mirror of
https://github.com/huggingface/diffusers.git
synced 2026-10-23 09:09:01 +03:00
v0.39.0
1300 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
Sayak Paul
|
4f37263ab7 |
[chore] update to 2026 finally. (#14079)
update to 2026 finally. |
||
Kashif Rasul
|
1f9c201083 |
Add doc pages for the DiffusionGemma schedulers (#14092)
* add doc pages for the discrete ddim and entropy bound schedulers Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * fix the cited paper titles to match the linked papers Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> |
||
|
Kashif Rasul
|
21ba39457d |
[discrete diffusion] Add DiffusionGemma pipeline and schedulers (#13986)
* Add discrete DDIM and entropy bound schedulers and a uniform mode for block refinement * Add DiffusionGemma block-diffusion pipeline * Add DiffusionGemma pipeline tests and docs * Put DiffusionGemma docs under the Text pipelines section * Add static cache and fullgraph-compiled decoder path to DiffusionGemma pipeline * Compile decoder externally for the static cache path instead of a pipeline flag * Prefill the encoder once into a reusable cache and sync default denoising steps * Support image prompts by forwarding pixel_values to the encoder prefill * Restyle docstrings to satisfy doc-builder * Sort the new scheduler and pipeline exports * Let any of the three schedulers drive the pipeline * Document the schedulers and updated defaults in the pipeline docs * Sort the scheduler dummy objects * Set scheduler sampling knobs on the scheduler config, not the pipeline call * Accept raw prompt/image/messages instead of pre-tokenized model inputs * Add leave-one-out predictor-corrector to DiscreteDDIM scheduler Adds optional Gibbs corrector sweeps after each predictor step for uniform diffusion, recovering the LOO denoiser in closed form so it works on the released checkpoint with no retraining. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Forward PEFT adapter API on the DiffusionGemma pipeline The denoiser is a Transformers model, so adapters (LoRA, DoRA, ...) load through its native PEFT integration rather than the diffusers LoRA loader. Also dispatch the predictor-corrector by scheduler capability instead of class. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Fix callback kwargs gathering on Python < 3.12 Build callback_kwargs with a loop instead of a dict comprehension, whose own scope hides locals() on pre-3.12 (PEP 709), causing KeyError: 'canvas'. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Speed up diffusion gemma sampling to match transformers Add adaptive stopping (stop a block once its prediction is stable and confident) and make the decoder compile cudagraph-safe via cudagraph_mark_step_begin + logits clone. ~175 -> ~372 tok/s. Also align the decoder mask with the new transformers#46654 layout. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Enable adaptive stopping by default Default confidence_threshold to 0.005 to match the released checkpoint and transformers, so the speedup is on out of the box. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Fold corrector sweeps into the step budget Run fewer predictor steps and spend the freed forwards on the corrector, so predictor-corrector sampling costs the same total forwards as plain ancestral (~2x faster), matching the paper. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Commit the converged prediction on adaptive stop Ancestral schedulers like DiscreteDDIM only clean the canvas on the final step, so stopping early left noise tokens. Use the denoiser argmax instead, which is the converged answer and matches the canvas for commit schedulers. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * anneal sampling temperature and fix static cache decoder mask Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Update src/diffusers/pipelines/diffusion_gemma/pipeline_diffusion_gemma.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update src/diffusers/schedulers/scheduling_entropy_bound.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update src/diffusers/pipelines/diffusion_gemma/pipeline_diffusion_gemma.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * address dg845 review comments * fix entropy scheduler temperature scaling * update decoder mask for new transformers * self-condition on the temperature-shaped logits * move temperature annealing into EntropyBoundScheduler * self-condition on the entropy scheduler's shaped logits * show torch.compile + static cache in the usage example * removed wrong commit * Update src/diffusers/pipelines/diffusion_gemma/pipeline_diffusion_gemma.py Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> * address review comments * expose pred_logits on all schedulers --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> |
||
apolinário
|
3993de59e3 |
Krea 2 LoRA DreamBooth trainer (#14046)
* Add Krea 2 LoRA DreamBooth trainer + Krea2 LoRA loader Adds examples/dreambooth/train_dreambooth_lora_krea2.py and the Krea2LoraLoaderMixin (Krea2Pipeline now inherits it) so LoRA adapters can be saved/loaded for the Krea 2 transformer. The transformer/pipeline forward now accept attention_kwargs for runtime LoRA scaling. Co-Authored-By: linoytsaban <linoy.tsaban@gmail.com> * inline _pack/_unpack_latents in trainer (pipeline methods are instance methods) * add Krea2LoraLoaderMixin to LoRA loaders docs * remove mu from validation call to align with pipeline on main --------- Co-authored-by: linoytsaban <linoy.tsaban@gmail.com> Co-authored-by: Linoy Tsaban <57615435+linoytsaban@users.noreply.github.com> |
||
YiYi Xu
|
7104cb43c3 |
Add Krea 2 (K2) text-to-image pipeline and transformer (#14045)
Adds Krea2Transformer2DModel and Krea2Pipeline: a flow-matching text-to-image MMDiT with grouped-query attention, Qwen3-VL text conditioning, and the Qwen-Image VAE. Co-authored-by: Elea Zhong <elea@krea.ai> Co-authored-by: Naga Sai Abhinay Devarinti <devarintinagasaiabhinay@gmail.com> |
||
Linoy Tsaban
|
3467efa65a |
Add Ideogram4LoraLoaderMixin (LoRA loading for Ideogram4) (#13921)
* add Ideogram4LoraLoaderMixin * Support loading non-diffusers Ideogram4 LoRA checkpoints (#13919) support loading non-diffusers Ideogram4 LoRAs * add Ideogram4 LoRA loader tests * support call-time LoRA scaling via attention_kwargs in Ideogram4 * fix and un-skip Ideogram4 LoRA loader tests * document attention_kwargs in Ideogram4 forward and pipeline * style Ideogram4 attention_kwargs docstrings * fix Ideogram4 LoRA loader CI test failures - pipeline: run the text encoder on its parameters' current device, then move features to the execution device, so encode_prompt works under enable_model_cpu_offload. The pipeline calls the text encoder's submodules directly to tap intermediate layers, which bypasses accelerate's onload hook, so the weights stay on CPU while inputs are on the execution device. Fixes test_lora_loading_model_cpu_offload. - tests: override test_lora_fuse_nan to corrupt a weight under Ideogram4's `layers` tower (the base test probes transformer_blocks/blocks/etc.). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * address review nits on Ideogram4 LoRA loader - pipeline: clarify the te_device comment (per review) — explain the CpuOffload hook attaches to forward, why submodule calls bypass it, and that te_device is the offload device under enable_model_cpu_offload. - tests: drop the unused `import sys` and `sys.path.append(".")`. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> |
||
David Bertoin
|
a487e2f131 |
Add PRXPixelPipeline: pixel-space PRX text-to-image pipeline (#13928)
Adds a pixel-space variant of PRX that denoises raw RGB directly (no VAE), conditioned on a Qwen3-VL text encoder: - PRXTransformer2DModel: new optional config args `bottleneck_size` (two-layer img_in projection for large patch dims) and `resolution_embeds` (PRXResolutionEmbedder conditions the timestep modulation on the latent resolution) - PRXPipeline: support for subclass-tuned tokenizer max length, light text cleaning, x-prediction flow matching (x0 -> velocity conversion), and non-unit initial noise scale - PRXPixelPipeline: thin subclass wiring the above together (vae optional/None, vae_scale_factor=1, 1024px default) - conversion script support for the pixel checkpoint format - registration in __init__ files + dummy objects, docs autodoc entry, fast pipeline tests Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
yzhautouskay
