @@ def get_gcc_version(run_lambda):
 
 
 def get_clang_version(run_lambda):
-    return run_and_parse_first_match(run_lambda, 'clang --version',
-                                     r'clang version (.*)')
+    return run_and_parse_first_match(run_lambda, 'clang --version', r'clang version (.*)')
 
 
 def get_cmake_version(run_lambda):
-    return run_and_parse_first_match(run_lambda, 'cmake --version',
-                                     r'cmake (.*)')
+    return run_and_parse_first_match(run_lambda, 'cmake --version', r'cmake (.*)')
 
 
 def get_nvidia_driver_version(run_lambda):
     if get_platform() == 'darwin':
         cmd = 'kextstat | grep -i cuda'
-        return run_and_parse_first_match(run_lambda, cmd,
-                                         r'com[.]nvidia[.]CUDA [(](.*?)[)]')
+        return run_and_parse_first_match(run_lambda, cmd, r'com[.]nvidia[.]CUDA [(](.*?)[)]')
     smi = get_nvidia_smi()
     return run_and_parse_first_match(run_lambda, smi, r'Driver Version: (.*?) ')
 
 
 def get_gpu_info(run_lambda):
-    if get_platform() == 'darwin' or (TORCH_AVAILABLE and hasattr(
-            torch.version, 'hip') and torch.version.hip is not None):
+    if get_platform() == 'darwin' or (TORCH_AVAILABLE and hasattr(torch.version, 'hip')
+                                      and torch.version.hip is not None):
         if TORCH_AVAILABLE and torch.cuda.is_available():
             if torch.version.hip is not None:
                 prop = torch.cuda.get_device_properties(0)
@@ -207,8 +201,7 @@ def get_gpu_info(run_lambda):
 
 
 def get_running_cuda_version(run_lambda):
-    return run_and_parse_first_match(run_lambda, 'nvcc --version',
-                                     r'release .+ V(.*)')
+    return run_and_parse_first_match(run_lambda, 'nvcc --version', r'release .+ V(.*)')
 
 
 def get_cudnn_version(run_lambda):
@@ -254,8 +247,7 @@ def get_nvidia_smi():
     if get_platform() == 'win32':
         system_root = os.environ.get('SYSTEMROOT', 'C:\\Windows')
         program_files_root = os.environ.get('PROGRAMFILES', 'C:\\Program Files')
-        legacy_path = os.path.join(program_files_root, 'NVIDIA Corporation',
-                                   'NVSMI', smi)
+        legacy_path = os.path.join(program_files_root, 'NVIDIA Corporation', 'NVSMI', smi)
         new_path = os.path.join(system_root, 'System32', smi)
         smis = [new_path, legacy_path]
         for candidate_smi in smis:
@@ -387,10 +379,8 @@ def get_cpu_info(run_lambda):
     if get_platform() == 'linux':
         rc, out, err = run_lambda('lscpu')
     elif get_platform() == 'win32':
-        rc, out, err = run_lambda(
-            'wmic cpu get Name,Manufacturer,Family,Architecture,ProcessorType,DeviceID, \
-        CurrentClockSpeed,MaxClockSpeed,L2CacheSize,L2CacheSpeed,Revision /VALUE'
-        )
+        rc, out, err = run_lambda('wmic cpu get Name,Manufacturer,Family,Architecture,ProcessorType,DeviceID, \
+        CurrentClockSpeed,MaxClockSpeed,L2CacheSize,L2CacheSpeed,Revision /VALUE')
     elif get_platform() == 'darwin':
         rc, out, err = run_lambda("sysctl -n machdep.cpu.brand_string")
     cpu_info = 'None'
@@ -415,27 +405,22 @@ def get_platform():
 
 
 def get_mac_version(run_lambda):
-    return run_and_parse_first_match(run_lambda, 'sw_vers -productVersion',
-                                     r'(.*)')
+    return run_and_parse_first_match(run_lambda, 'sw_vers -productVersion', r'(.*)')
 
 
 def get_windows_version(run_lambda):
     system_root = os.environ.get('SYSTEMROOT', 'C:\\Windows')
     wmic_cmd = os.path.join(system_root, 'System32', 'Wbem', 'wmic')
     findstr_cmd = os.path.join(system_root, 'System32', 'findstr')
-    return run_and_read_all(
-        run_lambda,
-        '{} os get Caption | {} /v Caption'.format(wmic_cmd, findstr_cmd))
+    return run_and_read_all(run_lambda, '{} os get Caption | {} /v Caption'.format(wmic_cmd, findstr_cmd))
 
 
 def get_lsb_version(run_lambda):
-    return run_and_parse_first_match(run_lambda, 'lsb_release -a',
-                                     r'Description:\t(.*)')
+    return run_and_parse_first_match(run_lambda, 'lsb_release -a', r'Description:\t(.*)')
 
 
 def check_release_file(run_lambda):
-    return run_and_parse_first_match(run_lambda, 'cat /etc/*-release',
-                                     r'PRETTY_NAME="(.*)"')
+    return run_and_parse_first_match(run_lambda, 'cat /etc/*-release', r'PRETTY_NAME="(.*)"')
 
 
 def get_os(run_lambda):
@@ -495,17 +480,13 @@ def get_pip_packages(run_lambda, patterns=None):
 
         if pip_available:
             cmd = [sys.executable, '-mpip', 'list', '--format=freeze']
-        elif os.environ.get("UV") is not None:
-            print("uv is set")
+        elif shutil.which("uv") is not None:
             cmd = ["uv", "pip", "list", "--format=freeze"]
         else:
-            raise RuntimeError(
-                "Could not collect pip list output (pip or uv module not available)"
-            )
+            raise RuntimeError("Could not collect pip list output (pip or uv module not available)")
 
         out = run_and_read_all(run_lambda, cmd)
-        return "\n".join(line for line in out.splitlines()
-                         if any(name in line for name in patterns))
+        return "\n".join(line for line in out.splitlines() if any(name in line for name in patterns))
 
     pip_version = 'pip3' if sys.version[0] == '3' else 'pip'
     out = run_with_pip()
@@ -537,8 +518,7 @@ def is_xnnpack_available():
 def get_env_vars():
     env_vars = ''
     secret_terms = ('secret', 'token', 'api', 'access', 'password')
-    report_prefix = ("TORCH", "NCCL", "PYTORCH", "CUDA", "CUBLAS", "CUDNN",
-                     "OMP_", "MKL_", "NVIDIA")
+    report_prefix = ("TORCH", "NCCL", "PYTORCH", "CUDA", "CUBLAS", "CUDNN", "OMP_", "MKL_", "NVIDIA")
     for k, v in os.environ.items():
         if any(term in k.lower() for term in secret_terms):
             continue
@@ -559,8 +539,7 @@ def get_env_info():
         debug_mode_str = str(torch.version.debug)
         cuda_available_str = str(torch.cuda.is_available())
         cuda_version_str = torch.version.cuda
-        if not hasattr(torch.version,
-                       'hip') or torch.version.hip is None:  # cuda version
+        if not hasattr(torch.version, 'hip') or torch.version.hip is None:  # cuda version
             hip_compiled_version = hip_runtime_version = miopen_runtime_version = 'N/A'
         else:  # HIP version
 
@@ -588,9 +567,8 @@ def get_env_info():
     return SystemEnv(
         torch_version=version_str,
         is_debug_build=debug_mode_str,
-        python_version='{} ({}-bit runtime)'.format(
-            sys_version,
-            sys.maxsize.bit_length() + 1),
+        python_version='{} ({}-bit runtime)'.format(sys_version,
+                                                    sys.maxsize.bit_length() + 1),
         python_platform=get_python_platform(),
         is_cuda_available=cuda_available_str,
         cuda_compiled_version=cuda_version_str,
@@ -714,10 +692,8 @@ def pretty_str(envinfo):
         'nvidia_driver_version',
     ]
     all_cuda_fields = dynamic_cuda_fields + ['cudnn_version']
-    all_dynamic_cuda_fields_missing = all(mutable_dict[field] is None
-                                          for field in dynamic_cuda_fields)
-    if TORCH_AVAILABLE and not torch.cuda.is_available(
-    ) and all_dynamic_cuda_fields_missing:
+    all_dynamic_cuda_fields_missing = all(mutable_dict[field] is None for field in dynamic_cuda_fields)
+    if TORCH_AVAILABLE and not torch.cuda.is_available() and all_dynamic_cuda_fields_missing:
         for field in all_cuda_fields:
             mutable_dict[field] = 'No CUDA'
         if envinfo.cuda_compiled_version is None:
@@ -730,19 +706,15 @@ def pretty_str(envinfo):
     mutable_dict = replace_nones(mutable_dict)
 
     # If either of these are '', replace with 'No relevant packages'
-    mutable_dict['pip_packages'] = replace_if_empty(
-        mutable_dict['pip_packages'])
-    mutable_dict['conda_packages'] = replace_if_empty(
-        mutable_dict['conda_packages'])
+    mutable_dict['pip_packages'] = replace_if_empty(mutable_dict['pip_packages'])
+    mutable_dict['conda_packages'] = replace_if_empty(mutable_dict['conda_packages'])
 
     # Tag conda and pip packages with a prefix
     # If they were previously None, they'll show up as ie '[conda] Could not collect'
     if mutable_dict['pip_packages']:
-        mutable_dict['pip_packages'] = prepend(
-            mutable_dict['pip_packages'], '[{}] '.format(envinfo.pip_version))
+        mutable_dict['pip_packages'] = prepend(mutable_dict['pip_packages'], '[{}] '.format(envinfo.pip_version))
     if mutable_dict['conda_packages']:
-        mutable_dict['conda_packages'] = prepend(mutable_dict['conda_packages'],
-                                                 '[conda] ')
+        mutable_dict['conda_packages'] = prepend(mutable_dict['conda_packages'], '[conda] ')
     mutable_dict['cpu_info'] = envinfo.cpu_info
     return env_info_fmt.format(**mutable_dict)
 
@@ -756,18 +728,13 @@ def main():
     output = get_pretty_env_info()
     print(output)
 
-    if TORCH_AVAILABLE and hasattr(torch, 'utils') and hasattr(
-            torch.utils, '_crash_handler'):
+    if TORCH_AVAILABLE and hasattr(torch, 'utils') and hasattr(torch.utils, '_crash_handler'):
         minidump_dir = torch.utils._crash_handler.DEFAULT_MINIDUMP_DIR
         if sys.platform == "linux" and os.path.exists(minidump_dir):
-            dumps = [
-                os.path.join(minidump_dir, dump)
-                for dump in os.listdir(minidump_dir)
-            ]
+            dumps = [os.path.join(minidump_dir, dump) for dump in os.listdir(minidump_dir)]
             latest = max(dumps, key=os.path.getctime)
             ctime = os.path.getctime(latest)
-            creation_time = datetime.datetime.fromtimestamp(ctime).strftime(
-                '%Y-%m-%d %H:%M:%S')
+            creation_time = datetime.datetime.fromtimestamp(ctime).strftime('%Y-%m-%d %H:%M:%S')
             msg = "\n*** Detected a minidump at {} created on {}, ".format(latest, creation_time) + \
                   "if this is related to your bug please include it when you file a report ***"
             print(msg, file=sys.stderr)

comfyui/__init__.py

@@ -1,5 +1,4 @@
-from .video_generator.nodes import (NODE_CLASS_MAPPINGS,
-                                    NODE_DISPLAY_NAME_MAPPINGS)
+from .video_generator.nodes import (NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS)
 
 WEB_DIRECTORY = "./web"
-__all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS', 'WEB_DIRECTORY']
+__all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS', 'WEB_DIRECTORY']

comfyui/examples/FastHunyuan-diffusers.json

@@ -443,7 +443,6 @@
         1025,
         "fixed",
         24,
-        -99999,
         -99999
       ],
       "auto_widget_states": {
@@ -491,11 +490,6 @@
           "isAuto": true,
           "value": -99999,
           "cachedValue": "X://insert/path/here.mp4"
-        },
-        "enable_teacache": {
-          "isAuto": true,
-          "value": -99999,
-          "cachedValue": true
         }
       }
     },
@@ -642,4 +636,4 @@
     "VHS_KeepIntermediate": true
   },
   "version": 0.4
-}
+}

comfyui/examples/Wan2.1-I2V-14B-480P-Diffusers.json

@@ -353,8 +353,7 @@
         1024,
         "fixed",
         24,
-        "X://insert/path/here.mp4",
-        true
+        "X://insert/path/here.mp4"
       ],
       "auto_widget_states": {
         "height": {
@@ -401,11 +400,6 @@
           "isAuto": true,
           "value": "X://insert/path/here.mp4",
           "cachedValue": "X://insert/path/here.mp4"
-        },
-        "enable_teacache": {
-          "isAuto": true,
-          "value": true,
-          "cachedValue": true
         }
       }
     },
@@ -694,4 +688,4 @@
     "VHS_KeepIntermediate": true
   },
   "version": 0.4
-}
+}

comfyui/video_generator/inference_args.py

@@ -31,9 +31,6 @@ class InferenceArgs:
                 "image_path": ("STRING", {
                     "default": "X://insert/path/here.mp4"
                 }),
-                "enable_teacache": ([True, False], {
-                    "default": False
-                }),
             }
         }
 
@@ -57,7 +54,6 @@ class InferenceArgs:
         seed,
         fps,
         image_path,
-        enable_teacache,
     ):
         raw_args = {
             "height": height,
@@ -69,7 +65,6 @@ class InferenceArgs:
             "seed": seed,
             "fps": fps,
             "image_path": image_path,
-            "enable_teacache": enable_teacache,
         }
 
         # Filter out keys where value is -99999, handling different types properly

comfyui/video_generator/load_image.py

@@ -14,10 +14,7 @@ class LoadImagePath:
     @classmethod
     def INPUT_TYPES(s):
         input_dir = folder_paths.get_input_directory()
-        files = [
-            f for f in os.listdir(input_dir)
-            if os.path.isfile(os.path.join(input_dir, f))
-        ]
+        files = [f for f in os.listdir(input_dir) if os.path.isfile(os.path.join(input_dir, f))]
         files = folder_paths.filter_files_content_types(files, ["image"])
         return {
             "required": {
@@ -65,13 +62,10 @@ class LoadImagePath:
                 None,
             ]
             if 'A' in processed_image.getbands():
-                mask = np.array(processed_image.getchannel('A')).astype(
-                    np.float32) / 255.0
+                mask = np.array(processed_image.getchannel('A')).astype(np.float32) / 255.0
                 mask = 1. - torch.from_numpy(mask)
             elif processed_image.mode == 'P' and 'transparency' in processed_image.info:
-                mask = np.array(
-                    processed_image.convert('RGBA').getchannel('A')).astype(
-                        np.float32) / 255.0
+                mask = np.array(processed_image.convert('RGBA').getchannel('A')).astype(np.float32) / 255.0
                 mask = 1. - torch.from_numpy(mask)
             else:
                 mask = torch.zeros((64, 64), dtype=torch.float32, device="cpu")

comfyui/video_generator/node_helpers.py

@@ -9,8 +9,7 @@ from PIL import ImageFile, UnidentifiedImageError
 T = TypeVar('T')
 
 
-def conditioning_set_values(conditioning: list[Any],
-                            values: dict[str, Any] | None = None) -> list[Any]:
+def conditioning_set_values(conditioning: list[Any], values: dict[str, Any] | None = None) -> list[Any]:
     if values is None:
         values = {}
     c = []
@@ -27,8 +26,7 @@ def pillow(fn: Callable[[Any], T], arg: Any) -> T:
     prev_value = None
     try:
         x = fn(arg)
-    except (OSError, UnidentifiedImageError, ValueError
-            ):  #PIL issues #4472 and #2445, also fixes ComfyUI issue #3416
+    except (OSError, UnidentifiedImageError, ValueError):  #PIL issues #4472 and #2445, also fixes ComfyUI issue #3416
         prev_value = ImageFile.LOAD_TRUNCATED_IMAGES
         ImageFile.LOAD_TRUNCATED_IMAGES = True
         x = fn(arg)
@@ -39,12 +37,7 @@ def pillow(fn: Callable[[Any], T], arg: Any) -> T:
 
 
 def hasher() -> Callable[[], Any]:
-    hashfuncs = {
-        "md5": hashlib.md5,
-        "sha1": hashlib.sha1,
-        "sha256": hashlib.sha256,
-        "sha512": hashlib.sha512
-    }
+    hashfuncs = {"md5": hashlib.md5, "sha1": hashlib.sha1, "sha256": hashlib.sha256, "sha512": hashlib.sha512}
     return hashfuncs[args.default_hashing_function]
 
 
@@ -58,8 +51,7 @@ def string_to_torch_dtype(string: str) -> torch.dtype | None:
     return None
 
 
-def image_alpha_fix(destination: torch.Tensor,
-                    source: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
+def image_alpha_fix(destination: torch.Tensor, source: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
     if destination.shape[-1] < source.shape[-1]:
         source = source[..., :destination.shape[-1]]
     elif destination.shape[-1] > source.shape[-1]:

comfyui/video_generator/text_encoder_config.py

@@ -26,11 +26,7 @@ class TextEncoderConfig:
     CATEGORY = "fastvideo"
 
     def set_args(self, prefix, quant_config, lora_config):
-        raw_args = {
-            "prefix": prefix,
-            "quant_config": quant_config,
-            "lora_config": lora_config
-        }
+        raw_args = {"prefix": prefix, "quant_config": quant_config, "lora_config": lora_config}
 
         # Filter out keys where value is -99999
         args = {k: v for k, v in raw_args.items() if str(int(v)) != str(-99999)}

comfyui/video_generator/video_generator.py

@@ -13,11 +13,7 @@ from comfy.model_management import processing_interrupted
 from fastvideo import PipelineConfig
 from fastvideo import VideoGenerator as FastVideoGenerator
 
-sys.path.insert(
-    0,
-    os.path.dirname(
-        os.path.dirname(
-            os.path.dirname(os.path.dirname(os.path.abspath(__file__))))))
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))))
 
 
 # Custom exception for interruption
@@ -28,8 +24,7 @@ class GenerationInterruptedException(Exception):
 # Custom exception for interruption that ComfyUI will recognize
 class GenerationCancelledException(Exception):
 
-    def __init__(self,
-                 message: str = "Generation was cancelled by user") -> None:
+    def __init__(self, message: str = "Generation was cancelled by user") -> None:
         self.message = message
         super().__init__(self.message)
 
@@ -139,8 +134,7 @@ class VideoGenerator:
                 self._generation_interrupted = True
 
                 # Try to send interrupt signal to worker processes
-                if self.generator is not None and hasattr(
-                        self.generator, 'executor'):
+                if self.generator is not None and hasattr(self.generator, 'executor'):
                     try:
                         # The MultiprocExecutor has a workers attribute
                         if hasattr(self.generator.executor, 'workers'):
@@ -156,16 +150,12 @@ class VideoGenerator:
                 break
             time.sleep(0.5)
 
-    def _run_generation(self, prompt: str, output_path: str,
-                        inference_args: dict[str, Any]) -> None:
+    def _run_generation(self, prompt: str, output_path: str, inference_args: dict[str, Any]) -> None:
         """Thread function to run the generation"""
         try:
             if self.generator is not None:
-                self.generator.generate_video(prompt=prompt,
-                                              output_path=output_path,
-                                              **inference_args)
-                self._generation_result = os.path.join(output_path,
-                                                       f"{prompt[:100]}.mp4")
+                self.generator.generate_video(prompt=prompt, output_path=output_path, **inference_args)
+                self._generation_result = os.path.join(output_path, f"{prompt[:100]}.mp4")
             else:
                 raise RuntimeError("Generator is not initialized")
         except Exception as e:
@@ -226,8 +216,7 @@ class VideoGenerator:
             update_config_from_args(pipeline_config.vae_config, vae_config)
 
         if text_encoder_config is not None:
-            update_config_from_args(pipeline_config.text_encoder_configs,
-                                    text_encoder_config)
+            update_config_from_args(pipeline_config.text_encoder_configs, text_encoder_config)
 
         # Update top-level pipeline config with remaining arguments
         raw_pipeline_args = {}
@@ -245,10 +234,7 @@ class VideoGenerator:
             raw_pipeline_args['text_encoder_precision'] = text_encoder_precision
 
         # Filter out any value explicitly set to -99999 (auto values)
-        pipeline_args = {
-            k: v
-            for k, v in raw_pipeline_args.items() if str(int(v)) != str(-99999)
-        }
+        pipeline_args = {k: v for k, v in raw_pipeline_args.items() if str(int(v)) != str(-99999)}
 
         update_config_from_args(pipeline_config, pipeline_args)
 
@@ -262,38 +248,30 @@ class VideoGenerator:
         if dit_cpu_offload is not None:
             raw_generation_args['dit_cpu_offload'] = dit_cpu_offload
 
-        generation_args = {
-            k: v
-            for k, v in raw_generation_args.items()
-            if str(int(v)) != str(-99999)
-        }
+        generation_args = {k: v for k, v in raw_generation_args.items() if str(int(v)) != str(-99999)}
 
         if self.generator is None:
             print('generation_args', generation_args)
             print('pipeline_config', pipeline_config)
-            self.generator = FastVideoGenerator.from_pretrained(
-                model_path=model_path,
-                **generation_args,
-                pipeline_config=pipeline_config)
+            self.generator = FastVideoGenerator.from_pretrained(model_path=model_path,
+                                                                **generation_args,
+                                                                pipeline_config=pipeline_config)
 
         print('inference_args', inference_args)
 
         # Start a thread to run the generation
         self._generation_thread = threading.Thread(target=self._run_generation,
-                                                   args=(prompt, output_path,
-                                                         inference_args),
+                                                   args=(prompt, output_path, inference_args),
                                                    daemon=True)
         self._generation_thread.start()
 
         # Start a background thread to monitor for interruptions
         self._generation_active = True
-        self._interrupt_thread = threading.Thread(
-            target=self._monitor_for_interruption, daemon=True)
+        self._interrupt_thread = threading.Thread(target=self._monitor_for_interruption, daemon=True)
         self._interrupt_thread.start()
 
         # Wait for either completion or interruption
-        while self._generation_thread.is_alive(
-        ) and not self._interrupt_event.is_set():
+        while self._generation_thread.is_alive() and not self._interrupt_event.is_set():
             self._generation_thread.join(timeout=0.5)
 
         self._generation_active = False

comfyui/web/js/FastVideo.core.js

@@ -552,7 +552,7 @@ app.registerExtension({
                 ]
                 const floatWidgetNames = ["embedded_cfg_scale", "guidance_scale"]
                 const comboWidgetNames = ["vae_tiling", "vae_precision", "vae_sp", "text_encoder_precision", "precision",
-                    "load_encoder", "load_decoder", "use_tiling", "use_temporal_tiling", "use_parallel_tiling", "dit_cpu_offload", "enable_teacache"
+                    "load_encoder", "load_decoder", "use_tiling", "use_temporal_tiling", "use_parallel_tiling", "dit_cpu_offload"
                 ]
                 const stringWidgetNames = ["prefix", "quant_config", "lora_config", "image_path"]
 

docs/README.md

@@ -36,4 +36,4 @@ Then open your browser to: http://localhost:8000
 
 ## Automatic Deployment
 
-Documentation is automatically built and deployed to GitHub Pages when changes are pushed to the `main` branch via the `.github/workflows/docs.yml` workflow.
+Documentation is automatically built and deployed to GitHub Pages when changes are pushed to the `main` branch via the `.github/workflows/infra-docs.yml` workflow.

docs/api/fastvideo.md

@@ -62,9 +62,9 @@ This page contains the complete API reference for the FastVideo library.
       show_root_toc_entry: true
       heading_level: 4
 
-#### fastvideo.configs.sample
+#### fastvideo.api.sampling_param
 
-::: fastvideo.configs.sample
+::: fastvideo.api.sampling_param
     options:
       show_source: true
       show_root_heading: true

docs/attention/developer/index.md

@@ -54,7 +54,8 @@ To use this:
 1. **Set Context**: In your pipeline or generation loop, use the `set_forward_context` context manager.
 2. **Access Context**: Inside your attention backend, use `get_forward_context()`.
 
-See [`docs/attention/sta/index.md`](../sta/index.md) (Sliding Tile Attention) for an example of how complex configuration (window sizes) is passed this way.
+See [`docs/attention/sta/index.md`](../sta/index.md) for a legacy STA example
+of passing complex configuration (window sizes) through `ForwardContext`.
 
 ## 3. Adding Compiled Kernels (C++/CUDA)
 

docs/attention/index.md

@@ -5,13 +5,16 @@ FastVideo provides highly optimized custom attention kernels to accelerate video
 ## Supported Kernels
 
 * **[Video Sparse Attention (VSA)](vsa/index.md)**: Sparse attention mechanism selecting top-k blocks.
-* **[Sliding Tile Attention (STA)](sta/index.md)**: Optimized attention for window-based video generation.
+* **[Sliding Tile Attention (STA)](sta/index.md)**: STA kernel support is kept in
+  `fastvideo-kernel`; full FastVideo STA pipeline workflow is archived in
+  `sta_do_not_delete`.
 * **Backend development guide**: See the developer guide at
   [Attention Backend Development](../contributing/attention_backend.md).
 
 ## General Build Instructions
 
-These instructions apply to building the `fastvideo-kernel` package from source, which includes both STA and VSA kernels.
+These instructions apply to building the `fastvideo-kernel` package from
+source, which includes both STA and VSA kernels.
 
 ### Prerequisites
 

docs/attention/sta/index.md

@@ -1,27 +1,84 @@
 # Sliding Tile Attention (STA)
 
-Optimized attention for window-based video generation (e.g., HunyuanVideo).
+STA inference integration is archived from `main`.
 
-## Installation
+The full STA pipeline code (including mask search and STA inference wiring in
+`fastvideo/`) is preserved in:
 
-STA is included in the `fastvideo-kernel` package. See the [main Attention page](../index.md) for build instructions.
+- https://github.com/hao-ai-lab/FastVideo/tree/sta_do_not_delete
 
-## Usage
+In this branch, STA kernels in `fastvideo-kernel` are still kept.
 
-```python
-from fastvideo_kernel import sliding_tile_attention
+## Why STA is not in `main`
 
-# q, k, v: [batch_size, num_heads, seq_length, head_dim]
-# window_size: List of (t, h, w) tiles. Tile size is (6, 8, 8).
-# text_length: Number of text tokens (0-256)
+We do not keep STA pipeline integration in `main` because we believe Video
+Sparse Attention (VSA) is strictly better than STA for the actively maintained
+FastVideo inference path.
 