|
bcbfcd43da |
Add Cosmos3 video2video generation support (#13896)
* Init v2v cosmos3 commit * Add user quide; prompt upsampling is TBD * Apply style fixes --------- Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> |
||
|
YiYi Xu
|
12fc496d3c |
keep the agent symlinks (#13968)
* Commit root AGENTS.md/CLAUDE.md symlinks so agents auto-load conventions Track the root AGENTS.md/CLAUDE.md symlinks (-> .ai/AGENTS.md) instead of generating them via `make claude`/`make codex`, so a fresh clone auto-loads the agent guide with no setup step (matching transformers). The make targets now only wire up skills; .ai/AGENTS.md, the contribution doc, and the PR template are updated to match. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Add documentation-impact check to PR review rules During review, scan related usage docs and agent docs and flag (advisory) any that should be updated based on the PR — outdated/missing usage docs, or a rule/pattern/gotcha worth recording in the agent guides. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
YiYi Xu
|
4757c7c465 |
Make root PHILOSOPHY.md a symlink to the docs philosophy page (#13954)
* Make root PHILOSOPHY.md a symlink to the docs philosophy page The root PHILOSOPHY.md had drifted into a stale duplicate of docs/source/en/conceptual/philosophy.md (which was rewritten in #13808 to cover the single-file model policy and Modular Diffusers). Replace it with a symlink to the docs file, mirroring CONTRIBUTING.md -> contribution.md, so there's a single source of truth. A symlink renders as a path stub (not content) on GitHub, so repoint the links that pointed at blob/main/PHILOSOPHY.md to the rendered docs site: the PR template and the English contribution guide. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Point contributor-guide links at the rendered docs site CONTRIBUTING.md is a symlink, so blob/main/CONTRIBUTING.md renders as a path stub on GitHub rather than the guide. Repoint the PR template and README links to the rendered docs page, matching the philosophy/AI-agents links. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Repoint remaining symlink-stub doc links to the rendered docs site The stale-bot comment and the ko/zh contribution guides still linked to blob/main/{CONTRIBUTING,PHILOSOPHY}.md, which render as path stubs because those root files are symlinks. Point them at the rendered docs pages (localized /ko/ and /zh/ for the translations). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> |
||
liwd190019
|
93d3296b54 |
Add tutorial translations in Chinese (#13932)
Add Chinese tutorial translations |
||
|
YiYi Xu
|
141852f532 |
[.ai] add self-review skill (#13917)
* [.ai] add self-review skill, retire parity-testing skill, and tighten the agent guides - New `self-review` skill mirroring the `@claude` CI review (rubric from review-rules.md, call-path dead-code analysis), report-only, with the report flagging what to fix before submitting (blocking + dead code) vs what to leave for the actual review. - Remove the WIP `parity-testing` skill; preserve its pitfalls as `model-integration/pitfalls.md` (numerical-discrepancy reference). - model-integration: restructure around a grouped checklist, default-to-modular, an overall file-structure sketch (details deferred to the guides), a fresh-conversion `Model parity test` example (internal, not shipped), and a filled-in weight/checkpoint-conversion section. - Centralize the loading rule (from_pretrained / from_single_file, no custom loaders) in models.md; add per-folder File structure sections to models.md / pipelines.md; default-to-modular note in pipelines.md. - AGENTS.md: dedicated 'Self-review before a PR' and 'Reference guides' sections. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * [.ai] simplify pitfalls #6 and drop the model-storage / injection-test entries Trim pitfall #6 to the essential point (small dtype diffs compound into a large final difference), remove the `/tmp` model-storage and incomplete-injection-test pitfalls, and renumber 1-16 with cross-references updated. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * [.ai] drop parity-harness-specific pitfalls With the parity-testing skill gone, remove the stale-test-fixtures pitfall (saved tensors / cross-pipeline fixtures no longer apply) and de-jargon the noise-dtype detection note. Keeps the pitfalls list generic to numerical discrepancy. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * [.ai] trim pitfalls to a concise possible-causes reference Drop the variable-shadowing and decoder-config pitfalls and the noise-dtype 'Detection' aside, tighten the remaining entries, renumber 1-12 (cross-refs updated), and reframe the intro as a non-checklist reference list of possible causes to consult only when outputs don't match. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Apply suggestion from @yiyixuxu * Apply suggestion from @yiyixuxu * [docs] update contributing guide for the self-review skill Replace the retired parity-testing skill with self-review in the skills list, and add a 'Self-review before opening' step to the AI-assisted contributions section: run the self-review skill / review-rules, fix blocking issues + dead code, and treat the @claude CI review as a non-authoritative helper (note any intentional skips in the PR for the reviewer). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Apply suggestions from code review Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> * [.ai] fix dangling pitfalls ref and broaden self-review scope - Drop the broken 'pitfalls.md #10' reference in the conversion step (the /tmp model-storage pitfall was removed); save to a local path instead. - Self-review now reviews the whole diff, not just src/diffusers/ and .ai/ — a contributor should review their own tests/docs/scripts too (the CI's scoping is a safety measure for untrusted PRs). Reword to 'same rubric as the CI'. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> |
||
Kailai Feng
|
a50ade4926 |
[Pipelines] Add DreamLite text-to-image and image-edit pipelines (#13815)
* feat(pipelines): add DreamLite text-to-image and image-edit pipelines
Add ByteDance's DreamLite model family to diffusers. DreamLite is a
UNet-based diffusion model that supports both text-to-image generation
and reference-image editing through a shared 3-branch dual-CFG design.
Two pipelines are shipped:
* DreamLitePipeline - full 3-branch dual CFG (negative,
reference, prompt); supports T2I and
I2I editing at 1024x1024.
* DreamLiteMobilePipeline - distilled single-branch variant for
on-device inference; no CFG.
New model code (all isolated under *_dreamlite.py / unet_dreamlite.py
to avoid touching shared upstream files):
* models/transformers/transformer_2d_dreamlite.py - DreamLite 2D
transformer block.
* models/unets/unet_dreamlite.py - DreamLiteUNetModel.
* models/unets/unet_2d_blocks_dreamlite.py - DreamLite-specific
down/up/mid blocks.
* models/resnet_dreamlite.py - DreamLite ResNet
variants.
* models/attention_processor.py - add
DreamLiteAttnProcessor2_0 (pure addition, no existing processor
modified).
Pipeline + tests + docs:
* pipelines/dreamlite/{__init__.py, pipeline_dreamlite.py,
pipeline_dreamlite_mobile.py, pipeline_output.py}.
* tests/pipelines/dreamlite/{test_pipeline_dreamlite.py,
test_pipeline_dreamlite_mobile.py} with the standard
PipelineTesterMixin suite; setUp/tearDown auto-patches encode_prompt
with a fake so MagicMock text encoders work without per-test
boilerplate.
* Skip 8 mixin tests that don't apply to DreamLite (MagicMock
serialisation, custom attention processor, encode_prompt return
shape, batch_size > 1 sweep), mirroring SD3 / Flux conventions.
* docs/source/en/api/pipelines/dreamlite.md + _toctree.yml entry
(alphabetically between DiT and EasyAnimate).
* Register exports in 6 __init__.py files.
Two real bugs surfaced by the mixin test suite are fixed in this
commit:
* num_images_per_prompt > 1: prompt_embeds and text_attention_mask
are now repeated along the batch dimension in both pipelines'
T2I and I2I branches before being passed to the UNet.
* vae=None: __init__ now guards the encoder_block_out_channels
lookup so encode_prompt can be tested in isolation per
PipelineTesterMixin convention.
SlowTests real-checkpoint resolution is set to 1024x1024 (the only
size DreamLite is trained for).
Test result: 27 passed, 50 skipped, 0 failed on CPU fast suite.
make style && make quality: clean.
* docs+tests(pipelines/dreamlite): pin Hub repos to `diffusers` branch
The `carlofkl/DreamLite-{base,mobile}` Hub repos host two flavours of the
same checkpoint:
* `main` branch - keeps `model_index.json` pointing at ByteDance's
internal package path so the original (non-diffusers)
reference code can still load these weights.
* `diffusers` branch - rewrites the `unet` entry of `model_index.json` to
`["diffusers", "DreamLiteUNetModel"]` so this
integration loads correctly from `diffusers`.
This commit pins every `from_pretrained(...)` call shipped with the
diffusers integration (docs examples, pipeline docstrings, SlowTests) to
`revision="diffusers"`. Local-override env vars (DREAMLITE_BASE_PATH /
DREAMLITE_MOBILE_PATH) still bypass the revision pin.
* chore(pipelines/dreamlite): sync `# Copied from` blocks + dummy objects after rebase
Mechanical changes after rebasing onto current `main`:
* `pipeline_dreamlite.py::retrieve_timesteps` — re-synced from
`diffusers.pipelines.flux.pipeline_flux.retrieve_timesteps` (PEP 604
type hints, expanded docstring, plus the new
`accepts_timesteps` / `accept_sigmas` introspection guards). DreamLite's
default code path uses `num_inference_steps` (uniform schedule) and never
passes custom `timesteps` / `sigmas`, so the added guards are dead-code
for this pipeline — behaviour is unchanged.
* `dummy_pt_objects.py` / `dummy_torch_and_transformers_objects.py` —
registered the dummy classes auto-generated by `make fix-copies` for
`DreamLiteTransformer2DModel`, `DreamLiteUNetModel`, `DreamLitePipeline`,
`DreamLiteMobilePipeline`, `DreamLitePipelineOutput`.
Generated by `make fix-copies`. No hand edits.
* docs(dreamlite): register attention processor + split combined docstring entries
- Register DreamLiteAttnProcessor2_0 in docs/source/en/api/attnprocessor.md
(fixes check_support_list.py).
- Split combined 'height / width' and 'guidance_scale / image_guidance_scale'
entries in the two pipeline docstrings; add a complete Args block to
DreamLiteTransformer2DModel.forward
(fixes check_forward_call_docstrings.py).
No behavioral change.
* refactor(dreamlite): address review feedback from #13815
- Inline the down/up block factories and define DreamLiteCrossAttn{,NoSelfAttn}{Down,Up}Block2D directly (review #1, #2)
- Rename DownBlock2DDreamLite/UpBlock2DDreamLite to DreamLiteDownBlock2D/DreamLiteUpBlock2D to match diffusers naming conventions (review #3, #4)
- Merge unet_2d_blocks_dreamlite.py into unet_dreamlite.py to mirror recent transformer model files (review #5)
- Wire max_sequence_length into the tokenizer call for generate mode (review #6)
- Replace hard-coded drop_idx values (64/34) with self.prompt_template_encode_*_start_idx attributes plus a comment explaining how the offsets are derived (review #7, #8)
- Drop the manual Image.resize call and rely on VaeImageProcessor's LANCZOS default in preprocess(image, height, width) (review #9)
- Use self.guidance_scale / self.image_guidance_scale properties in the CFG combine instead of the underscore-prefixed attributes (review #10, #11)
- Inline retrieve_latents / retrieve_timesteps / calculate_shift in the mobile pipeline with `# Copied from` markers, removing the cross-pipeline imports (review #12)
- Add `# Copied from` marker to _extract_masked_hidden in the mobile pipeline (review #13)
* refactor(dreamlite): address dg845 follow-up review
- Merge resnet_dreamlite.py (DepthwiseSeparableConv + ResnetBlock2DDreamLite)
into unet_dreamlite.py and delete the standalone module (review #1)
- Move DreamLiteAttnProcessor2_0 from attention_processor.py into
unet_dreamlite.py to keep all DreamLite-specific code in one place;
update docs autodoc reference accordingly (review #2)
- Drop the PyTorch 2.0 hasattr/ImportError check in
DreamLiteAttnProcessor2_0.__init__ (diffusers already requires
torch>=2.0; matches Wan deprecation) (review #3)
- Drop the deprecated `scale` argument handling from
DreamLiteAttnProcessor2_0.__call__ (new model, no legacy callers)
(review #4)
- Switch SDPA call to dispatch_attention_fn so all diffusers attention
backends (FlashAttention, FlashAttention-3, sageattention, etc.) are
selectable (review #5)
- Rename block dispatch keys in _get_{down,mid,up}_block_dreamlite to
match the Python class names (DreamLiteCrossAttn{Down,Up}Block2D /
DreamLiteCrossAttnNoSelfAttn{Down,Up}Block2D /
DreamLiteUNetMidBlock2DCrossAttn / DreamLite{Down,Up}Block2D);
default down/up/mid block_types in DreamLiteUNetModel and the test
fixtures are updated to the new keys (review #6, #7); the
carlofkl/DreamLite-{base,mobile} (diffusers branch) Hub configs are
being updated in lock-step
- Localize retrieve_latents inside pipeline_dreamlite.py with a
`# Copied from` marker, removing the cross-pipeline import; mirrors
the mobile pipeline (review #8)
- Add a check_inputs() method to both DreamLitePipeline and
DreamLiteMobilePipeline (mobile uses `# Copied from`); call it from
__call__; pulls the image-type validation out of prepare_image_latents
and adds prompt-type and h/w-divisibility checks (review #9)
* fix(dreamlite): correct Q/K/V layout for dispatch_attention_fn
dispatch_attention_fn expects (batch, seq, heads, head_dim) and handles the transpose internally; the previous code passed (batch, heads, seq, head_dim), which collided with the dispatch's internal transpose and broke inference (RuntimeError: tensor size mismatch at non-singleton dimension 1).
* test(dreamlite): swap MagicMock for tiny real Qwen3-VL fixture
Address dg845's review: rebuild the DreamLite fast-test fixture around a
real (tiny) Qwen3VLForConditionalGeneration + Qwen3VLProcessor so the
standard PipelineTesterMixin save/load, dtype, and offload tests run
end-to-end against the actual encode_prompt code path. Override
DreamLiteUNetModel.set_default_attn_processor to reinstall the GQA
processor so mixin utilities that round-trip through it keep working.
* Apply style fixes
* fix(dreamlite): address blocking review issues from #13815
- Override _no_split_modules / _repeated_blocks on DreamLiteUNetModel
with the actual DreamLite class names (BasicTransformerBlockDreamLite,
ResnetBlock2DDreamLite, DreamLiteCrossAttnUpBlock2D,
DreamLiteUpBlock2D) so device_map="auto" and compile_repeated_blocks()
match correctly.
- Keep attention masks as bool tensors in DreamLiteTransformer2DModel
instead of converting them to dense additive float biases. The dense
format hard-raises on flash / _flash_3 / _sage backends in
dispatch_attention_fn (which requires dtype == torch.bool).
- Add explicit parentheses around each clause in check_inputs's mixed
and/or condition (both pipelines) for readability.
- Replace nn.Module.__init__(self) with ModelMixin.__init__(self) in
DreamLiteUNetModel.__init__ so mixin state (e.g.
_gradient_checkpointing_func) is properly initialised. ConfigMixin /
PushToHubMixin don't define their own __init__, so this covers the
full chain without re-running UNet2DConditionModel.__init__.
* fix(dreamlite): forward all processor outputs to Qwen3VL text encoder
Recent versions of Qwen3VLProcessor add an mm_token_type_ids output, and
Qwen3VLModel.compute_3d_position_ids raises ValueError whenever
multimodal inputs are present (image_grid_thw is not None) but
mm_token_type_ids is None.
encode_prompt previously forwarded only input_ids / attention_mask /
pixel_values / image_grid_thw, dropping the new field and breaking the
fast pipeline tests against transformers main.