-out = sliding_tile_attention(
-    q, k, v, 
-    window_size=[(3, 3, 3)],  # Example window
-    text_length=256
-)
+## What to checkout for STA workflows
+
+To run the full STA workflow, switch to the archived branch:
+
+```bash
+git fetch origin
+git checkout sta_do_not_delete
 ```
 
+## Mask Search (archive branch)
+
+The reference script is:
+
+- `examples/inference/sta_mask_search/inference_wan_sta.sh`
+
+It runs two stages:
+
+1. `STA_searching` (full search), output at
+   `inference_results/sta/mask_search_full`
+2. `STA_tuning` (sparse tuning), output at
+   `inference_results/sta/mask_search_sparse`
+
+Run:
+
+```bash
+export FASTVIDEO_ATTENTION_BACKEND=SLIDING_TILE_ATTN
+export FASTVIDEO_ATTENTION_CONFIG=assets/mask_strategy_wan.json
+
+bash examples/inference/sta_mask_search/inference_wan_sta.sh
+```
+
+## STA Inference (archive branch)
+
+With a selected mask strategy, run inference with:
+
+```bash
+export FASTVIDEO_ATTENTION_BACKEND=SLIDING_TILE_ATTN
+export FASTVIDEO_ATTENTION_CONFIG=assets/mask_strategy_wan.json
+
+fastvideo generate \
+  --model-path Wan-AI/Wan2.1-T2V-14B-Diffusers \
+  --num-gpus 2 \
+  --tp-size 2 \
+  --sp-size 2 \
+  --height 768 \
+  --width 1280 \
+  --num-frames 69 \
+  --num-inference-steps 50 \
+  --prompt "A cinematic wildlife shot of a lion walking in golden grasslands." \
+  --output-path outputs_video/STA/
+```
+
+Python usage on the archive branch can also set `STA_mode` in
+`VideoGenerator.from_pretrained(...)`:
+
+- `STA_searching`
+- `STA_tuning`
+- `STA_inference`
+
+## Kernel-level API (current branch)
+
+STA kernels remain available from `fastvideo-kernel`. See
+[Attention overview](../index.md) for build instructions.
+
 ## Citation
 
 If you use Sliding Tile Attention in your research, please cite:

docs/contributing/ci_architecture.md

@@ -0,0 +1,331 @@
+# CI Architecture
+
+## Overview
+
+FastVideo uses a three-tier CI pipeline designed to keep feedback fast on every push while
+protecting `main` through a full GPU regression suite before any merge.
+
+```
+PR push
+  │
+  ├─► Tier 1: Pre-commit (~2 min)
+  │     GitHub Actions / ubuntu-latest
+  │     yapf, ruff, mypy, codespell, pymarkdown, actionlint, check-filenames
+  │
+  └─► Tier 2: Fastcheck (~10-20 min, path-filtered)
+        Buildkite / Modal GPU instances
+        Only runs tests for paths you changed
+
+              │ (developer comments /merge or maintainer adds 'ready' label)
+              ▼
+        Tier 3: Full Suite (~60-90 min)
+          Buildkite / Modal GPU instances
+          All integration, SSIM, training, and performance tests
+          Runs on the PR branch directly
+              │
+          pass ──► Mergify auto-squash-merges to main, branch deleted
+          fail ──► fix the regression, push, and /merge again
+```
+
+---
+
+## CI Tiers
+
+### Tier 1: Pre-commit (every PR push)
+
+| Attribute | Value |
+|-----------|-------|
+| Triggered by | Every push to any PR branch, plus pushes to `main` |
+| Runs on | GitHub Actions, `ubuntu-latest` |
+| Duration | ~2 minutes |
+
+**Checks run** (from `.pre-commit-config.yaml`, stage: `manual`):
+
+| Hook | What it checks |
+|------|---------------|
+| `yapf` | Python code formatting |
+| `ruff` | Python linting and auto-fixable style issues |
+| `codespell` | Spelling errors in code and docs |
+| `pymarkdown` | Markdown formatting |
+| `actionlint` | GitHub Actions workflow syntax |
+| `mypy` | Static type checking (Python 3.10 target) |
+| `check-filenames` | No spaces in tracked filenames |
+
+A failure here means code style or type issues. Run `pre-commit run --all-files` locally to
+replicate CI results before pushing.
+
+---
+
+### Tier 2: Fastcheck (path-filtered, every PR push)
+
+| Attribute | Value |
+|-----------|-------|
+| Triggered by | Every push; the monorepo-diff plugin skips tests for unchanged paths |
+| Runs on | Buildkite, Modal GPU instances |
+| Duration | ~10-20 minutes per test (run in parallel) |
+
+**Tests and their path triggers:**
+
+| Buildkite label | `TEST_TYPE` | Triggers when you change |
+|-----------------|-------------|--------------------------|
+| Encoder Tests | `encoder` | `fastvideo/models/encoders/**`, `fastvideo/models/loader/**`, `fastvideo/tests/encoders/**`, `pyproject.toml`, `docker/Dockerfile.python3.12` |
+| VAE Tests | `vae` | `fastvideo/models/vaes/**`, `fastvideo/models/loader/**`, `fastvideo/tests/vaes/**`, `pyproject.toml`, `docker/Dockerfile.python3.12` |
+| Transformer Tests | `transformer` | `fastvideo/models/dits/**`, `fastvideo/models/loader/**`, `fastvideo/tests/transformers/**`, `fastvideo/layers/**`, `fastvideo/attention/**`, `pyproject.toml`, `docker/Dockerfile.python3.12` |
+| Kernel Tests | `kernel_tests` | `fastvideo-kernel/**`, `pyproject.toml`, `docker/Dockerfile.python3.12` |
+| Unit Tests | `unit_test` | `fastvideo/**`, `.buildkite/**`, `.github/**`, `pyproject.toml`, `docker/Dockerfile.python3.12` |
+
+A Fastcheck failure means a component-level regression. Check the Buildkite build log for the
+failing test's output.
+
+---
+
+### Tier 3: Full Test Suite (triggered by `ready` label)
+
+| Attribute | Value |
+|-----------|-------|
+| Triggered by | Adding the `ready` label to the PR (via `/merge` command), or a `/test full` command |
+| Runs on | Buildkite, Modal GPU instances |
+| Duration | 60-90 minutes total (tests run in parallel, path-filtered) |
+
+**Tests:**
+
+| Buildkite label | `TEST_TYPE` | Timeout |
+|-----------------|-------------|---------|
+| SSIM Tests | `ssim` | 90 min |
+| LoRA Inference Tests | `inference_lora` | 20 min |
+| Training Tests | `training` | 15 min |
+| Distillation DMD Tests | `distillation_dmd` | 15 min |
+| Self-Forcing Tests | `self_forcing` | 15 min |
+| LoRA Training Tests | `training_lora` | 15 min |
+| Training Tests VSA | `training_vsa` | 15 min |
+| Inference Tests VMoBA | `inference_vmoba` | 15 min |
+| Performance Tests | `performance` | 30 min |
+| API Server Tests | `api_server` | 30 min |
+
+If a Full Suite test fails, check the Buildkite build log for the failing step's output.
+Fix the regression, push, and comment `/merge` again to re-trigger.
+
+---
+
+## Auto-merge Flow
+
+Mergify prevents untested code from landing on `main` by gating squash-merge on the Full
+Suite passing directly on the PR branch.
+
+**How it works:**
+
+1. A developer comments `/merge` on an approved PR (or a maintainer adds the `ready` label).
+2. The `ready` label triggers `ci-trigger-full-suite.yml`, which calls the Buildkite API to
+   run the Full Suite on the PR branch itself.
+3. While the Full Suite runs, Mergify also auto-rebases the PR branch against `main` if it
+   is behind and has no conflicts.
+4. Once the Full Suite posts `full-suite-passed`, Mergify checks all **merge conditions**:
+   - `pre-commit` check is green
+   - `fastcheck-passed` check is green
+   - `full-suite-passed` check is green
+   - At least 1 approved review (`#approved-reviews-by>=1`)
+   - PR title starts with a valid `[type]` tag
+   - PR is not a draft
+   - No merge conflicts
+5. If all conditions pass, Mergify squash-merges to `main` automatically. The branch is
+   deleted after merge.
+6. If the Full Suite fails, the developer fixes the issue, pushes, and comments `/merge`
+   again to re-trigger.
+
+**Merge conditions summary:**
+
+| Condition | Meaning |
+|-----------|---------|
+| `check-success~=pre-commit` | Tier 1 pre-commit must be green |
+| `check-success=fastcheck-passed` | Tier 2 Fastcheck must be green |
+| `check-success=full-suite-passed` | Tier 3 Full Suite must be green |
+| `#approved-reviews-by>=1` | At least one approved review |
+| `title~=(?i)^\[(feat|bugfix|...)` | PR title has a valid type tag |
+| `-draft` | PR is not in draft state |
+| `-conflict` | No merge conflicts with base branch |
+| `-closed` | PR is still open |
+
+---
+
+## Label System
+
+Labels are applied automatically. You don't need to set them manually.
+
+### Type Labels (from PR title prefix)
+
+Applied by Mergify based on the `[tag]` at the start of the PR title.
+
+| Label | Matched title prefix | Meaning |
+|-------|---------------------|---------|
+| `type: feat` | `[feat]` or `[feature]` | New feature or capability |
+| `type: bugfix` | `[bugfix]` or `[fix]` | Bug fix |
+| `type: refactor` | `[refactor]` | Code restructuring, no behavior change |
+| `type: perf` | `[perf]` | Performance improvement |
+| `type: ci` | `[ci]` | CI/CD or tooling changes |
+| `type: docs` | `[doc]` or `[docs]` | Documentation only |
+| `type: misc` | `[misc]` or `[chore]` | Housekeeping, dependency bumps |
+| `type: new-model` | `[new-model]` | Adding a new model |
+
+### Scope Labels (from changed files)
+
+Applied by Mergify based on which paths you modified. Multiple scope labels can be added.
+
+| Label | File paths that trigger it |
+|-------|---------------------------|
+| `scope: training` | `fastvideo/train/`, `fastvideo/training/`, `fastvideo/distillation/`, `examples/train/`, `examples/training/`, `examples/distill/` |
+| `scope: inference` | `fastvideo/pipelines/basic/`, `fastvideo/pipelines/stages/`, `fastvideo/pipelines/samplers/`, `fastvideo/entrypoints/`, `fastvideo/worker/`, `fastvideo/api/sampling_param.py`, `fastvideo/configs/pipelines/`, `examples/inference/` |
+| `scope: attention` | `fastvideo/attention/` |
+| `scope: kernel` | `fastvideo-kernel/`, `csrc/` |
+| `scope: data` | `fastvideo/dataset/`, `fastvideo/pipelines/preprocess/`, `examples/preprocessing/` |
+| `scope: infra` | `.github/`, `.buildkite/`, `fastvideo/tests/`, `docker/` |
+| `scope: distributed` | `fastvideo/distributed/` |
+| `scope: docs` | `docs/` |
+| `scope: ui` | `ui/` |
+| `scope: model` | `fastvideo/models/`, `fastvideo/layers/`, `fastvideo/configs/models/` |
+
+### Process Labels
+
+| Label | Who sets it | Meaning |
+|-------|-------------|---------|
+| `ready` | Developer (`/merge` command) or maintainer | Triggers Full Suite and enables auto-merge |
+| `needs-rebase` | Mergify (automatic) | PR has merge conflicts; rebase needed |
+| `do-not-merge` | Maintainer | Blocks queue entry regardless of other conditions |
+
+---
+
+## PR Title Format
+
+All PR titles targeting `main` must start with a bracketed type tag. This is enforced by a
+Mergify merge protection rule and is required before a PR can be squash-merged.
+
+**Format:**
+
+```
+[type] Short description
+```
+
+**Valid type tags:**
+
+`feat`, `feature`, `bugfix`, `fix`, `refactor`, `perf`, `ci`, `doc`, `docs`, `misc`, `chore`,
+`kernel`, `new-model`
+
+**Valid examples:**
+
+```
+[feat] Add causal Wan pipeline
+[bugfix] Fix VAE temporal tiling corruption
+[refactor] Restructure training framework
+[perf] Optimize FlashAttention kernel dispatch
+[docs] Add inference guide for LoRA
+[new-model] Port HunyuanVideo 1.5 to FastVideo
+```
+
+**Invalid examples (will block merge):**
+
+```
+Add causal Wan pipeline          ← missing type tag
+FEAT: Add pipeline               ← wrong format
+feat: Add pipeline               ← square brackets required
+```
+
+If your title is invalid, Mergify posts a comment explaining the required format and the
+merge protection check will remain failed until you update the title.
+
+---
+
+## Slash Commands
+
+Slash commands let contributors and maintainers trigger CI actions directly from PR comments.
+**Write permission to the repository is required.**
+
+The command is recognized within a few seconds. The workflow reacts with a 🚀 emoji to confirm.
+
+### `/merge`
+
+```
+/merge
+```
+
+Adds the `ready` label to the PR, which triggers the Full Suite on your PR branch and
+enables Mergify to auto-squash-merge once all conditions pass.
+
+The command first removes the `ready` label if it is already present, then re-adds it. This
+ensures the `labeled` event fires and a fresh Full Suite build is started even on a re-try.
+
+### `/test <name>`
+
+```
+/test <name>
+```
+
+Triggers a specific Buildkite test or suite on the current PR branch.
+
+| Command | Runs | Maps to `TEST_TYPE` |
+|---------|------|---------------------|
+| `/test encoder` | Encoder Tests (Fastcheck) | `encoder` |
+| `/test vae` | VAE Tests (Fastcheck) | `vae` |
+| `/test transformer` | Transformer Tests (Fastcheck) | `transformer` |
+| `/test kernel` | Kernel Tests (Fastcheck) | `kernel_tests` |
+| `/test unit` | Unit Tests (Fastcheck) | `unit_test` |
+| `/test ssim` | SSIM regression tests | `ssim` |
+| `/test training` | Training pipeline tests | `training` |
+| `/test lora-inference` | LoRA inference tests | `inference_lora` |
+| `/test lora-training` | LoRA training tests | `training_lora` |
+| `/test distillation` | DMD distillation tests | `distillation_dmd` |
+| `/test self-forcing` | Self-Forcing tests | `self_forcing` |
+| `/test vsa` | VSA training tests | `training_vsa` |
+| `/test vmoba` | VMoBA inference tests | `inference_vmoba` |
+| `/test performance` | Performance benchmarks | `performance` |
+| `/test api` | API server integration tests | `api_server` |
+| `/test full` | Entire Full Suite | all (with `TEST_SCOPE=full`) |
+| `/test fastcheck` | Entire Fastcheck suite | fastcheck (with `TEST_SCOPE=fastcheck`) |
+| `/test pre-commit` | Pre-commit checks on PR code | — (runs `ci-precommit.yml` via `workflow_call`) |
+
+**Re-running failed tests:** When you use `/test <name>` to re-run a specific failed test,
+the resulting Buildkite check uses the same name as the original (e.g., `/test encoder`
+creates `buildkite/ci/microscope-encoder-tests`). This overwrites the failed check status.
+Once all tests in a tier pass, the aggregate status (`fastcheck-passed` or
+`full-suite-passed`) is automatically updated to `success` by the `ci-aggregate-status.yml`
+workflow.
+
+**How aggregate status refresh works:**
+
+1. `/test <name>` triggers a Buildkite build with `TEST_SCOPE=direct`. The test step uses
+   the same label as its fastcheck/full-suite counterpart, so the resulting GitHub check
+   overwrites the original.
+2. When the build completes, Buildkite's `notify` posts a `direct-test-completed` commit
+   status. This is the only signal that triggers the aggregate workflow — intermediate step
+   status updates do not trigger it.
+3. `ci-aggregate-status.yml` fires, calls `getCombinedStatusForRef` to fetch the latest
+   status for every context on that commit (each context returns only its most recent
+   state), groups them by prefix (`microscope-*` → fastcheck, `test-tube-*`/`bar-chart-*`
+   → full suite), and posts `fastcheck-passed: success` or `full-suite-passed: success` if
+   all entries in the group are `success`.
+4. Tests that were never triggered (skipped by monorepo-diff) have no status entry and do
+   not block the aggregate.
+
+---
+
+## Auto Branch Cleanup
+
+After a PR is squash-merged to `main`, Mergify automatically deletes the head branch.
+Protected branches (`main`, `master`, `release/*`) are never deleted.
+
+---
+
+## Workflow File Reference
+
+| Filename | Trigger | What it does |
+|----------|---------|-------------|
+| `ci-precommit.yml` | Every push / PR against `main` | Runs pre-commit hooks (yapf, ruff, mypy, codespell, pymarkdown, actionlint, check-filenames) |
+| `ci-trigger-full-suite.yml` | `ready` label added to a PR | Calls Buildkite API to run Full Suite on the PR branch |
+| `ci-slash-commands.yml` | PR comment starting with `/merge` or `/test` | Handles slash commands; adds `ready` label or triggers Buildkite |
+| `ci-aggregate-status.yml` | Any Buildkite commit status update | Checks if all tests in a tier passed; updates `fastcheck-passed` or `full-suite-passed` |
+| `community-issue-labeler.yml` | Issue opened or edited | Auto-labels issues by keyword matching against title and body |
+| `community-welcome.yml` | First contribution | Posts a welcome comment for first-time contributors |
+| `community-stale.yml` | Scheduled | Marks and closes stale issues and PRs |
+| `infra-build-image.yml` | Manual (`workflow_dispatch`) | Builds Docker images for CI |
+| `infra-docs.yml` | Changes to `docs/` merged to `main` | Builds and deploys documentation to GitHub Pages |
+| `publish-fastvideo.yml` | Version bump | Publishes `fastvideo` package to PyPI |
+| `publish-kernel.yml` | Version bump | Publishes `fastvideo-kernel` package to PyPI |
+| `publish-comfyui.yml` | Version bump | Publishes ComfyUI node package to PyPI |

docs/contributing/coding_agents.md

@@ -44,7 +44,7 @@ FastVideo maps a Diffusers-style repo into a pipeline like:
 - `fastvideo/configs/models/*`: arch configs and `param_names_mapping` for
   weight name translation.
 - `fastvideo/configs/pipelines/*`: pipeline wiring (component classes + names).
-- `fastvideo/configs/sample/*`: default runtime sampling parameters.
+- `fastvideo/api/sampling_param.py`: runtime sampling parameters.
 - `fastvideo/pipelines/basic/*`: end-to-end pipeline logic built from stages.
 - `model_index.json`: the HF repo entrypoint that maps component names to
   classes and weight files.
@@ -55,7 +55,7 @@ Minimal usage example (based on `examples/inference/basic/basic.py`):
 
 ```python
 from fastvideo import VideoGenerator
-from fastvideo.configs.sample import SamplingParam
+from fastvideo.api.sampling_param import SamplingParam
 
 model_id = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"  # or official_weights/<model_name>/
 generator = VideoGenerator.from_pretrained(model_id, num_gpus=1)
@@ -319,7 +319,8 @@ Purpose:
 
 - `fastvideo/configs/pipelines/` describes pipeline wiring and model module
   names.
-- `fastvideo/configs/sample/` defines default runtime parameters.
+- `fastvideo/api/sampling_param.py` defines runtime sampling parameters.
+  Defaults come from profiles in `fastvideo/pipelines/basic/<family>/profiles.py`.
 