Switch to ``self.text_encoder(**tk_out, output_hidden_states=True)``
(matching NucleusMoEImagePipeline) so all processor outputs are
forwarded automatically and future additions don't regress this path.
* Apply style fixes
* docs(dreamlite): address final review nits from #13815
- Replace broken cat.png URL in editing examples (both base and mobile)
with the standard `huggingface/documentation-images` source used
elsewhere in the diffusers docs.
- Promote the recommended guidance_scale=3.5 / image_guidance_scale=1.5
to the default values of DreamLitePipeline.__call__, and drop the
now-redundant explicit args from the docs examples.
- Switch the EXAMPLE_DOC_STRING examples in both pipelines from
torch.float16 to torch.bfloat16 for consistency with the rest of the
docs.
---------
Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
|
||
Xin He
|
c31bb1c148 |
Integrate AutoRound into Diffusers (#13552)
* support auto_round Signed-off-by: Xin He <xin3.he@intel.com> * add document and unit tests Signed-off-by: Xin He <xin3.he@intel.com> * fix CI Signed-off-by: Xin He <xin3.he@intel.com> * Apply suggestions from code review Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * update document and overwrite the default quantization_config with specified backend. Signed-off-by: Xin He <xin3.he@intel.com> * add UT and fix bug Signed-off-by: Xin He <xin3.he@intel.com> * update per comments Signed-off-by: Xin He <xin3.he@intel.com> * update per comments Signed-off-by: Xin He <xin3.he@intel.com> * fix compile error in doc Signed-off-by: Xin He <xin3.he@intel.com> * Apply style fixes * small nits * Add auto_round dependency to the versions table Signed-off-by: Xin He <xin3.he@intel.com> * fix make deps_table_check_updated Signed-off-by: Xin He <xin3.he@intel.com> * fix CI Signed-off-by: Xin He <xin3.he@intel.com> --------- Signed-off-by: Xin He <xin3.he@intel.com> Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> |
||
|
YiYi Xu
|
5d10b4de3b |
[docs] update philosophy.md (finally) (#13808)
* [docs] refresh philosophy.md for current conventions Update the framing to match how Diffusers actually works now: models follow the single-file policy (with UNet/older transformers as legacy exceptions, not the model), add a Modular Pipelines section parallel to standard Pipelines, and slim Pipelines/Models bullets to remove stale guidance (per-task pipeline subclassing, generic-class checkpoint adaptation, dead links to deprecated repos). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * Update docs/source/en/conceptual/philosophy.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com> Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> |
||
|
yzhautouskay
|
bcc20e452b |
Add Cosmos3 action generation support (#13823)
* Add Cosmos3 action generation support * Add README action examples * Use do_classifier_free_guidance property * Remove unused method * Add action policy example to pipelines doc * Adding model selection for action example doc. * Remove redundant casts * Rename _pack_action_tokens to _prepare_action_segment * Move validation checks to check_inputs * Add action arguments in the __call__ docstring * Move action mode check to check_inputs * Rename action to action_tokens * Add warning for num_frames ovewrite attempt * Rename action_tokens to raw_actions * Remove scheduler config override * Refactor action to use CosmosActionCondition * Fix examples script to support flow_shift arg * Apply styling fixes * Remove CosmosActionCondition properties, move to pipeline * Replace validate wiht post init * Set height/width/num_frames to None, raise error if set ofr action * Fix action_dim default setting * Remove video argument before v2v is added * Fix None args * Add _EMBODIMENT_TO_RAW_ACTION_DIM mapping * Remove --raw-action-dim from README.md * Added prompt upsampler docs and examples * Apply style fixes * Bugfix for action mrope compression factor * Add action prompt json formatting, update docs * Add hand_pose raw_action_dim value * Add missing args to docstrings --------- Co-authored-by: Atharva Joshi <atjoshi@nvidia.com> Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: YiYi Xu <yixu310@gmail.com> |
||
|
apolinário
|
3e83f4348f |
Add structured prompt upsampling to Ideogram4 (#13860)
* Add structured prompt upsampling to Ideogram4 Rewrite prompts into Ideogram4's native structured JSON caption before encoding, opt-in via `prompt_upsampling=True` in `__call__` or the standalone `upsample_prompt`. Upsampling is driven by a generative `text_encoder` (`Qwen3VLForConditionalGeneration`, which carries the LM head); the head-less `Qwen3VLModel` is still supported for plain conditioning, and `upsample_prompt` raises an instructive error when the encoder cannot generate. Captions are schema-constrained via `outlines` when installed, and the modular pipeline gains a matching prompt-enhancer block. * Remove LM-head grafting; modular block uses a generative text_encoder Drop `graft_lm_head` and drive the modular `Ideogram4PromptUpsampleStep` off a generative `text_encoder` (`Qwen3VLForConditionalGeneration`), matching the standard pipeline: guard with `can_generate()` and an instructive error, and build the outlines logits processor lazily. Updates the copied `_get_text_encoder_hidden_states` to resolve the decoder for both encoder classes. * Style docstrings with doc-builder; mark prompt strings docstyle-ignore Reflow the Ideogram4 prompt-enhancer docstrings to the 119-col doc-builder style, and add `# docstyle-ignore` to the functional `CAPTION_SYSTEM_MESSAGE` and `CAPTION_USER_TEMPLATE` strings so the styler doesn't rewrap them (matching Flux2's `system_messages.py`). * Use an Ideogram4PromptEnhancerHead component for prompt upsampling Add `Ideogram4PromptEnhancerHead`, a small `ModelMixin` holding the Qwen3-VL LM head, as an optional `prompt_enhancer_head` pipeline component. Upsampling loads the head via a normal `from_pretrained` (its own repo, or bundled later) instead of an in-pipeline download, and grafts it onto the shared head-less `text_encoder` body so no second 8B body is loaded. Both the standard and modular pipelines build the generative model from `text_encoder` + the head; `upsample_prompt` raises an instructive error when the head component is absent. * Update src/diffusers/pipelines/ideogram4/pipeline_ideogram4.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update src/diffusers/pipelines/ideogram4/pipeline_ideogram4.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Apply suggestion from @apolinario * Apply suggestion from @dg845 Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Apply suggestion from @apolinario * Fix trailing whitespace * docs: add prompt-upsampling examples (remote API + local head) for Ideogram4 --------- Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> |
||
|
apolinário
|
9b0818cf87 |
Add Ideogram 4 (#13859)
* Add Ideogram 4 Adds the Ideogram 4 text-to-image model: transformer, standard pipeline, modular pipeline, docs, and tests. Checkpoint: ideogram-ai/ideogram-4-nf4 Co-Authored-By: YiYi Xu <yiyi@huggingface.co> * Use split q/k/v projections in Ideogram4 attention Replace the fused `qkv`/`o` linears with canonical `to_q`/`to_k`/`to_v`/`to_out` projections, matching the standard diffusers attention layout and the split checkpoint format. Mathematically equivalent to the fused form (q/k/v are contiguous row-slices of the fused weight). Drops the now-inapplicable fuse/unfuse overrides. --------- Co-authored-by: Jin <jin.li@ideogram.ai> Co-authored-by: YiYi Xu <yiyi@huggingface.co> |
||
Lancer
|
b003a47354 |
[Feat] support AutoPipelineForText2Audio (#13511)
* [Feat] support AutoPipelineForText2Audio Signed-off-by: Lancer <maruixiang6688@gmail.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * Apply style fixes --------- Signed-off-by: Lancer <maruixiang6688@gmail.com> Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> |
||
rmatif
|
c943837899 |
Add Anima modular pipeline (#13732)
* Add Anima pipeline * Fix empty Anima negative prompts * Fix Anima registration * Clean up Anima conditioner * Refactor Anima to modular * Use modular loader in Anima docs * Move Anima text conditioner * Apply style fixes * Address Anima review nits * Fix Anima rotary autocast * Clean up Anima LoRA checks * Update Anima model id * Document Anima LoRA loader * Document Anima conditioner forward --------- Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: YiYi Xu <yixu310@gmail.com> |
||
Vinh H. Pham
|
32ecbe3836 |
Fix redundant Z-Image terminal timestep (#13730)
* Fix redundant Z-Image terminal timestep * Update src/diffusers/modular_pipelines/z_image/before_denoise.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * unify sigmas calculation * update z-image turbo examples --------- Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> |
||
Guian Fang
|
d94ad81b87 |
[AnyFlow] FAR: standalone causal-mask builder + torch.compile follow-up (#13792)
* [AnyFlow] FAR: standalone causal-mask builder + torch.compile follow-up
Follow-up to #13745. Extracts FAR mask construction to a module-level
helper and adds an `attention_mask` forward kwarg so
AnyFlowFARTransformer3DModel can be wrapped in
`torch.compile(fullgraph=True)`. The pipeline pre-builds the mask during
KV-cache prefill so users get end-to-end fullgraph compile.