 Action:
 
@@ -474,7 +475,7 @@ FastVideo integration.
 3. Pipeline wiring.
    - Pipeline: `fastvideo/pipelines/basic/wan/wan_pipeline.py`
    - Pipeline config: `fastvideo/configs/pipelines/wan.py`
-   - Sampling defaults: `fastvideo/configs/sample/wan.py`
+   - Sampling defaults: `fastvideo/pipelines/basic/wan/profiles.py`
 
 4. Minimal example.
    - Script: `examples/inference/basic/basic.py`

docs/contributing/developer_env/runpod.md

@@ -1,65 +1,195 @@
+# RunPod Development Environment
 
-# 📦 Developing FastVideo on RunPod
+RunPod gives you on-demand cloud GPUs for FastVideo development. It's useful when you need a beefy GPU to test training runs, benchmark inference, or reproduce results without waiting for shared cluster time.
 
-You can easily use the FastVideo Pod Template on [RunPod](https://www.runpod.io) for development or experimentation.
+## Prerequisites
 
-## Creating a new pod
+- A [RunPod](https://www.runpod.io) account with billing configured
+- An SSH key pair. If you don't have one, generate it with `ssh-keygen -t ed25519`
+- Your public key (`~/.ssh/id_ed25519.pub`) ready to paste into RunPod
+
+## Step 1: Create a Pod
+
+**1. Verify your account**
+
+Make sure you're logged into the right RunPod account before spending credits.
 
-- Make sure you are using the correct RunPod account.
 ![RunPod Account Selection](../../assets/images/runpod_account.png)
 
-- Use "Additional Filters" to select CUDA 12.8.
+**2. Filter by CUDA version**
+
+Use "Additional Filters" to select CUDA 12.8.
+
 ![RunPod CUDA selection](../../assets/images/runpod_cuda.png)
 
-- Click "Deploy" and Pick a single A40 or RTX 4090 GPU.
+**3. Select a GPU**
+
+Click "Deploy" and pick a GPU. See [GPU Recommendations](#gpu-recommendations) below for guidance on which GPU to choose.
+
 ![RunPod GPU Selection](../../assets/images/runpod_deploy.png)
 
-- Select the "FastVideo" or "fastvideo-dev" Pod Template.
+**4. Pick the FastVideo template**
+
+Select the "FastVideo" or "fastvideo-dev" Pod Template. This pulls the pre-built image that includes all dependencies, Flash Attention, and a ready-to-use `uv` environment.
+
 ![RunPod Pod Template Selection](../../assets/images/runpod_create.png)
 
-- Set the Pod name to "`<name>-<FastVideo>-<date>`".
+**5. Name your pod**
+
+Use a memorable name like `yourname-fastvideo-2026-03-28`. This helps if you have multiple pods running.
+
+**6. Add a persistent volume (recommended)**
+
+Attach a network volume to `/root/.cache` or `/models` for storing downloaded model weights. Models can be 10-50 GB each, and re-downloading them every session wastes time and bandwidth.
+
+**7. Deploy**
+
+Click Deploy. The pod takes a few minutes to start while the image pulls. You'll see it transition to "Running" in your dashboard.
+
+## Step 2: Connect via SSH
+
+Once the pod is running, find the "SSH over exposed TCP" connection string in the pod dashboard.
 
-- Finally, once the pod is deployed (will take a few minutes as the image is being pulled), you can SSH into it using "SSH exposed over TCP". You'll need to use the matching private ssh key you provided.
 ![RunPod SSH](../../assets/images/runpod_ssh.png)
 
-## Working with the pod
-
-After SSH'ing into your pod, you'll find the correct `uv` environment already activated and you should be in /FastVideo/ directory. Make sure to use /FastVideo/ for all your work.
-
-To pull in the latest changes from the GitHub repo:
+Connect with:
 
 ```bash
-cd /FastVideo
-git pull
+ssh root@<pod-ip> -p <port> -i ~/.ssh/id_ed25519
 ```
 
-Run your development workflows as usual:
+RunPod also supports VS Code Remote SSH if you prefer an IDE.
 
-```bash
-# Run linters
-pre-commit run --all-files
+### Custom template (advanced)
 
-# Run tests
-pytest tests/
-```
-
-Make sure to push your changes back to the GitHub repo as nothing will be saved to the pod when it is terminated.
-
-After you are done with your work, you can terminate the pod by clicking the "Terminate" and "Delete" buttons. Remember if the pod is not completely deleted, Runpod will keep charging you for it.
-
-## Extra Information:
-If you need to customize the pod template this section has some useful information. For the most part you can leave the defaults of the FastVideo Pod Template.
-
-When creating your pod template, use this image:
+If you're setting up a pod from scratch instead of the FastVideo template, use this image:
 
 ```
 ghcr.io/hao-ai-lab/fastvideo/fastvideo-dev:py3.12-latest
 ```
 
-Paste Container Start Command to support SSH ([RunPod Docs](https://docs.runpod.io/pods/configuration/use-ssh)):
+And paste this as the Container Start Command to enable SSH ([RunPod docs](https://docs.runpod.io/pods/configuration/use-ssh)):
 
 ```bash
 bash -c "apt update;DEBIAN_FRONTEND=noninteractive apt-get install openssh-server -y;mkdir -p ~/.ssh;cd $_;chmod 700 ~/.ssh;echo \"$PUBLIC_KEY\" >> authorized_keys;chmod 700 authorized_keys;service ssh start;sleep infinity"
 ```
 
 ![RunPod template configuration](../../assets/images/runpod_template.png)
+
+## Step 3: Set Up FastVideo
+
+After SSH'ing in, the `uv` virtual environment at `/opt/venv` is already activated (configured in `.bashrc` and `.profile`). You land in the `/FastVideo` directory.
+
+**Clone or pull the repo**
+
+If the pod already has the FastVideo repo:
+
+```bash
+cd /FastVideo
+git pull
+```
+
+If starting from a blank pod:
+
+```bash
+git clone https://github.com/hao-ai-lab/FastVideo.git /FastVideo
+cd /FastVideo
+```
+
+**Install the package**
+
+```bash
+uv pip install -e .[dev]
+```
+
+The Docker image already includes Flash Attention and most heavy dependencies, so this is fast.
+
+**Build the custom kernels (optional)**
+
+VSA and STA attention kernels aren't in the Docker image by default. Build them if you're working on attention backends or need maximum inference performance:
+
+```bash
+cd /FastVideo/fastvideo-kernel
+./build.sh
+```
+
+The build script detects your GPU architecture automatically. An A100 or H100 takes about 5-10 minutes.
+
+**Verify the setup**
+
+```bash
+cd /FastVideo
+python -c "import fastvideo; print('OK')"
+pytest tests/ -q --no-header
+```
+
+## Development Workflow
+
+### Editing code on RunPod
+
+Two common approaches:
+
+**Option A: Edit on RunPod directly**
+
+Use VS Code Remote SSH or `vim`/`nano` on the pod. Commit and push when you're ready:
+
+```bash
+cd /FastVideo
+git add .
+git commit -m "your change"
+git push
+```
+
+**Option B: Edit locally, sync to RunPod**
+
+Work in your local repo, then pull on the pod:
+
+```bash
+# On RunPod:
+cd /FastVideo
+git pull
+```
+
+This keeps your local tools (editor, linters) intact while running GPU workloads on the pod.
+
+### Running linters and tests
+
+```bash
+# Lint
+pre-commit run --all-files
+
+# Full test suite
+pytest tests/
+
+# Just package tests
+pytest fastvideo/tests/ -v
+```
+
+### Storing models
+
+If you attached a persistent volume, point your model downloads there:
+
+```bash
+export HF_HOME=/models/huggingface
+export TRANSFORMERS_CACHE=/models/huggingface
+```
+
+Add these to `/root/.bashrc` so they persist across SSH sessions. The volume survives pod termination, so you only download models once.
+
+### Terminating the pod
+
+When you're done, push any commits you want to keep. RunPod does not save pod storage after termination.
+
+Go to your RunPod dashboard, click "Terminate", then "Delete". A pod that's stopped but not deleted still charges you for storage. Fully delete it to stop all charges.
+
+## GPU Recommendations
+
+| GPU | VRAM | Good for |
+|-----|------|----------|
+| RTX 4090 | 24 GB | Inference testing, small model fine-tuning, quick iteration |
+| A40 | 48 GB | Mid-size training runs, 480p video generation |
+| A100 (40 GB) | 40 GB | Multi-GPU inference, training with sequence parallelism |
+| A100 (80 GB) | 80 GB | Large model training, 720p+ video generation |
+| H100 | 80 GB | Heavy training, benchmarking, kernel development |
+
+For most development work, a single RTX 4090 or A40 is sufficient and cost-effective. Use A100/H100 when you need to reproduce training results at scale or test multi-GPU features.

docs/contributing/overview.md

@@ -14,24 +14,11 @@ improving performance, or fixing a bug.
 
 For a full install checklist, see `docs/getting_started/installation/gpu.md`.
 
-## Local development (Conda + editable install)
+## Local development (UV + editable install)
 
-Install Miniconda:
+If you previously used Conda for local setup, switch to uv for a faster and more stable development environment.
 
-```bash
-wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
-bash Miniconda3-latest-Linux-x86_64.sh
-source ~/.bashrc
-```
-
-Create and activate a Conda environment:
-
-```bash
-conda create -n fastvideo python=3.12 -y
-conda activate fastvideo
-```
-
-Install `uv` (optional, but recommended):
+Install `uv`:
 
 ```bash
 curl -LsSf https://astral.sh/uv/install.sh | sh
@@ -39,6 +26,20 @@ curl -LsSf https://astral.sh/uv/install.sh | sh
 wget -qO- https://astral.sh/uv/install.sh | sh
 ```
 
+Create and activate a uv environment (recommended):
+
+```bash
+uv venv --python 3.12 --seed
+source .venv/bin/activate
+```
+
+Conda alternative (supported):
+
+```bash
+conda create -n fastvideo python=3.12 -y
+conda activate fastvideo
+```
+
 Clone the repo:
 
 ```bash
@@ -51,7 +52,7 @@ Install FastVideo in editable mode and set up hooks:
 uv pip install -e .[dev]
 
 # Optional: FlashAttention (builds native kernels)
-uv pip install flash-attn --no-build-isolation
+uv pip install flash-attn --no-build-isolation -v
 
 # Linting, formatting, static typing
 pre-commit install --hook-type pre-commit --hook-type commit-msg

docs/contributing/pull_requests.md

@@ -0,0 +1,223 @@
+# Contributing via Pull Requests
+
+This guide walks through the PR workflow: title format, labels, CI pipeline, and getting
+your changes merged.
+
+---
+
+## PR Title Format (Required)
+
+Every PR targeting `main` must start with a type tag in square brackets. This is checked by
+Mergify before any merge is allowed.
+
+**Format:**
+
+```
+[type] Short description of the change
+```
+
+**Valid type tags:**
+
+| Tag | When to use |
+|-----|-------------|
+| `[feat]` or `[feature]` | New feature or capability |
+| `[bugfix]` or `[fix]` | Bug fix |
+| `[refactor]` | Code restructuring with no behavior change |
+| `[perf]` | Performance improvement |
+| `[ci]` | CI/CD or build tooling changes |
+| `[doc]` or `[docs]` | Documentation only |
+| `[misc]` or `[chore]` | Housekeeping, dependency bumps, minor cleanup |
+| `[kernel]` | CUDA kernel changes in `fastvideo-kernel/` |
+| `[new-model]` | Adding a new model or pipeline |
+
+**Examples:**
+
+```
+[feat] Add causal Wan 2.2 I2V pipeline
+[bugfix] Fix VAE temporal tiling corruption on H100
+[refactor] Restructure distributed attention dispatch
+[docs] Add LoRA finetuning guide
+[new-model] Port HunyuanVideo 1.5 to FastVideo
+```
+
+If your title is missing the tag, Mergify will post a comment listing the valid formats.
+Update the title and the check will re-evaluate automatically.
+
+---
+
+## Labels
+
+Labels are applied automatically based on your PR title and the files you changed. You don't
+need to set them manually.
+
+**Type label** — set from the `[tag]` in your title:
+`type: feat`, `type: bugfix`, `type: refactor`, `type: perf`, `type: ci`, `type: docs`,
+`type: misc`, `type: new-model`
+
+**Scope labels** — set from which files you modified (multiple labels can apply):
+`scope: training`, `scope: inference`, `scope: attention`, `scope: kernel`, `scope: data`,
+`scope: infra`, `scope: distributed`, `scope: docs`, `scope: ui`, `scope: model`
+
+**Process labels** — set during review and merge:
+
+| Label | Who sets it | Meaning |
+|-------|-------------|---------|
+| `ready` | You (`/merge` comment) or a maintainer | Triggers Full Suite and enables auto-merge |
+| `needs-rebase` | Mergify (automatic) | Your PR has conflicts; rebase against `main` |
+| `do-not-merge` | Maintainer | Blocks merge regardless of CI status |
+
+---
+
+## CI Pipeline
+
+Three tiers run automatically on every PR.
+
+**Tier 1: Pre-commit (~2 min) — runs on every push**
+
+GitHub Actions checks formatting, linting, type correctness, and spelling using pre-commit
+hooks: yapf, ruff, mypy, codespell, pymarkdown, actionlint, and check-filenames.
+
+**Tier 2: Fastcheck (~10-20 min) — runs on every push, path-filtered**
+
+Buildkite runs GPU tests only for the components you changed. If you only modified
+`fastvideo/models/vaes/`, only VAE Tests run. Tests run in parallel.
+
+**Tier 3: Full Suite (~60-90 min) — triggered by the `ready` label**
+
+When you comment `/merge` (or a maintainer adds the `ready` label), Buildkite runs the
+complete test suite on your PR branch: SSIM regression, LoRA inference and training,
+distillation, self-forcing, VSA, VMoBA, performance benchmarks, and API server tests.
+
+---
+
+## Getting Your PR Merged
+
+**Step-by-step:**
+
+1. Open a PR with a title that starts with a valid `[type]` tag.
+2. Push your changes. Pre-commit and Fastcheck run automatically.
+3. Fix any pre-commit failures locally (`pre-commit run --all-files`) and push again.
+4. Wait for at least one approving review.
+5. Once approved and pre-commit is green, comment `/merge` on the PR.
+6. The `ready` label is added, which triggers the Full Suite on your PR branch.
+7. Mergify also auto-rebases your branch against `main` if it is behind and conflict-free.
+8. If all Full Suite tests pass and all merge conditions are met (approval, valid title,
+   pre-commit green, fastcheck green, no draft, no conflicts), Mergify squash-merges to
+   `main` automatically. Your branch is deleted.
+9. If a Full Suite test fails, check the Buildkite build log for the failing step. Fix the
+   issue, push, and comment `/merge` again. You can also re-run individual failed tests
+   with `/test <name>` — see below.
+
+!!! note
+    Only contributors with write permission to the repository can trigger slash commands.
+    If you're an external contributor, ask a maintainer to run `/merge` or add the `ready`
+    label for you.
+
+---
+
+## Running Tests On Demand
+
+Comment on your PR to trigger specific tests independently of the auto-merge flow.
+
+**Trigger the entire Full Suite:**
+
+```
+/test full
+```
+
+**Trigger the Fastcheck suite:**
+
+```
+/test fastcheck
+```
+
+**Trigger individual tests:**
+
+```
+/test encoder          # Encoder component tests
+/test vae              # VAE component tests
+/test transformer      # Transformer / DiT tests
+/test kernel           # CUDA kernel tests
+/test unit             # Unit tests
+
+/test ssim             # SSIM regression tests
+/test training         # Training pipeline tests
+/test lora-inference   # LoRA inference tests
+/test lora-training    # LoRA training tests
+/test distillation     # DMD distillation tests
+/test self-forcing     # Self-Forcing distillation tests
+/test vsa              # VSA training tests
+/test vmoba            # VMoBA inference tests
+/test performance      # Performance benchmarks
+/test api              # API server integration tests
+/test pre-commit       # Pre-commit checks on PR code
+```
+
+The workflow reacts with a 🚀 emoji to confirm the command was received.
+
+When you re-run an individual test with `/test <name>`, the new result overwrites the
+original failed check (same Buildkite check name). Once all tests in a tier pass, the
+`fastcheck-passed` or `full-suite-passed` status is automatically updated.
+
+---
+
+## Troubleshooting
+
+### Pre-commit fails
+
+Run locally to reproduce and auto-fix:
+
+```bash
+# Install pre-commit if needed
+uv pip install pre-commit
+pre-commit install
+
+# Run all checks on all files
+pre-commit run --all-files
+```
+
+Common quick fixes:
+
+- **yapf**: `yapf -i <file>` (Python formatting)
+- **ruff**: `ruff check --fix <file>` (linting)
+- **codespell**: `codespell --write-changes <file>` (spelling)
+
+### PR title format check fails
+
+Update your title to start with a valid type tag. The Mergify merge protection check
+re-evaluates automatically after you save the title.
+
+Valid tags: `feat`, `feature`, `bugfix`, `fix`, `refactor`, `perf`, `ci`, `doc`, `docs`,
+`misc`, `chore`, `kernel`, `new-model`
+
+### My PR has merge conflicts (`needs-rebase` label)
+
+Rebase against `main` and force-push:
+
+```bash
+git fetch origin main
+git rebase origin/main
+# Resolve any conflicts, then:
+git push --force-with-lease
+```
+
+Mergify removes the `needs-rebase` label automatically once conflicts are resolved.
+
+### Full Suite failed after `/merge`
+
+The Full Suite found a regression. Check the failing Buildkite step's output for assertion
+errors or tracebacks.
+
+Common causes:
+
+- Test failures caused by your code changes
+- Missing dependency in `pyproject.toml`
+- GPU memory issue (some tests require specific hardware like L40S or H100)
+- Kernel build failure (if you changed `fastvideo-kernel/`)
+
+After fixing, push and comment `/merge` again.
+
+### I'm an external contributor without write permission
+
+You can't use slash commands directly. After your PR is approved, ask a maintainer to
+comment `/merge` or add the `ready` label.

docs/contributing/testing.md

@@ -8,7 +8,7 @@ This guide explains how to add and run tests in FastVideo. The testing suite is
 * **Component Tests**: Located in `fastvideo/tests/encoders`, `fastvideo/tests/transformers`, and `fastvideo/tests/vaes`. These verify the loading and basic functionality of model components.
 * **SSIM Tests**: Located in `fastvideo/tests/ssim`. These are regression tests that compare generated videos against reference videos using the Structural Similarity Index Measure (SSIM) to detect quality degradation.
 * **Training Tests**: Located in `fastvideo/tests/training`. These validate training loops, loss calculations, and specific training techniques like LoRA, Distillation, and VSA.
-* **Inference Tests**: Located in `fastvideo/tests/inference`. These test specialized inference pipelines and optimizations (e.g., STA, V-MoBA).
+* **Inference Tests**: Located in `fastvideo/tests/inference`. These test specialized inference pipelines and optimizations (e.g., VSA, V-MoBA).
 
 For now, we will focus on **SSIM Tests**.
 
@@ -23,13 +23,18 @@ SSIM tests are located in `fastvideo/tests/ssim`. These tests generate videos us
 
 ```
 fastvideo/tests/ssim/
-├── <GPU>_reference_videos/   # Reference videos organized by GPU type (e.g., L40S_reference_videos)
-│   ├── <Model_Name>/
-│   │   ├── <Backend>/        # e.g., FLASH_ATTN, TORCH_SDPA
-│   │   │   └── <Video_File>
+├── reference_videos/
+│   ├── default/
+│   │   └── <GPU>_reference_videos/
+│   │       ├── <Model_Name>/
+│   │       │   ├── <Backend>/        # e.g., FLASH_ATTN, TORCH_SDPA
+│   │       │   │   └── <Video_File>
+│   └── full_quality/
+│       └── <GPU>_reference_videos/
 ├── test_causal_similarity.py
-├── test_inference_similarity.py
-├── update_reference_videos.sh
+├── test_wan_t2v_similarity.py
+├── test_wan_i2v_similarity.py
+├── reference_videos_cli.py
 └── ...
 ```
 
@@ -37,7 +42,7 @@ fastvideo/tests/ssim/
 
 To add a new SSIM test, follow these steps:
 
-1. **Create or Update a Test File**: You can add a new test function to an existing file (like `test_inference_similarity.py`) or create a new one if testing a distinct category of models.
+1. **Create or Update a Test File**: Prefer model-specific files (for example `test_wan_t2v_similarity.py`) and create a new one when testing a distinct model or pipeline.
 
 2. **Define Model Parameters**: Define the configuration for the model you want to test. This includes model path, dimensions, inference steps, and other generation parameters. **Note:** Consider using lower `num_inference_steps` or reduced resolution (e.g., 480p instead of 720p) to keep test execution time reasonable, provided it doesn't compromise the test's ability to detect regression.
 
@@ -81,10 +86,15 @@ To add a new SSIM test, follow these steps:
    ```
 
 4. **Reference Videos**:
-   * When running the test for the first time (or when updating the reference), the test will fail because the reference video is missing. The generated video will be saved in `fastvideo/tests/ssim/generated_videos`.
+   * When running the test for the first time (or when updating the reference), the test will fail because the reference video is missing. The generated video will be saved in `fastvideo/tests/ssim/generated_videos/<quality-tier>/<GPU>_reference_videos`.
    * Inspect the generated video to ensure it meets quality expectations.
-   * Move the generated video to the appropriate reference folder: `fastvideo/tests/ssim/<GPU>_reference_videos/<Model>/<Backend>/`.
-   * You can use the helper script `update_reference_videos.sh` to automate copying videos from `generated_videos` to `L40S_reference_videos`. Note: Check the script to ensure paths match your environment (it defaults to `L40S_reference_videos`).
+   * Move the generated video to the appropriate quality/GPU reference folder:
+     `fastvideo/tests/ssim/reference_videos/<quality-tier>/<GPU>_reference_videos/<Model>/<Backend>/`.
+   * You can use the helper CLI to copy generated videos into a reference folder:
+     `python fastvideo/tests/ssim/reference_videos_cli.py copy-local --quality-tier default --reference-dir fastvideo/tests/ssim/reference_videos/default/L40S_reference_videos`
+   * Upload/download can target both quality tiers and specific GPU folders:
+     `python fastvideo/tests/ssim/reference_videos_cli.py upload --quality-tier all`
+     `python fastvideo/tests/ssim/reference_videos_cli.py download --quality-tier full_quality --device-folder H200_reference_videos`
 
 ### Running Tests Locally
 
@@ -96,36 +106,113 @@ pytest fastvideo/tests/ssim/ -vs
 
 Ensure you have the necessary GPUs available as defined in your test parameters.
 
-## Modal Workflow
+## CI Integration
 
-FastVideo uses [Modal](https://modal.com/) for running tests in a CI environment. The workflow scripts are located in `fastvideo/tests/modal/`.
+FastVideo uses [Modal](https://modal.com/) for running tests in a CI environment. The
+workflow scripts are located in `fastvideo/tests/modal/`.
+
+### Buildkite Pipeline
+
+Tests are orchestrated by Buildkite (`.buildkite/pipeline.yml`) and executed on Modal GPU
+instances. The pipeline runs in two modes:
+
+**Fastcheck** — runs on every PR push, path-filtered. Only tests for the components you
+changed are triggered. Tests run in parallel.
+
+**Full Suite** — runs when a PR enters the Merge Queue (or when triggered manually via
+`/test full`). Covers SSIM regression, training, distillation, inference, and performance.
 
 ### `pr_test.py`
 
-The main entry point for CI tests is `fastvideo/tests/modal/pr_test.py`. This script defines Modal functions that execute the pytest suites on specific hardware (e.g., L40S, H100).
+The main entry point for CI tests is `fastvideo/tests/modal/pr_test.py`. This script defines
+Modal functions that execute the pytest suites on specific hardware (e.g., L40S, H100).
 
 ### Updating Modal Configuration
 
 If you add a new test that requires:
+
 * **Different GPU Hardware**: You may need to change the `@app.function(gpu=...)` decorator.
 * **Longer Execution Time**: Increase the `timeout` parameter.
-* **New Environment Variables/Secrets**: Add them to `secrets=[...]` or the image environment. For example, if your model is gated on Hugging Face, ensure `HF_API_KEY` is passed.
+* **New Environment Variables/Secrets**: Add them to `secrets=[...]` or the image
+  environment. For example, if your model is gated on Hugging Face, ensure `HF_API_KEY`
+  is passed.
 
-For SSIM tests, the `run_ssim_tests` function in `pr_test.py` currently runs:
+For SSIM tests, use `fastvideo/tests/modal/ssim_test.py`:
 
-```python
-@app.function(gpu="L40S:2", image=image, timeout=2700, secrets=[modal.Secret.from_dict({"HF_API_KEY": os.environ.get("HF_API_KEY", "")})])
-def run_ssim_tests():
-    run_test("hf auth login --token $HF_API_KEY && pytest ./fastvideo/tests/ssim -vs")
+```bash
+python -m modal run fastvideo/tests/modal/ssim_test.py::run_ssim_tests
 ```
 
-If your new test file is inside `fastvideo/tests/ssim`, it will automatically be picked up by this command. However, ensure that the `gpu="L40S:2"` configuration is sufficient for your model. If your model requires more GPUs (e.g., 4 or 8), you might need to create a separate Modal function or update the existing one.
+Target specific SSIM files/models:
+
+```bash
+python -m modal run fastvideo/tests/modal/ssim_test.py::run_ssim_tests \
+  --test-files test_wan_t2v_similarity.py \
+  --model-ids Wan2.1-T2V-1.3B-Diffusers
+```
+
+If HF token env vars are not set (`HF_API_KEY` / `HUGGINGFACE_HUB_TOKEN` /
+`HF_TOKEN`), the local entrypoint fails fast. To export raw `generated_videos`
+from Modal to the shared volume:
+
+```bash
+python -m modal run fastvideo/tests/modal/ssim_test.py::run_ssim_tests \
+  --sync-generated-to-volume
+```
+
+The raw export path is quality-tiered:
+
+* default params: `ssim_generated_videos/default/<subdir>/generated_videos`
+* full-quality params: `ssim_generated_videos/full_quality/<subdir>/generated_videos`
+
+The printed `modal volume get` command also downloads into a quality-specific
+local directory under `./generated_videos_modal/<quality-tier>`.
+
+To turn downloaded Modal outputs into local reference videos, use the matching
+quality tier with `copy-local`, for example:
+
+```bash
+python fastvideo/tests/ssim/reference_videos_cli.py copy-local \
+  --quality-tier full_quality \
+  --generated-dir ./generated_videos_modal/full_quality/L40S_reference_videos \
+  --device-folder L40S_reference_videos
+```
 
 ### Workflow Scripts
 
-The shell script that triggers these tests in the CI pipeline is located at `.buildkite/scripts/pr_test.sh`. If you add a new test category (e.g., a new folder outside of `ssim`), you will need to:
+The shell script that triggers tests in CI is `.buildkite/scripts/pr_test.sh`. If you add
+a new test category (e.g., a new folder outside of `ssim`), you will need to:
+
 1. Add a new function in `fastvideo/tests/modal/pr_test.py`.
 2. Add a new case in `.buildkite/scripts/pr_test.sh` to handle the new test type.
 
 !!! note
-    If you are a maintainer, you'll need to finally manually update the workflow script in Buildkite. Otherwise, a maintainer will help you update.
+    If you are a maintainer, update the workflow script in Buildkite after merging. Otherwise,
+    ask a maintainer for help.
+
+## Triggering Tests via Slash Commands
+
+Maintainers and contributors with write access can trigger individual test suites directly
+from a PR comment. The workflow reacts with a 🚀 emoji to confirm the command was received.
+
+```
+/test ssim              # SSIM regression tests
+/test training          # Training pipeline tests
+/test lora-training     # LoRA training tests
+/test lora-inference    # LoRA inference tests
+/test distillation      # DMD distillation tests
+/test self-forcing      # Self-Forcing tests
+/test vsa               # VSA training tests
+/test vmoba             # VMoBA inference tests
+/test performance       # Performance benchmarks
+/test api               # API server integration tests
+/test encoder           # Encoder component tests (Fastcheck)
+/test vae               # VAE component tests (Fastcheck)
+/test transformer       # Transformer / DiT tests (Fastcheck)
+/test kernel            # CUDA kernel tests (Fastcheck)
+/test unit              # Unit tests (Fastcheck)
+/test full              # Entire Full Suite
+/test fastcheck         # Entire Fastcheck suite
+```
+
+See [CI Architecture](ci_architecture.md) for the complete reference.