* Public method `AnyFlowFARTransformer3DModel.build_attention_mask(...)`
(modes: "train", "cache") plus private module-level helper
`_build_anyflow_far_causal_block_mask(...)`.
* `_build_freqs` cache lookup/write bypassed under
`torch.compiler.is_compiling()` to avoid a Dynamo guard recompile on
the second compiled call (applied in bidi source; synced to FAR via
`# Copied from`).
* `TestAnyFlowFARTransformer3DCompile(TorchCompileTesterMixin)` —
recompilation_and_graph_break, repeated_blocks, and group_offloading
pass on H200; AOT is `@pytest.mark.skip`'d (torch.export rejects
BlockMask as a pytree input).
* Base `get_dummy_inputs` omits `attention_mask` so every non-compile
test class exercises the in-forward fallback; the compile class
overrides to inject a pre-built mask.
* Bit-exact: pre-built path vs internal-build fallback max|Δ|=0.0e+00.
* [AnyFlow] docs: full author list, repo demo examples, slimmer pipeline page
* Full author list and NVIDIA → NUS → MIT institution order; TL;DR +
abstract + Available Models bullets.
* Rewritten pipeline-selection tip describing both pipelines symmetrically.
* T2V / I2V / V2V examples now use the canonical 81-frame setup and the
demo prompts / conditioning assets shipped under
`NVlabs/AnyFlow/assets/evaluation/` (linked via raw.githubusercontent.com).
* Drop the inline "Optimizing Memory" and "torch.compile" sections — those
notes will live in the NVlabs/AnyFlow repo's own performance guide rather
than the diffusers pipeline reference.
* Sync zh user guide and the two model-API stubs.
* [AnyFlow] FAR: move chunk_partition default into transformer config
- AnyFlowFARTransformer3DModel.__init__ now accepts chunk_partition via
@register_to_config (default (1, 3, 3, 3, 3, 3, 3, 2) for the released
81-frame checkpoints, matching the field on Hub).
- AnyFlowFARPipeline.__call__ no longer requires chunk_partition; defaults
to self.transformer.config.chunk_partition. Per-call override still
supported for V2V / non-default num_frames.
- Drop the AnyFlowFARPipeline.default_chunk_partition class attribute.
- Update docs (en pipelines/models, zh using-diffusers) and the conversion
script to match.
* [AnyFlow] FAR pipeline: fix `timesteps` shadowing across chunks
Inside the per-chunk rollout loop, the local variable `timesteps` was
reassigned to `self.scheduler.timesteps` after `set_timesteps()`. On the
next chunk iteration the same name was passed back into
`set_timesteps(timesteps=...)`, where a non-None value enters the
*custom-schedule* branch — `apply_shift` re-runs on already-shifted
values, double-shifting the schedule for every chunk after the first.
Concretely, with `shift=5` and `num_inference_steps=4`:
- chunk 0 timesteps: [1000, 937.5, 833.3, 625] (correct)
- chunk 1+ timesteps: [1000, 986.8, 961.3, 892.9] (double-shifted)
The later steps drift toward `t=1000` instead of toward `t=0`, the
flow-map model is conditioned on the wrong source sigma, and the chunk
KV cache accumulates errors that show up as artifacts in later video
frames.
Fix: rebind the cached schedule to a fresh local name
(`scheduler_timesteps`) so the outer-scope `timesteps` kwarg (the
user-provided custom schedule, when any) stays untouched across chunks.
Layer-by-layer verification against the NVlabs reference implementation
on H200 (elephant prompt, seed 0, 4 NFE, 81 frames):
- chunk 0 inference: bit-exact (0.0 mean diff)
- chunk 1 step 0: 0.194 → 0.014 (-93%)
- chunk 7 last step: 0.564 → 0.274 (-51%)
* [AnyFlow] FAR: doc-builder line wrap for chunk_partition docstrings
Pure rewrap to satisfy `doc-builder style --max_len 119`. Two docstrings
introduced in
|
||
Atharva Joshi
|
a1c7df4801 |
Adding Cosmos 3 to Diffusers (#13818)
* Adding Cosmos 3 * removed dead code * Change customer TimeEmbedding Layer to DIffusers Time Embedding * removed dependency on hugging face transformers * refactor 1 * Fixed Attention Pattern * Removed from Pretrain overrides * Removing normalization from the audio Tokenizer * fixed diffusers checkpoint * fixed video save uint conversion * added forward hook for cpu offload case * removed dead params for sound tokenizer * renaming audio encoder for readability * ruff format * Fix checkpoint conversion script for sound tokenizer * Audio Decoder trim and removing some dead code * removing dead sequence packing code * refactor pipeline to diffusers style formatting * removing use of cosmos3 audio encoder * Revert "removing use of cosmos3 audio encoder" This reverts commit 1b8b99a95e5acc749b5c100637ed382de138c606. * refactor audio encoder * inline remaining sequence packing functions and lint * Removed GenerationDataClean class and Action logic * inlined default args * removed dead code and refactoring * drop pipeline-helper @no_grad, inline derive helper, move guidance check to check_inputs Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * drop unused list bookkeeping from PackedSequence attn_modes was never read; sample_lens collapses into the existing sequence_length int (we only pack a single sample at a time); split_lens collapses into a single und_len int (only split_lens[0] was ever read). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * extract pack-time build state from PackedSequence dataclass curr, _use_mrope, _mrope_temporal_offset, _mrope_reset_spatial were transient counters used while building the joint sequence, not part of the finalized output. Thread them through _pack_*_tokens as positional args/returns so the dataclass only carries fields the pipeline actually reads back. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * drop private-API isinstance/shape asserts These were build-mode guards on PackedSequence internals and shape checks on tensors the pipeline itself constructs, both flagged as noise in private code per reviewer. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * build PackedSequence tensors on target device, drop to_cuda Thread device through pack_input_sequence and _pack_*_tokens helpers so all torch.tensor/zeros/arange calls land on the target device directly. Move CPU-side mRoPE tensors over with .to(device) at the append site. Pass device to finalize so list-to-tensor conversion lands on device too. Delete PackedSequence.to_cuda() and the helper _modality_to_cuda; drop the corresponding call sites in __call__. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * drop SequencePlan, skip_text_tokens, and bos_token_id branch SequencePlan's has_text and has_vision were True at every construction site and has_sound was derivable from x0_tokens_sound is not None. condition_frame_indexes_vision is now passed directly as a List[int] arg to pack_input_sequence. Removed the skip_text_tokens flag (never True) and the dead bos_token_id shift branch in _pack_text_tokens (special_tokens never carries one in this pipeline). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * use retrieve_latents helper in _encode_video Copied from stable_diffusion_img2img matching cosmos2_5 convention so make fix-copies keeps it synced. Functionally identical to the prior .latent_dist.mode() call but handles latent_dist/latents attribute variants uniformly. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * drop get_data_and_condition + data_batch dict scaffolding normalize_video_databatch_inplace, augment_image_dim_inplace and remove_padding_from_latent were no-ops once is_preprocessed=True (always set by this pipeline) and the pipeline never pads. get_data_and_condition just orchestrated those plus a never-taken multi-vision branch. Replaced the whole chain with a few lines inline in prepare_latents: build vision_tensor on device, call _encode_video, set fps_vision. prepare_latents no longer needs input_caption_key, input_video_key, input_image_key, or the prompt kwarg. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * drop _load_image_as_tensor; use VideoProcessor for conditioning frame Image loading is the caller's responsibility (load_image from diffusers.utils), matching the cosmos2_5 example. The pipeline registers a VideoProcessor in __init__ and calls preprocess() to resize + normalize caller-supplied PIL / np / tensor inputs to [1, 3, H, W] in [-1, 1]. prepare_latents fills the temporal dim in two lines (single frame at t=0, repeat-pad the rest) preserving the prior i2v behavior. Inference script updated to call load_image() before passing to the pipeline. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * move encode/decode helpers + transformer forward into Cosmos3OmniTransformer Pull encode_text / encode_vision / encode_sound_tokens / decode_vision / decode_sound_tokens (and their pure-tensor helpers patchify_and_pack_latents, unpatchify_and_unpack_latents, apply_timestep_embeds_to_noisy_tokens, _pack_sound_latents, _unpack_sound_latents) from the pipeline into Cosmos3OmniTransformer as methods. The transformer's forward(packed_seq) now runs the full per-step pass: encode text/vision/sound, rotary + layer loop, decode vision/sound — returns (preds_vision, preds_sound). The pipeline's CFG loop drops the encode_*/decode_* method calls and the manual und/gen split/concat; each pass is now a single self.transformer(packed_seq) call. No self.transformer.{embed_tokens, vae2llm, llm2vae, sound2llm, llm2sound, time_embedder, time_proj, sound_modality_embed} access remains in the pipeline. Cosmos3VLTextModel is kept as a structural wrapper for now — flattening it would break the published checkpoint layout. Tracked separately. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * remove Cosmos3VLTextModel; flatten transformer layout embed_tokens / layers / norm / norm_moe_gen / rotary_emb are now direct attributes of Cosmos3OmniTransformer. The converter strips the leading `model.` prefix from the source language_model state-dict so new conversions land at the flat layout natively. Published Hub artifact (nvidia/Cosmos3-Nano) needs its transformer safetensors + index.json re-keyed with the same prefix strip before this code can load it. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * move save_img_or_video/save_wav to cosmos/export_utils.py Per reviewer guidance, custom video/audio export helpers belong in a pipeline-local export_utils.py (mirroring pipelines/ltx2/export_utils.py) rather than living inside the pipeline file. Pipeline imports trim the now-unused pathlib/numpy/export_to_video; inference example updated to import from the new location. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * drop @torch.no_grad on pipeline __call__ Diffusers pipeline convention: __call__ does not wear a torch.no_grad decorator; the responsibility for grad context sits with the caller. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * follow standard transformer conventions in Cosmos3OmniTransformer - Declare _no_split_modules, _repeated_blocks, _skip_layerwise_casting_patterns, _keep_in_fp32_modules, _supports_gradient_checkpointing on the transformer. - Wire self.gradient_checkpointing + the _gradient_checkpointing_func branch in forward so the flag is honest (models.md gotcha #3). - Add PeftAdapterMixin and AttentionMixin to the mixin set so LoRA loading and the attention-backend setters work. - CosmosAttnProcessor3_0 now declares _attention_backend / _parallel_config and forwards them to dispatch_attention_fn, matching the pattern in models.md and transformer_wan.py. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * restore @torch.no_grad() on Cosmos3OmniDiffusersPipeline.__call__ Reverting bea4eecf9 — removing the decorator causes GPU OOM during inference because the autograd graph accumulates across the full denoising loop (35 steps × dual cond/uncond passes × full transformer). pipelines.md gotcha #2 documents this exact failure mode and the convention is upheld by every sibling pipeline (pipeline_flux.py:652, pipeline_qwenimage.py:462, pipeline_wan.py:381, pipeline_ltx.py:535). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * drop training-only dummy projection paths in transformer decode helpers _decode_vision and _decode_sound both guarded a "no noisy tokens" branch that ran zeroed projections to keep the autograd graph intact. Those branches only fire when a pure-conditioning step has no MSE-loss tokens, which never happens in the inference pipeline — every workflow has at least one noisy vision frame, and _decode_sound is gated on has_sound which itself requires noisy sound tokens. Deleting per CLAUDE.md: "delete training-time code paths… only keep the inference path." Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * inline modality encode/decode helpers into transformer forward models.md "Coding style": all layer calls should be visible directly in forward — avoid helper functions that hide nn.Module calls. Inlines _encode_text / _encode_vision / _encode_sound / _decode_vision / _decode_sound into forward so embed_tokens, vae2llm, sound2llm, llm2vae, llm2sound, and time_embedder are all visible at the call site. Pure- tensor helpers (_patchify_and_pack_latents, _unpatchify_and_unpack_latents, _pack_sound_latents, _unpack_sound_latents, _apply_timestep_embeds_to_noisy_tokens) stay as methods since they don't hide layer state. Also drops the inference-unreachable guards while collapsing the helpers: the "vision is None" / "sound is None" / "mse_loss_indexes.numel() > 0" branches never fire because the pipeline always packs vision, only routes to the sound branch when sound is present, and condition_frame_indexes never covers the entire stream. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * trim dead flags and idioms in Cosmos3 pipeline __call__ - Drop assert config.use_moe — use_moe is never read by the model and asserts vanish under python -O; if !use_moe is unsupported the place to surface it is check_inputs, not a stripped assert. - Delete the joint_attn_implementation == "flex" path entirely (the include_end_of_generation_token branch in pack_input_sequence, the include_eog hoist, and both call-site kwargs). The published config is "two_way"; the flex branch and the end_of_generation special token were dead under every shipped checkpoint. - Drop torch._inductor.cudagraph_mark_step_begin() from the step loop — cudagraph stepping belongs in the caller's torch.compile wrapper, not the inference pipeline. - Replace four int(torch.prod(torch.tensor(shape))) idioms with math.prod(shape) — no tensor allocation, no .item() sync, and math is already imported. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * introduce Cosmos3Condition for prepare_latents return shape Replace the 7-tuple return from prepare_latents with a Cosmos3Condition dataclass that carries the encoded conditioning latents (vision + optional sound), their fps tensors, the conditioning frame indices, and the num_vision_items count. The denoising loop and _postprocess_latents now read these as named attributes instead of positional tuple unpacking. Addresses reviewer thread huggingface/diffusers-new-model-addition-cosmos#1 comment 3278569263 ("create something like Cosmos3Condition class for condition input"). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * restructure pack helpers as data-returning pipeline methods Replace the four module-level pack functions (_pack_text_tokens, _pack_vision_tokens, _pack_sound_tokens, pack_input_sequence) with methods on Cosmos3OmniDiffusersPipeline. The three per-segment methods now build and return their own data (text_ids/mrope_ids tuple for text, a populated ModalityData for vision/sound) instead of mutating a shared PackedSequence builder; pack_input_sequence orchestrates them. Other cleanups along the way: - Drop dead branches that the published config never exercises: use_mrope=False (model is always unified_3d_mrope), has_generation=False (always True), multi-vision-items (num_vision_items always 1), and the curr_rope_id non-mrope path. - Move the bf16 cast of per-step noisy tokens to before pack_input_sequence so the build-then-mutate pattern on packed_seq.vision.tokens disappears. - Drop the latent_patch_size / config hoists in __call__ that are now read directly inside the pack methods. Addresses reviewer threads huggingface/diffusers-new-model-addition-cosmos#1 comments 3278871807, 3278908766, 3278918514. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * collapse builder-pattern PackedSequence/ModalityData into flat dataclasses ModalityData and PackedSequence carried a list-or-tensor union for every field so they could double as builders during packing, with finalize() converting the lists to tensors at the end. Now that the pack methods each build their segment in one shot, finalize() is just a list->tensor conversion the pack methods can do themselves. - Rename ModalityData to _ModalityData (internal) with all-tensor fields (lists only for per-item entries like tokens / condition_mask). - Rename PackedSequence to Cosmos3PackedSequence and drop its finalize() method; fields are direct tensors at construction time. - _pack_vision_tokens / _pack_sound_tokens now build finalized _ModalityData directly via torch.arange / torch.tensor / torch.full. - pack_input_sequence builds the final Cosmos3PackedSequence in one return statement; no more two-stage build-then-finalize. - Update the transformer's forward docstring to reference the new name. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * exclude dtype from saved transformer config via ignore_for_config ModelMixin.from_pretrained injects dtype into init_dict (configuration_utils.py:289) whenever it appears in the loader's unused_kwargs, so Cosmos3OmniTransformer.__init__ has to accept dtype. But the default @register_to_config decorator was also serializing it into config.json on every save — leaving a stray "dtype": "bfloat16" key that doesn't describe the architecture, just the load-time runtime preference. Adding ignore_for_config = ["dtype"] keeps the decorator from registering dtype while still accepting it on the init signature. New saves omit dtype; existing checkpoints that have it log a warning at load time but the value is re-injected and __init__ ignores it, so loading is unchanged. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * add Cosmos3 pipeline and transformer docs pages; retire JSON example inputs - New docs/source/en/api/pipelines/cosmos3.md with copy-pasteable text-to-image, text-to-video, image-to-video, and text-to-video-with-sound example snippets. The docs page is now the canonical reference for application code instead of the JSON-driven inputs/ directory. - New docs/source/en/api/models/cosmos3_omni_transformer.md describing the MoT dual-pathway architecture and showing a from_pretrained snippet. - Wire both pages into docs/source/en/_toctree.yml. - Export Cosmos3OmniDiffusersPipeline from the top-level diffusers package (matching every sibling pipeline) and add the corresponding dummy class for torch/transformers-unavailable environments. - Delete examples/cosmos3/inputs/omni/{t2i,t2v,i2v}.json and rewrite inference_cosmos3.py to take --prompt / --vision-path / --num-frames directly as CLI args. The script stays as a development smoke-test runner; canonical usage now lives in the docs. - Refresh examples/cosmos3/README.md to point at the docs page and reflect the new CLI surface. Addresses reviewer thread huggingface/diffusers-new-model-addition-cosmos#1 comment 3278258567. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * restore full prompts in cosmos3 docs from original example JSONs The text-to-video and image-to-video example prompts were shortened when porting from examples/cosmos3/inputs/omni/*.json into the docs page. Restore them verbatim from the JSONs so the docs reflect the prompts the model was actually demonstrated against, and so users copying from the docs get the same conditioning the example was tuned for. Also align the text-to-video-with-sound example: it now reuses the exact same prompt as the text-to-video block with only enable_sound=True added, instead of a hand-written waterfall prompt. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * Renamed Cosmos3 module attributes Signed-off-by: Maciej Bala <mbala@nvidia.com> * bugfix Signed-off-by: Maciej Bala <mbala@nvidia.com> * Removed unnecessary helper function; added extra comments Signed-off-by: Maciej Bala <mbala@nvidia.com> * Moved from encode_prompt to tokenize_prompt Signed-off-by: Maciej Bala <mbala@nvidia.com> * Bring back video system prompt Signed-off-by: Maciej Bala <mbala@nvidia.com> * Remove multi frame conditioning for now Signed-off-by: Maciej Bala <mbala@nvidia.com> * Removed default negative prompts from code Signed-off-by: Maciej Bala <mbala@nvidia.com> * Remove Cosmos3condition; simplify sequence pack Signed-off-by: Maciej Bala <mbala@nvidia.com> * Clean up multiple parameters Signed-off-by: Maciej Bala <mbala@nvidia.com> * Simplify decode video; remove remainings of batching Signed-off-by: Maciej Bala <mbala@nvidia.com> * simple renames Signed-off-by: Maciej Bala <mbala@nvidia.com> * Refactored schedulers for sound Signed-off-by: Maciej Bala <mbala@nvidia.com> * Remove unnecessary autocast Signed-off-by: Maciej Bala <mbala@nvidia.com> * Update sound example Signed-off-by: Maciej Bala <mbala@nvidia.com> * Simplify loops in transformer_cosmos3.py Signed-off-by: Maciej Bala <mbala@nvidia.com> * Remove unused config attributes Signed-off-by: Maciej Bala <mbala@nvidia.com> * Cleanup audio decoder Signed-off-by: Maciej Bala <mbala@nvidia.com> * Reuse encoder_video from LTX2 for Cosmos3 Signed-off-by: Maciej Bala <mbala@nvidia.com> * Fixed a few nits Signed-off-by: Maciej Bala <mbala@nvidia.com> * Moved to RMSNorm for Cosmos3 Signed-off-by: Maciej Bala <mbala@nvidia.com> * Remove meta_tensor usage Signed-off-by: Maciej Bala <mbala@nvidia.com> * Improved rope handling Signed-off-by: Maciej Bala <mbala@nvidia.com> * Improved prompt templates Signed-off-by: Maciej Bala <mbala@nvidia.com> * Added extra docs for templatete Signed-off-by: Maciej Bala <mbala@nvidia.com> * remove dataclasses * Cleanup after merging Signed-off-by: Maciej Bala <mbala@nvidia.com> * Added guardrails Signed-off-by: Maciej Bala <mbala@nvidia.com> * bugfixed guardrails Signed-off-by: Maciej Bala <mbala@nvidia.com> * Bugfix guardrails v2 Signed-off-by: Maciej Bala <mbala@nvidia.com> * Simplified input_timestep Signed-off-by: Maciej Bala <mbala@nvidia.com> * Add TODO Co-authored-by: YiYi Xu <yixu310@gmail.com> * simplify conditional mask generation * Inlined _postprocess_latents Signed-off-by: Maciej Bala <mbala@nvidia.com> * removed pack_input_sequence helper * restore export utils with deprecation warning * moved sampling rate to pipeline attribute * inlined sound and image condition mask * seperating static and timestep based sound and vision token packing * unpack transformer args * enabled selection of cosmos3 super * Update src/diffusers/pipelines/cosmos/pipeline_cosmos3_omni.py Co-authored-by: YiYi Xu <yixu310@gmail.com> * Update src/diffusers/pipelines/cosmos/pipeline_cosmos3_omni.py Co-authored-by: YiYi Xu <yixu310@gmail.com> * Apply suggestions from code review Co-authored-by: YiYi Xu <yixu310@gmail.com> * fixed sound and vision conditioning use from prepare latents * ruff format and doc builder * ran fix copies * move typing to python3.10 * move special token application from pack_text_tokens to tokenize_prompt * rename packing methods to process methods * remove guidance_scale check * fix nits * respect vae dtype in the pipeline * use vae dtype for vae normalization stats * skip CFG if guidance_scale is 1 * Remove unnecessary parameter Signed-off-by: Maciej Bala <mbala@nvidia.com> * fix CFG for sound * bugfix for sound CFG * ruff format * Fix apply_chat_template return dict arg to return BatchEncoding * added option to select attention processor * docs: refresh Cosmos 3 pipeline intro Replace the terse architectural lede with the launch-style positioning (unified WFM for Physical AI, consolidating Predict/Reason/Transfer into one omni-model) and split out "What's new" and "Available checkpoints" sections so the page leads with capability rather than repo IDs. * docs: document Cosmos3OmniPipeline.__call__ arguments Add the missing Args block to Cosmos3OmniPipeline.__call__ so utils/check_forward_call_docstrings.py passes — covers all 21 parameters from prompt through enable_safety_check, plus a Returns section describing the output dataclass. * style: doc-builder reflow on Cosmos3OmniPipeline.__call__ docstring --------- Signed-off-by: Maciej Bala <mbala@nvidia.com> Co-authored-by: Yuliya Zhautouskaya <yzhautouskay@nvidia.com> Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Co-authored-by: Maciej Bala <mbala@nvidia.com> Co-authored-by: Dima Zhylko <dzhylko@nvidia.com> Co-authored-by: YiYi Xu <yixu310@gmail.com> |
||
|
Guian Fang
|
e39aecff57 |
Add AnyFlow Any-Step Video Diffusion Pipelines (Bidirectional + FAR Causal) (#13745)
* [Pipelines] AnyFlow: scaffold pipelines/anyflow + register all top-level imports
This is the lazy-loader scaffolding only. Body files (pipeline_anyflow.py,
pipeline_anyflow_causal.py, transformer_anyflow.py,
scheduling_flow_map_euler_discrete.py) come in subsequent commits.