docs/design/inference_schema_parity_inventory.yaml

@@ -0,0 +1,449 @@
+status_definitions:
+  kept: "Public field remains on a public adapter surface with the same meaning."
+  moved: "Public field remains supported but normalizes into a different nested path."
+  preset_owned: "Public field remains supported only through a model/preset-specific surface."
+  compatibility_only: "Legacy public field remains adapter-only during migration and is not part of the canonical typed schema."
+  private_only: "Field should only be handled by private adapters and is not a public FastVideo compatibility promise."
+  internal_only: "Field is runtime/config plumbing and should not be part of the new public typed inference API."
+
+surfaces:
+  fastvideo_args:
+    moved:
+      model_path: generator.model_path
+      workload_type: generator.pipeline.workload_type
+      distributed_executor_backend: generator.engine.execution_backend
+      trust_remote_code: generator.trust_remote_code
+      revision: generator.revision
+      num_gpus: generator.engine.num_gpus
+      tp_size: generator.engine.parallelism.tp_size
+      sp_size: generator.engine.parallelism.sp_size
+      hsdp_replicate_dim: generator.engine.parallelism.hsdp_replicate_dim
+      hsdp_shard_dim: generator.engine.parallelism.hsdp_shard_dim
+      dist_timeout: generator.engine.parallelism.dist_timeout
+      lora_path: generator.pipeline.components.lora_path
+      dit_cpu_offload: generator.engine.offload.dit
+      use_fsdp_inference: generator.engine.use_fsdp_inference
+      dit_layerwise_offload: generator.engine.offload.dit_layerwise
+      text_encoder_cpu_offload: generator.engine.offload.text_encoder
+      image_encoder_cpu_offload: generator.engine.offload.image_encoder
+      vae_cpu_offload: generator.engine.offload.vae
+      pin_cpu_memory: generator.engine.offload.pin_cpu_memory
+      enable_torch_compile: generator.engine.compile.enabled
+      torch_compile_kwargs: generator.engine.compile.backend,fullgraph,mode,dynamic,extras
+      disable_autocast: generator.engine.disable_autocast
+      enable_stage_verification: generator.engine.enable_stage_verification
+      prompt_txt: request.inputs.prompt_path
+      override_text_encoder_safetensors: generator.pipeline.components.text_encoder_weights
+      override_text_encoder_quant: generator.engine.quantization.text_encoder_quant
+      override_transformer_cls_name: generator.pipeline.components.override_transformer_cls_name
+      init_weights_from_safetensors: generator.pipeline.components.transformer_weights
+      init_weights_from_safetensors_2: generator.pipeline.components.transformer_2_weights
+      override_pipeline_cls_name: generator.pipeline.components.override_pipeline_cls_name
+      boundary_ratio: request.sampling.boundary_ratio
+      ltx2_vae_tiling: generator.pipeline.vae_tiling
+    preset_owned:
+      ltx2_vae_spatial_tile_size_in_pixels: generator.pipeline.preset_overrides.ltx2.vae.spatial_tile_size_in_pixels
+      ltx2_vae_spatial_tile_overlap_in_pixels: generator.pipeline.preset_overrides.ltx2.vae.spatial_tile_overlap_in_pixels
+      ltx2_vae_temporal_tile_size_in_frames: generator.pipeline.preset_overrides.ltx2.vae.temporal_tile_size_in_frames
+      ltx2_vae_temporal_tile_overlap_in_frames: generator.pipeline.preset_overrides.ltx2.vae.temporal_tile_overlap_in_frames
+      ltx2_initial_latent_path: request.extensions.ltx2.initial_latent_path
+    compatibility_only:
+      mode: "Legacy multi-mode FastVideoArgs switch; typed inference config should not expose execution mode."
+      inference_mode: "Legacy boolean mirror of mode; kept only through adapters while FastVideoArgs remains."
+      lora_nickname: "Legacy adapter-selection surface pending LoRA API cleanup."
+      lora_target_modules: "Legacy LoRA configuration surface pending dedicated component API."
+      output_type: "Legacy output formatting surface pending GenerationResult cleanup."
+      VSA_sparsity: "Model-specific inference optimization not yet represented in the typed public schema."
+      moba_config_path: "Model-specific MoBA optimization surface not yet represented in the typed public schema."
+      master_port: "Executor/bootstrap compatibility field; not part of the canonical inference schema."
+    private_only:
+      ray_placement_group: "Ray deployment-only field."
+      ray_runtime_env: "Ray deployment-only field."
+    internal_only:
+      pipeline_config: "Legacy internal carrier object."
+      preprocess_config: "Legacy preprocess carrier object."
+      moba_config: "Derived runtime config loaded from moba_config_path."
+      model_paths: "Runtime bookkeeping."
+      model_loaded: "Runtime bookkeeping."
+
+  pipeline_config_base:
+    moved:
+      pipeline_config_path: generator.pipeline.components.pipeline_config_path
+    preset_owned:
+      embedded_cfg_scale: generator.pipeline.preset_overrides.embedded_cfg_scale
+      flow_shift: generator.pipeline.preset_overrides.flow_shift
+      flow_shift_sr: generator.pipeline.preset_overrides.flow_shift_sr
+      is_causal: generator.pipeline.preset_overrides.is_causal
+      vae_tiling: generator.pipeline.preset_overrides.vae_tiling
+      vae_sp: generator.pipeline.preset_overrides.vae_sp
+      dmd_denoising_steps: generator.pipeline.preset_overrides.dmd_denoising_steps
+      ti2v_task: generator.pipeline.preset_overrides.ti2v_task
+      boundary_ratio: generator.pipeline.preset_overrides.boundary_ratio
+    compatibility_only:
+      model_path: "Redundant with generator.model_path."
+      disable_autocast: "Duplicated by generator.engine.disable_autocast during migration."
+      dit_precision: "Precision override pending dedicated typed component precision design."
+      upsampler_precision: "Precision override pending dedicated typed component precision design."
+      vae_precision: "Precision override pending dedicated typed component precision design."
+      image_encoder_precision: "Precision override pending dedicated typed component precision design."
+      text_encoder_precisions: "Precision override pending dedicated typed component precision design."
+    internal_only:
+      dit_config: "Legacy internal component config object."
+      upsampler_config: "Legacy internal component config object."
+      vae_config: "Legacy internal component config object."
+      image_encoder_config: "Legacy internal component config object."
+      text_encoder_configs: "Legacy internal component config object."
+      preprocess_text_funcs: "Internal text preprocessing hooks."
+      postprocess_text_funcs: "Internal text postprocessing hooks."
+
+  pipeline_config_extensions:
+    preset_owned:
+      conditioning_strategy:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      max_num_conditional_frames:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      min_num_conditional_frames:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      sigma_conditional:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      sigma_data:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      state_ch:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      state_t:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      text_encoder_class:
+        sources:
+          - fastvideo.configs.pipelines.cosmos.CosmosConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      autoregressive_chunk_frames:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      autoregressive_overlap_frames:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      cfg_behavior:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      default_camera_rotation:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      default_movement_distance:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      default_negative_prompt:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      default_trajectory_type:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      filter_points_threshold:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      fps:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      frame_buffer_max:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      moge_model_name:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      noise_aug_strength:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      num_frames:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      offload_moge_after_depth:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      use_moge_depth:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      video_resolution:
+        sources:
+          - fastvideo.configs.pipelines.gen3c.Gen3CConfig
+          - fastvideo.configs.pipelines.gen3c.Gen3CInferenceConfig
+      text_encoder_crop_start:
+        sources:
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15I2V480PStepDistilledConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15I2V720PConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15SR1080PConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15T2V480PConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15T2V720PConfig
+          - fastvideo.configs.pipelines.hyworld.HYWorldConfig
+          - fastvideo.configs.pipelines.hyworld.Hunyuan15T2V480PConfig
+      text_encoder_max_lengths:
+        sources:
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15I2V480PStepDistilledConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15I2V720PConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15SR1080PConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15T2V480PConfig
+          - fastvideo.configs.pipelines.hunyuan15.Hunyuan15T2V720PConfig
+          - fastvideo.configs.pipelines.hyworld.HYWorldConfig
+          - fastvideo.configs.pipelines.hyworld.Hunyuan15T2V480PConfig
+      precision:
+        sources:
+          - fastvideo.configs.pipelines.lingbotworld.LingBotWorldI2V480PConfig
+          - fastvideo.configs.pipelines.lingbotworld.Wan2_2_I2V_A14B_Config
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionI2VConfig
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionI2V_A14B_Config
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionT2VConfig
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionT2V_14B_Config
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionT2V_1_3B_Config
+          - fastvideo.configs.pipelines.wan.FastWan2_1_T2V_480P_Config
+          - fastvideo.configs.pipelines.wan.FastWan2_2_TI2V_5B_Config
+          - fastvideo.configs.pipelines.wan.MatrixGameBaseI2V480PConfig
+          - fastvideo.configs.pipelines.wan.MatrixGameI2V480PConfig
+          - fastvideo.configs.pipelines.wan.SelfForcingWan2_2_T2V480PConfig
+          - fastvideo.configs.pipelines.wan.SelfForcingWanT2V480PConfig
+          - fastvideo.configs.pipelines.wan.WANV2VConfig
+          - fastvideo.configs.pipelines.wan.Wan2_2_I2V_A14B_Config
+          - fastvideo.configs.pipelines.wan.Wan2_2_T2V_A14B_Config
+          - fastvideo.configs.pipelines.wan.Wan2_2_TI2V_5B_Config
+          - fastvideo.configs.pipelines.wan.WanI2V480PConfig
+          - fastvideo.configs.pipelines.wan.WanI2V720PConfig
+          - fastvideo.configs.pipelines.wan.WanT2V480PConfig
+          - fastvideo.configs.pipelines.wan.WanT2V720PConfig
+      warp_denoising_step:
+        sources:
+          - fastvideo.configs.pipelines.lingbotworld.LingBotWorldI2V480PConfig
+          - fastvideo.configs.pipelines.lingbotworld.Wan2_2_I2V_A14B_Config
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionI2VConfig
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionI2V_A14B_Config
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionT2VConfig
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionT2V_14B_Config
+          - fastvideo.configs.pipelines.turbodiffusion.TurboDiffusionT2V_1_3B_Config
+          - fastvideo.configs.pipelines.wan.FastWan2_1_T2V_480P_Config
+          - fastvideo.configs.pipelines.wan.FastWan2_2_TI2V_5B_Config
+          - fastvideo.configs.pipelines.wan.MatrixGameBaseI2V480PConfig
+          - fastvideo.configs.pipelines.wan.MatrixGameI2V480PConfig
+          - fastvideo.configs.pipelines.wan.SelfForcingWan2_2_T2V480PConfig
+          - fastvideo.configs.pipelines.wan.SelfForcingWanT2V480PConfig
+          - fastvideo.configs.pipelines.wan.WANV2VConfig
+          - fastvideo.configs.pipelines.wan.Wan2_2_I2V_A14B_Config
+          - fastvideo.configs.pipelines.wan.Wan2_2_T2V_A14B_Config
+          - fastvideo.configs.pipelines.wan.Wan2_2_TI2V_5B_Config
+          - fastvideo.configs.pipelines.wan.WanI2V480PConfig
+          - fastvideo.configs.pipelines.wan.WanI2V720PConfig
+          - fastvideo.configs.pipelines.wan.WanT2V480PConfig
+          - fastvideo.configs.pipelines.wan.WanT2V720PConfig
+      bsa_cdf_threshold:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      bsa_chunk_k:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      bsa_chunk_q:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      bsa_params:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      bsa_sparsity:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      enable_bsa:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      enable_kv_cache:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      enhance_hf:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      offload_kv_cache:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      t_thresh:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      use_distill:
+        sources: [fastvideo.configs.pipelines.longcat.LongCatT2V480PConfig, fastvideo.configs.pipelines.longcat.LongCatT2V704PConfig]
+      scheduler_arch:
+        sources: [fastvideo.configs.pipelines.sd35.SD35Config]
+      text_encoder_archs:
+        sources: [fastvideo.configs.pipelines.sd35.SD35Config]
+      tokenizer_archs:
+        sources: [fastvideo.configs.pipelines.sd35.SD35Config]
+      transformer_arch:
+        sources: [fastvideo.configs.pipelines.sd35.SD35Config]
+      vae_arch:
+        sources: [fastvideo.configs.pipelines.sd35.SD35Config]
+      expand_timesteps:
+        sources:
+          - fastvideo.configs.pipelines.wan.FastWan2_2_TI2V_5B_Config
+          - fastvideo.configs.pipelines.wan.Wan2_2_TI2V_5B_Config
+      context_noise:
+        sources: [fastvideo.configs.pipelines.wan.MatrixGameI2V480PConfig]
+      num_frames_per_block:
+        sources: [fastvideo.configs.pipelines.wan.MatrixGameI2V480PConfig]
+    compatibility_only:
+      batch_size: "Gen3C inference-only tuning field pending typed batching design."
+      gradient_checkpointing: "Gen3C inference-only compatibility field pending typed batching design."
+      guidance_scale: "Gen3C pipeline-level default pending preset/default-request cleanup."
+      num_inference_steps: "Gen3C pipeline-level default pending preset/default-request cleanup."
+    internal_only:
+      audio_decoder_config: "Legacy internal component config object."
+      audio_decoder_precision: "Precision override pending dedicated component precision design."
+      vocoder_config: "Legacy internal component config object."
+      vocoder_precision: "Precision override pending dedicated component precision design."
+
+  sampling_param_base:
+    moved:
+      image_path: request.inputs.image_path
+      pil_image: request.inputs.pil_image
+      video_path: request.inputs.video_path
+      mouse_cond: request.inputs.mouse_cond
+      keyboard_cond: request.inputs.keyboard_cond
+      grid_sizes: request.inputs.grid_sizes
+      pose: request.inputs.pose
+      c2ws_plucker_emb: request.inputs.c2ws_plucker_emb
+      refine_from: request.inputs.refine_from
+      stage1_video: request.inputs.stage1_video
+      prompt: request.prompt
+      negative_prompt: request.negative_prompt
+      prompt_path: request.inputs.prompt_path
+      output_path: request.output.output_path
+      output_video_name: request.output.output_video_name
+      num_videos_per_prompt: request.sampling.num_videos_per_prompt
+      seed: request.sampling.seed
+      num_frames: request.sampling.num_frames
+      height: request.sampling.height
+      width: request.sampling.width
+      height_sr: request.sampling.height_sr
+      width_sr: request.sampling.width_sr
+      fps: request.sampling.fps
+      num_inference_steps: request.sampling.num_inference_steps
+      num_inference_steps_sr: request.sampling.num_inference_steps_sr
+      guidance_scale: request.sampling.guidance_scale
+      guidance_scale_2: request.sampling.guidance_scale_2
+      guidance_rescale: request.sampling.guidance_rescale
+      boundary_ratio: request.sampling.boundary_ratio
+      sigmas: request.sampling.sigmas
+      enable_teacache: request.runtime.enable_teacache
+      save_video: request.output.save_video
+      return_frames: request.output.return_frames
+      return_trajectory_latents: request.runtime.return_trajectory_latents
+      return_trajectory_decoded: request.runtime.return_trajectory_decoded
+      continuation_state: request.state
+      return_continuation_state: request.output.return_state
+    preset_owned:
+      t_thresh: request.stage_overrides.refine.t_thresh
+      spatial_refine_only: request.stage_overrides.refine.spatial_refine_only
+      num_cond_frames: request.stage_overrides.refine.num_cond_frames
+      trajectory_type: request.extensions.gen3c.trajectory_type
+      movement_distance: request.extensions.gen3c.movement_distance
+      camera_rotation: request.extensions.gen3c.camera_rotation
+      prompt_attention_mask: request.extensions.hyworld.prompt_attention_mask
+      negative_attention_mask: request.extensions.hyworld.negative_attention_mask
+      camera_states: request.extensions.hunyuangamecraft.camera_states
+      camera_trajectory: request.extensions.hunyuangamecraft.camera_trajectory
+      action_list: request.extensions.hunyuangamecraft.action_list
+      action_speed_list: request.extensions.hunyuangamecraft.action_speed_list
+      gt_latents: request.extensions.hunyuangamecraft.gt_latents
+      conditioning_mask: request.extensions.hunyuangamecraft.conditioning_mask
+      ltx2_cfg_scale_video: request.extensions.ltx2.cfg_scale_video
+      ltx2_cfg_scale_audio: request.extensions.ltx2.cfg_scale_audio
+      ltx2_modality_scale_video: request.extensions.ltx2.modality_scale_video
+      ltx2_modality_scale_audio: request.extensions.ltx2.modality_scale_audio
+      ltx2_rescale_scale: request.extensions.ltx2.rescale_scale
+      ltx2_stg_scale_video: request.extensions.ltx2.stg_scale_video
+      ltx2_stg_scale_audio: request.extensions.ltx2.stg_scale_audio
+      ltx2_stg_blocks_video: request.extensions.ltx2.stg_blocks_video
+      ltx2_stg_blocks_audio: request.extensions.ltx2.stg_blocks_audio
+    internal_only:
+      data_type: "Derived from the request shape and not a public input."
+
+  sampling_param_extensions: {}
+
+  openai_image_request:
+    kept:
+      model: "HTTP adapter model-routing field."
+      response_format: "HTTP adapter response formatting field."
+      output_format: "HTTP adapter output-format field."
+      background: "HTTP adapter output-format field."
+      quality: "Compatibility field currently accepted by the adapter."
+      style: "Compatibility field currently accepted by the adapter."
+      user: "Compatibility field currently accepted by the adapter."
+    moved:
+      prompt: request.prompt
+      n: request.sampling.num_videos_per_prompt
+      size:
+        target: request.sampling.width,height
+        note: "Adapter parses OpenAI size strings as WIDTHxHEIGHT and forwards width then height."
+      num_inference_steps: request.sampling.num_inference_steps
+      guidance_scale: request.sampling.guidance_scale
+      true_cfg_scale: request.sampling.true_cfg_scale
+      seed: request.sampling.seed
+      negative_prompt: request.negative_prompt
+      enable_teacache: request.runtime.enable_teacache
+
+  openai_video_request:
+    kept:
+      model: "HTTP adapter model-routing field."
+    moved:
+      prompt: request.prompt
+      input_reference: request.inputs.image_path
+      reference_url: request.inputs.image_path
+      size:
+        target: request.sampling.width,height
+        note: "Adapter parses OpenAI size strings as WIDTHxHEIGHT and forwards width then height."
+      fps: request.sampling.fps
+      num_frames: request.sampling.num_frames
+      seed: request.sampling.seed
+      num_inference_steps: request.sampling.num_inference_steps
+      guidance_scale: request.sampling.guidance_scale
+      guidance_scale_2: request.sampling.guidance_scale_2
+      true_cfg_scale: request.sampling.true_cfg_scale
+      negative_prompt: request.negative_prompt
+      enable_teacache: request.runtime.enable_teacache
+      output_path: request.output.output_path
+    compatibility_only:
+      seconds:
+        target: request.sampling.num_frames
+        note: "HTTP adapter duration convenience field. If num_frames is omitted, the adapter computes num_frames = fps * seconds."
+
+cli:
+  notes:
+    - "CLI parity is checked against the actual generate/serve parser dest sets."
+    - "The inventory tracks parser dest names, excluding argparse's implicit help action."
+    - "The refactored inference CLI is config-only: subcommands expose only --config, and any additional CLI input must use dotted override paths."
+  generate:
+    explicit_local_fields:
+      - config
+    expected_dests:
+      - config
+  serve:
+    explicit_local_fields:
+      - config
+    expected_dests:
+      - config

docs/design/overview.md

@@ -12,7 +12,7 @@ FastVideo maps a Diffusers-style repo into a pipeline like this:
 - `fastvideo/configs/models/*`: arch configs and `param_names_mapping` for
   weight name translation.
 - `fastvideo/configs/pipelines/*`: pipeline wiring (component classes + names).
-- `fastvideo/configs/sample/*`: default runtime sampling parameters.
+- `fastvideo/api/sampling_param.py`: runtime sampling parameters.
 - `fastvideo/pipelines/basic/*`: end-to-end pipelines.
 - `fastvideo/pipelines/stages/*`: reusable pipeline stages.
 - `fastvideo/models/loader/*`: component loaders for Diffusers-style repos.
@@ -26,7 +26,7 @@ Minimal usage (from `examples/inference/basic/basic.py`):
 
 ```python
 from fastvideo import VideoGenerator
-from fastvideo.configs.sample import SamplingParam
+from fastvideo.api.sampling_param import SamplingParam
 
 model_id = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"  # or official_weights/<model_name>/
 generator = VideoGenerator.from_pretrained(model_id, num_gpus=1)
@@ -49,8 +49,9 @@ runtime parameters consistent:
 - `fastvideo/configs/models/`: architecture definitions, layer shapes, and
   `param_names_mapping` rules for key renaming.
 - `fastvideo/configs/pipelines/`: pipeline wiring and required components.
-- `fastvideo/configs/sample/`: default sampling parameters (steps, frames,
-  guidance scale, resolution, fps).
+- `fastvideo/api/sampling_param.py`: sampling parameters (steps, frames,
+  guidance scale, resolution, fps). Defaults come from profiles in
+  `fastvideo/pipelines/basic/<family>/profiles.py`.
 - `fastvideo/registry.py`: unified registry for pipeline config + sampling
   defaults and model metadata resolution, defined via explicit
   `register_configs(...)` blocks (no separate dict registries).
@@ -142,7 +143,7 @@ How this maps to FastVideo:
 - `T5TokenizerFast` -> loaded via HF in `fastvideo/models/loader/`
 - `UniPCMultistepScheduler` -> loaded via Diffusers scheduler utilities
 - Pipeline defaults -> `fastvideo/configs/pipelines/wan.py`
-- Sampling defaults -> `fastvideo/configs/sample/wan.py`
+- Sampling defaults -> `fastvideo/pipelines/basic/wan/profiles.py`
 
 ## Pipeline system
 

docs/design/server_contracts/streaming.md

@@ -0,0 +1,177 @@
+# Streaming WebSocket Server Contract
+
+The streaming server (`fastvideo/entrypoints/streaming/server.py`) speaks
+a JSON-over-WebSocket protocol with binary fMP4 chunks for media. This
+document is the authoritative spec for the message catalogue and the
+session state machine. Any change to either must update this document
+in the same PR that touches `protocol.py` or `session.py`.
+
+## Endpoint
+
+| Path | Protocol | Purpose |
+|---|---|---|
+| `WS /v1/stream` | WebSocket (JSON + binary) | Per-session realtime streaming |
+| `GET /health` | HTTP | Liveness probe (`status`, `stream_mode`, active `sessions`) |
+
+The server is launched by `fastvideo serve --config <serve.yaml>` when
+the config carries a `streaming:` block. Without that block the same CLI
+launches the OpenAI stateless HTTP server instead.
+
+## Connection lifecycle
+
+Every WebSocket connection holds exactly one `Session`. Sessions move
+through the states in `SessionState` (`fastvideo/entrypoints/streaming/session.py`).
+
+```
+                    ┌──────────────┐
+                    │ INITIALIZING │  ← WebSocket accepted, before init frame
+                    └──────┬───────┘
+                           │ session_init_v2 received
+            ┌──────────────┼──────────────┐
+            ▼              ▼              ▼
+         QUEUED      GPU_BINDING       REJECTED
+            │              │              ↑
+            │ slot ready   │              │ max-sessions hit
+            ▼              ▼              │ or invalid init
+                       ┌────────┐         │
+                       │ ACTIVE │ ────────┘
+                       └────┬───┘
+              segment loop  │
+                            │
+                ┌───────────┼───────────┐
+                ▼           ▼           ▼
+            COMPLETE      ERROR      TIMEOUT
+        (clean leave)  (any failure)  (idle / segment_cap reached)
+```
+
+Terminal states (`COMPLETE`, `ERROR`, `TIMEOUT`, `REJECTED`) are sinks —
+no transitions out. The transition matrix is enforced in
+`session.py::_VALID_TRANSITIONS`; bad transitions raise.
+
+`SessionManager` enforces the per-process budgets pulled from
+`StreamingConfig`:
+
+- `session_timeout_seconds` — idle reaper drops sessions that haven't
+  advanced; non-terminal sessions transition to `TIMEOUT`.
+- `generation_segment_cap` — a session that hits the cap transitions to
+  `COMPLETE` after the last segment ships.
+
+## Message catalogue
+
+Every JSON frame carries `{"type": <str>, ...}`. Pydantic models in
+`protocol.py` are the source of truth; this table is the human-readable
+view.
+
+### Client → server
+
+| `type` | Required fields | Purpose |
+|---|---|---|
+| `session_init_v2` | — | Opening frame. Carries preset, curated prompts, optional initial image, feature toggles, optional `continuation_state` to resume from a snapshot. |
+| `segment_prompt_source` | `prompt` | Request the next segment using the supplied prompt; optional sampling overrides (`seed`, `num_inference_steps`, `guidance_scale`, `negative_prompt`). |
+| `seed_prompts_updated` | `seed_prompts` | Replace the session's seed-prompt list; takes effect on the next segment. |
+| `enhancement_updated` | `enabled` | Toggle prompt enhancement for subsequent segments. |
+| `auto_extension_updated` | `enabled` | Toggle automatic per-segment prompt extension. |
+| `loop_generation_updated` | `enabled` | Toggle loop-generation mode. |
+| `generation_paused_updated` | `paused` | Pause/resume segment generation; queued requests defer. |
+| `snapshot_state` | — | Request the current `ContinuationState` for export; server replies with `continuation_state_snapshot`. |
+
+The opening frame must be `session_init_v2`. Any other first frame is
+rejected with an `error` (code `invalid_message`) and the WebSocket is
+closed.
+
+### Server → client
+
+| `type` | Carries | When emitted |
+|---|---|---|
+| `queue_status` | `position`, `queue_depth` | After `session_init_v2` accepted, before GPU binding. |
+| `gpu_assigned` | GPU id, model id | Once a generator slot is bound. |
+| `ltx2_stream_start` | session-level metadata | Once the session enters `ACTIVE`. |
+| `ltx2_segment_start` | `segment_idx`, `prompt`, prompt source | When a `segment_prompt_source` request begins generation. |
+| `step_complete` | `segment_idx`, denoise timings | After the segment's denoising loop finishes (before media emission). |
+| `media_init` | `segment_idx`, mime, stream id | First frame of fMP4 output for the segment. |
+| binary frame | fMP4 fragment bytes | Subsequent media chunks; the protocol enforces that `media_init` precedes any binary frames. |
+| `media_segment_complete` | `segment_idx`, chunk count, byte count | Last media chunk for the segment. |
+| `ltx2_segment_complete` | `segment_idx`, segment summary | Segment fully shipped; ready for the next `segment_prompt_source`. |
+| `ltx2_stream_complete` | session summary | Session reached `generation_segment_cap` or client requested clean shutdown. |
+| `session_timeout` | reason | Session hit `session_timeout_seconds`; immediately followed by close. |
+| `continuation_state_snapshot` | `kind`, `payload` | Reply to `snapshot_state`. The payload is the same shape produced by `LTX2ContinuationState.to_continuation_state(...)`. |
+| `error` | `code`, `message` | Any validation/runtime error. Non-fatal errors keep the connection open; fatal errors precede a `close`. |
+
+## Continuation state
+
+The session optionally accepts a `continuation_state` dict inside the
+opening `session_init_v2` frame. When present, the server hydrates it
+into a `ContinuationState(kind, payload)` envelope and feeds it as the
+`request.state` on the first segment's `GenerationRequest` — letting a
+client resume after a disconnect, migrate sessions across processes,
+or replay a prior session.
+
+After every segment, if the runtime returns a fresh state, the server
+persists it to the `SessionStore` so a `snapshot_state` request can
+export it. The store and serialization contracts live with the model
+family (e.g. `fastvideo/pipelines/basic/ltx2/continuation.py` for LTX-2).
+
+## Example flow
+
+```
+client                                                    server
+──────                                                    ──────
+WS /v1/stream  ─────── connect ─────────────────────────►
+               ◄────── (accept)
+
+{"type": "session_init_v2",
+ "preset": "ltx2_two_stage",
+ "curated_prompts": ["a fox in snow", "the fox jumps"],
+ "initial_image": {...},
+ "stream_mode": "av_fmp4"} ─────────────────────────────►
+
+                                                          (validate, queue, bind)
+               ◄──── {"type": "queue_status",
+                       "position": 0, "queue_depth": 0}
+               ◄──── {"type": "gpu_assigned",
+                       "gpu_id": 0, "model_id": "..."}
+               ◄──── {"type": "ltx2_stream_start", ...}
+
+{"type": "segment_prompt_source",
+ "prompt": "a fox in snow",
+ "source": "curated"} ───────────────────────────────────►
+                                                          (run pipeline)
+               ◄──── {"type": "ltx2_segment_start",
+                       "segment_idx": 1, ...}
+               ◄──── {"type": "step_complete",
+                       "segment_idx": 1, "timings": {...}}
+               ◄──── {"type": "media_init",
+                       "segment_idx": 1,
+                       "mime": "video/mp4", ...}
+               ◄──── <binary fMP4 init segment>
+               ◄──── <binary fMP4 fragment>
+               ◄──── <binary fMP4 fragment>
+               ◄──── {"type": "media_segment_complete",
+                       "segment_idx": 1, "chunks": 12}
+               ◄──── {"type": "ltx2_segment_complete",
+                       "segment_idx": 1, ...}
+
+{"type": "segment_prompt_source",
+ "prompt": "the fox jumps"} ─────────────────────────────►
+                                                          (segment 2 …)
+
+{"type": "snapshot_state"} ──────────────────────────────►
+               ◄──── {"type": "continuation_state_snapshot",
+                       "kind": "ltx2.v1",
+                       "payload": {"schema_version": 1, ...}}
+
+(close) ──────────────────────────────────────────────────►
+                                                          (session → COMPLETE)
+```
+
+## Backward / forward compatibility
+
+- Adding a new client message: append a Pydantic model to `protocol.py`
+  with a unique `type`; add the discriminator entry to `ClientMessage`;
+  add a row to the table above. Old clients that don't send the new
+  message remain compatible.
+- Adding a new server message: emit only when a new feature flag is
+  enabled (or always emit, since clients ignore unknown types).
+- Changing an existing message: bump the `type` (e.g. `session_init_v2`
+  → `session_init_v3`) and accept both for one release cycle. Never
+  silently change field semantics under the same `type`.

docs/design/training_architecture.md

@@ -0,0 +1,369 @@
+# Training Architecture
+
+!!! warning "Work in Progress"
+    This training architecture (`fastvideo/train/`) is under active development
+    and is replacing the older `fastvideo/training/` module. APIs, config
+    formats, and supported methods may change. See the
+    [Current Status](#current-status) section for what is implemented so far.
+
+FastVideo's training framework (`fastvideo/train/`) is built around a
+**pluggable, YAML-driven architecture** that cleanly separates **models**,
+**training methods**, and **infrastructure** into independent, composable
+layers. A single YAML config file is all that is needed to train any supported
+model with any supported algorithm — no code changes required to mix and match.
+
+---
+
+## Motivation
+
+Training video diffusion models involves a tangle of concerns: model loading,
+noise scheduling, distillation algorithms, distributed strategies,
+checkpointing, and validation. Existing training scripts tend to hard-wire
+these together, making it painful to:
+
+1. **Try a new distillation algorithm** on an existing model (requires forking
+   the training loop).
+2. **Add a new model** to an existing algorithm (requires re-implementing
+   boilerplate).
+3. **Switch distributed strategies** (FSDP, TP, SP) without touching algorithm
+   code.
+4. **Resume, checkpoint, and validate** uniformly across all combinations.
+
+The training framework solves this by making each axis of variation an
+independent plugin.
+
+---
+
+## Architecture Overview
+
+```
+YAML Config
+    |
+    v
++------------------+     +---------------------+     +------------------+
+|   Models Layer   |     |   Methods Layer     |     | Infrastructure   |
+|  (per-role)      |     |   (algorithm)       |     |   Layer          |
+|                  |     |                     |     |                  |
+| - ModelBase      |<----|  - TrainingMethod   |---->| - Trainer        |
+| - CausalModelBase|     |  - single_train_step|     | - Callbacks      |
+|                  |     |  - backward         |     | - Checkpoint     |
+| Roles:           |     |  - optimizers       |     | - Tracker (W&B)  |
+|  student         |     |                     |     | - Dataloader     |
+|  teacher         |     | Algorithms:         |     |                  |
+|  critic          |     |  DMD2, SelfForcing, |     | Distributed:     |
+|                  |     |  SFT, DFSFT         |     |  HSDP, TP, SP    |
++------------------+     +---------------------+     +------------------+
+```
+
+### Three Layers
+
+| Layer | Responsibility | Extension point |
+|-------|---------------|-----------------|
+| **Models** (`fastvideo/train/models/`) | Load transformer + scheduler, define `predict_noise`, `predict_x0`, `add_noise`, `backward`. Each training role (student/teacher/critic) is an independent instance. | Subclass `ModelBase` (or `CausalModelBase` for streaming). |
+| **Methods** (`fastvideo/train/methods/`) | Implement the training algorithm: own role models, define `single_train_step` + `backward`, manage optimizers/schedulers. | Subclass `TrainingMethod`. |
+| **Infrastructure** (`fastvideo/train/trainer.py`, `utils/`, `callbacks/`) | Training loop, gradient accumulation, distributed setup, checkpointing (DCP), W&B tracking, validation, EMA, grad clipping. | Add callbacks; everything else is shared. |
+
+---
+
+## YAML-Driven Configuration
+
+Everything is configured declaratively. The `_target_` field selects the Python
+class to instantiate:
+
+```yaml
+models:
+  student:
+    _target_: fastvideo.train.models.wan.WanModel
+    init_from: Wan-AI/Wan2.1-T2V-1.3B-Diffusers
+    trainable: true
+  teacher:
+    _target_: fastvideo.train.models.wan.WanModel
+    init_from: Wan-AI/Wan2.1-T2V-1.3B-Diffusers
+    trainable: false
+  critic:
+    _target_: fastvideo.train.models.wan.WanModel
+    init_from: Wan-AI/Wan2.1-T2V-1.3B-Diffusers
+    trainable: true
+
+method:
+  _target_: fastvideo.train.methods.distribution_matching.dmd2.DMD2Method
+  rollout_mode: simulate
+  dmd_denoising_steps: [1000, 850, 700, 550, 350, 275, 200, 125]
+  generator_update_interval: 5
+  real_score_guidance_scale: 3.5
+  # ...
+
+training:
+  distributed: { num_gpus: 8, sp_size: 1, tp_size: 1 }
+  data: { data_path: ..., num_latent_t: 20, num_frames: 77 }
+  optimizer: { learning_rate: 2.0e-6, betas: [0.0, 0.999] }
+  loop: { max_train_steps: 4000 }
+  checkpoint: { output_dir: outputs/my_run }
+
+callbacks:
+  grad_clip: { max_grad_norm: 1.0 }
+  validation: { pipeline_target: ..., every_steps: 100 }
+```
+
+To switch from DMD2 to SFT, change the `method._target_` and remove the
+teacher/critic — no code changes needed.
+
+---
+
+## Model Abstraction
+
+### `ModelBase` — Standard (Bidirectional) Models
+
+Every role gets its own `ModelBase` instance owning a `transformer` and
+`noise_scheduler`. The base class defines:
+
+- **`prepare_batch()`** — Convert raw dataloader output into forward-ready
+  `TrainingBatch`.
+- **`add_noise()`** — Apply forward-process noise at a given timestep.
+- **`predict_noise()` / `predict_x0()`** — Run the transformer and return
+  predictions.
+- **`backward()`** — Backward pass that restores forward context (attention
+  metadata, timesteps).
+- **`init_preprocessors()`** — Lazy-load VAE, build dataloader (called only on
+  the student).
+
+### `CausalModelBase` — Streaming / Causal Models
+
+Extends `ModelBase` with streaming inference primitives for causal video
+generation:
+
+```python
+class CausalModelBase(ModelBase):
+    def clear_caches(self, *, cache_tag: str = "pos") -> None: ...
+    def predict_noise_streaming(
+        self, ..., cache_tag, store_kv, cur_start_frame
+    ) -> Tensor | None: ...
+    def predict_x0_streaming(
+        self, ..., cache_tag, store_kv, cur_start_frame
+    ) -> Tensor | None: ...
+```
+
+KV caches are **internal** to the model instance, keyed by `cache_tag`. The
+method controls when to store (`store_kv=True`) vs. read-only
+(`store_kv=False`), enabling block-by-block causal rollout during training.
+
+---
+
+## Training Methods
+
+### DMD2 (Distribution Matching Distillation)
+
+**Roles:** student (trainable) + teacher (frozen) + critic (trainable)
+
+The student learns to generate clean video in few steps by matching the
+teacher's score function, with a critic network providing a learned fake-score
+baseline.
+
+- **Rollout modes:**
+  - `simulate` — Student starts from pure noise and iteratively denoises
+    through the full step schedule.
+  - `data_latent` — Student denoises from a single randomly-noised data
+    sample.
+- **Losses:** Generator loss (DMD gradient) + critic flow-matching loss, with
+  alternating updates (`generator_update_interval`).
+
+### Self-Forcing (Causal DMD)
+
+**Roles:** student (causal, trainable) + teacher (frozen) + critic (trainable)
+
+Extends DMD2 for **streaming/causal video generation**. The key idea: during
+training, the student processes video in temporal chunks, using its own
+previously-denoised outputs as context for future chunks — simulating online
+autoregressive rollout.
+
+- Video is split into blocks of `chunk_size` latent frames.
+- Each block is denoised through the student's step schedule; a random
+  early-exit step is sampled per block.
+- After denoising a block, its output is fed back (with optional
+  `context_noise`) as KV cache context for subsequent blocks via
+  `predict_noise_streaming(store_kv=True)`.
+- Supports SDE and ODE sampling during rollout.
+- Selective gradient control: `enable_gradient_in_rollout`,
+  `start_gradient_frame`.
+
+### Supervised Fine-Tuning (SFT)
+
+**Roles:** student only
+
+Standard flow-matching loss between predicted and ground-truth noise/x0.
+
+### Diffusion-Forcing SFT (DFSFT)
+
+**Roles:** student only
+
+SFT with **inhomogeneous (per-chunk) timesteps** — each temporal chunk in a
+video gets a different noise level. This trains the model to handle mixed-noise
+inputs, which is a prerequisite for causal/streaming inference where earlier
+frames are cleaner than later ones.
+
+---
+
+## Training Loop
+
+The `Trainer` runs a standard loop with pluggable method and callbacks:
+
+```
+for step in range(start_step, max_steps):
+    for accum_iter in range(grad_accum_steps):
+        batch <- dataloader
+        loss_map, outputs, metrics <- method.single_train_step(batch, step)
+        method.backward(loss_map, outputs)
+
+    callbacks.on_before_optimizer_step()   # grad clipping
+    method.optimizers_schedulers_step()
+    method.optimizers_zero_grad()
+    callbacks.on_training_step_end()       # logging
+    checkpoint_manager.maybe_save(step)
+    callbacks.on_validation_begin()        # periodic inference
+```
+
+### Callbacks
+
+- **GradNormClipCallback** — Per-module gradient norm logging + global
+  clipping.
+- **ValidationCallback** — Periodic inference sampling with configurable
+  pipeline, sampling steps, and guidance scale.
+- **EMACallback** — Exponential moving average of student weights.
+
+### Checkpointing
+
+- DCP (Distributed Checkpoint) format, compatible with FSDP/HSDP.
+- Saves: model weights, optimizer states, scheduler states, RNG states (per
+  role).
+- Full resume support: auto-restores step counter and all RNG states.
+
+---
+
+## Getting Started
+
+```bash
+# Install
+uv pip install -e .[dev]
+
+# Run DMD2 distillation on Wan 2.1
+torchrun --nproc_per_node=8 -m fastvideo.train.entrypoint.train \
+    --config examples/train/distill_wan2.1_t2v_1.3B_dmd2.yaml
+
+# Run SFT fine-tuning
+torchrun --nproc_per_node=8 -m fastvideo.train.entrypoint.train \
+    --config examples/train/finetune_wan2.1_t2v_1.3B_vsa_phase3.4_0.9sparsity.yaml
+```
+
+Example configs are in `examples/train/`.
+
+---
+
+## File Structure
+
+```
+fastvideo/train/
+  trainer.py                  # Training loop
+  models/
+    base.py                   # ModelBase, CausalModelBase ABCs
+    wan/wan.py                # Wan 2.1 T2V model plugin
+    wangame/wangame.py        # WanGame 2.1 I2V model plugin
+    wangame/wangame_causal.py # WanGame causal (streaming) plugin
+  methods/
+    base.py                   # TrainingMethod ABC
+    distribution_matching/
+      dmd2.py                 # DMD2 distillation
+      self_forcing.py         # Self-Forcing (causal DMD)
+    fine_tuning/
+      finetune.py             # Supervised fine-tuning
+      dfsft.py                # Diffusion-forcing SFT
+  callbacks/
+    grad_clip.py              # Gradient clipping + norm logging
+    validation.py             # Periodic inference validation
+    ema.py                    # EMA weight averaging
+  entrypoint/
+    train.py                  # CLI entrypoint (torchrun)
+  utils/
+    config.py                 # YAML parser -> RunConfig
+    builder.py                # build_from_config: model/method instantiation
+    training_config.py        # TrainingConfig dataclass
+    dataloader.py             # Dataset/dataloader construction
+    optimizer.py              # Optimizer/scheduler construction
+    checkpoint.py             # DCP save/resume
+    tracking.py               # W&B tracker
+```
+
+---
+
+## Current Status
+
+| Component | Status |
+|-----------|--------|
+| Core framework (trainer, config, callbacks) | Implemented and tested |
+| `WanModel` (Wan 2.1 T2V) | Implemented and tested |
+| `WanGameModel` (WanGame 2.1 I2V) | Implemented and tested |
+| `WanGameCausalModel` (streaming) | Implemented and tested |
+| `WanCausalModel` (Wan T2V causal) | In progress |
+| DMD2 method | Implemented and tested |
+| Self-Forcing method | Implemented and tested |
+| SFT method | Implemented and tested |
+| DFSFT method | Implemented and tested |
+| DCP checkpointing + resume | Implemented and tested |
+| EMA callback | Implemented |
+| Validation callback | Implemented and tested |
+| Causal DMD inference pipeline | Implemented |
+
+---
+
+## Open Questions
+
+We welcome community feedback on the following topics:
+
+### Model Plugin API
+
+The current `ModelBase` interface requires implementing 6 methods. Is this the
+right granularity?
+
+- Should `prepare_batch` be split into separate concerns (noise sampling,
+  timestep sampling, attention metadata)?
+- Should `backward` be lifted out of the model and into the method/trainer?
+
+### Causal Streaming Interface
+
+`CausalModelBase` adds `predict_noise_streaming` / `predict_x0_streaming` with
+cache management. Alternatives considered:
+
+- **(a) Current:** Cache is internal to the model, keyed by `cache_tag`.
+  Simple but couples cache lifecycle to model.
+- **(b) External cache:** Method owns the cache dict, passes it into predict
+  calls. More explicit but verbose.
+- **(c) Context manager:** `with model.streaming_context(tag) as ctx: ...` —
+  cleaner lifecycle but harder to compose.
+
+### Method Composition
+
+Currently, methods are monolithic classes. Should we support composing methods
+(e.g., DFSFT pre-training followed by Self-Forcing distillation) within a
+single config? Or is sequential training with checkpoint handoff sufficient?
+
+### New Models and Methods
+
+What models and training methods should we prioritize next?
+
+- **Models:** HunyuanVideo, CogVideoX, other Wan variants?
+- **Methods:** Consistency models, progressive distillation, reward-based
+  fine-tuning?
+
+### Distributed Strategy
+
+Currently supports HSDP (hybrid sharded data parallel) + TP + SP. Are there
+scenarios where the current distributed setup is insufficient? Should we add
+pipeline parallelism for very large models?
+
+---
+
+## References
+
+- [Self-Forcing paper](https://arxiv.org/abs/2406.05477) — Chen et al., 2024.
+- [DMD2 paper](https://arxiv.org/abs/2405.14867) — Yin et al., 2024.
+- [Diffusion Forcing paper](https://arxiv.org/abs/2407.01392) — Chen et al.,
+  2024.

docs/distillation/dmd.md

@@ -16,7 +16,8 @@ Both models are trained on **61×448×832** resolution but support generating vi
 First install [VSA](../attention/vsa/index.md). Set `MODEL_BASE` to your own model path and run:
 
 ```bash
-bash scripts/inference/v1_inference_wan_dmd.sh
+FASTVIDEO_ATTENTION_BACKEND=VIDEO_SPARSE_ATTN \
+  fastvideo generate --config scripts/inference/inference_wan_VSA_DMD_1_3B.yaml
 ```
 
 ## 🗂️ Dataset
@@ -85,3 +86,25 @@ sbatch examples/distill/Wan2.2-TI2V-5B-Diffusers/Data-free/distill_dmd_t2v_5B.sh
 - Learning rate: 2e-5
 - Training steps: 3000 (~12 hours)
 - HSDP shard dim: 1
+
+## 🧭 Note on `real_score_guidance_scale`
+
+The teacher CFG used inside the DMD loss follows the DMD2 reference
+implementation and uses the parameterization
+
+```
+x = x_cond + w * (x_cond - x_uncond)
+```
+
+rather than the Ho & Salimans form `x_uncond + w * (x_cond - x_uncond)`. The
+two are mathematically equivalent up to a constant offset:
+
+| `real_score_guidance_scale` (`w`) | Equivalent standard CFG (`w + 1`) | Output                |
+|-----------------------------------|-----------------------------------|-----------------------|
+| `-1`                              | `0`                               | unconditional         |
+| `0`                               | `1`                               | conditional           |
+| `3.5` (default)                   | `4.5`                             | strong guidance       |
+
+So `real_score_guidance_scale` should be read as the **extra** guidance
+strength added on top of the conditional prediction. When porting values
+from a paper that uses the Ho & Salimans form, subtract 1.