* [Schedulers] AnyFlow: add FlowMapEulerDiscreteScheduler
The flow-map scheduler advances samples from timestep t to caller-provided
target r in a single Euler step, supporting any-step sampling on flow-map-
distilled checkpoints. It is a general-purpose scheduler — not specific to the
AnyFlow checkpoints.
Tests: 12 standalone tests covering instantiation, set_timesteps endpoints,
shift identity/monotonicity, step shape preservation, zero-interval identity,
one-shot sampling, train weight schemes, scale_noise endpoints.
Docs: api/schedulers/flow_map_euler_discrete.md
* [Models] AnyFlow: add AnyFlowTransformer3DModel
A 3D DiT extending the v0.35.1 Wan2.1 backbone with two config-toggled modules:
* FAR causal blocks (init_far_model=True): block-sparse causal attention via
flex_attention + compressed-frame patch embedding for frame-level
autoregressive generation (Gu et al., 2025, arXiv:2503.19325).
* Dual-timestep flow-map embedding (init_flowmap_model=True): adds a delta
timestep embedder enabling flow-map sampling z_t -> z_r over arbitrary
intervals (AnyFlow).
With both flags off, the model reduces to stock Wan2.1.
The class is intentionally self-contained rather than annotated with
'# Copied from diffusers.models.transformers.transformer_wan' because upstream
Wan has been refactored extensively since v0.35.1 (new WanAttention class,
different processor architecture).
Tests: 9 unit tests covering construction in 3 modes, bidi forward shape and
determinism, return_dict variants, save/load round-trip with and without
init_far_model, gradient checkpointing toggle.
Docs: api/models/anyflow_transformer3d.md
* [Pipelines] AnyFlow: add AnyFlowPipeline and AnyFlowCausalPipeline
* AnyFlowPipeline (pipeline_anyflow.py, ~590 LOC): bidirectional T2V using
flow-map sampling. Loads checkpoints from nvidia/AnyFlow-Wan2.1-T2V-{1.3B,14B}.
* AnyFlowCausalPipeline (pipeline_anyflow_causal.py, ~700 LOC): FAR-based
causal pipeline supporting T2V/I2V/TV2V via task_type kwarg. Loads checkpoints
from nvidia/AnyFlow-FAR-Wan2.1-{1.3B,14B}-Diffusers.
Both pipelines reuse stock WanLoraLoaderMixin, AutoencoderKLWan, UMT5EncoderModel,
and AutoTokenizer from upstream. The transformer is the AnyFlowTransformer3DModel
introduced in the previous commit. The scheduler is FlowMapEulerDiscreteScheduler.
Tests:
* tests/pipelines/anyflow/test_anyflow.py: PipelineTesterMixin fast tests +
slow integration test against nvidia/AnyFlow-Wan2.1-T2V-1.3B-Diffusers.
* tests/pipelines/anyflow/test_anyflow_causal.py: same structure for FAR variant.
Reference slices for slow integration tests are deferred to Phase 7
(Final quality pass) where the user runs them on a real GPU.
* [Docs] AnyFlow: add main pipeline documentation page
Modeled on the Helios pipeline doc (PR #13208). Sections: paper link + abstract,
supported checkpoints table, memory/speed optimization tabs, T2V/I2V/TV2V
examples for both bidirectional and causal variants, autodoc trailers.
* [Auto/Scripts] AnyFlow: register AutoPipelineForText2Video + add conversion script
* Register AnyFlowPipeline in AUTO_TEXT2VIDEO_PIPELINES_MAPPING.
* AnyFlowCausalPipeline is intentionally NOT registered for AutoPipeline because
its task switch (t2v / i2v / tv2v) is too rich for a single auto-resolve key.
* scripts/convert_anyflow_to_diffusers.py: convert .pt training checkpoints
(with 'ema' state dict) into a diffusers save_pretrained layout. Supports all
4 released NVIDIA AnyFlow variants. Replaces the omegaconf-based config in the
upstream repo with argparse to match other diffusers conversion scripts.
* [Quality] AnyFlow: ruff-format + regenerated dummy stubs
* ruff format pass on all 5 source files (long lines + trailing comma fixes)
* check_dummies.py --fix_and_overwrite regenerated:
- dummy_pt_objects.py: AnyFlowTransformer3DModel + FlowMapEulerDiscreteScheduler
- dummy_torch_and_transformers_objects.py: AnyFlowPipeline + AnyFlowCausalPipeline
Local fast tests: 21/21 passed
- 12 scheduler tests (FlowMapEulerDiscreteScheduler)
- 9 transformer tests (AnyFlowTransformer3DModel construction + bidi forward + save/load)
The pipeline fast tests in tests/pipelines/anyflow/ require a local dev install
that matches the diffusers main branch's transformers >= compatibility floor.
The reference slices for slow integration tests (real GPU + 1.3B/14B
checkpoints) are intentionally left as TODO stubs to be captured by the user
on a real GPU machine before opening the PR.
* [AnyFlow] address review feedback: bug fixes + DMD wording + EN/ZH tutorials
Critical bug fixes (verified against precision-validation review):
* pipeline_anyflow.py / pipeline_anyflow_causal.py: replace hardcoded
transformer_dtype = torch.bfloat16 with self.transformer.dtype, so
pipe.to("cpu") and PipelineTesterMixin save/load tests do not crash on a
dtype mismatch in the patch_embedding conv3d.
* transformer_anyflow.py: drop the duplicate `base = base = ...` assignment in
_build_causal_mask (was a copy-paste typo carried over from FAR-Dev).
* transformer_anyflow.py: drop unused `q_is_context` / `k_is_context` locals
and the `# noqa: F841` markers that were silencing the dead-store warning.
* transformer_anyflow.py: remove `CacheMixin` from the inheritance list — the
pipeline manages KV cache directly, the mixin's interface is unused.
* transformer_anyflow.py: guard the module-level `torch.compile(flex_attention)`
with try/except so the file imports cleanly on CPU CI / no-Triton machines.
* convert_anyflow_to_diffusers.py: replace ad-hoc print warnings with the
stdlib logger (warning_once-style) and a module-level basicConfig.
Documentation accuracy:
* AnyFlowCausalPipeline class docstring + main pipeline doc + EN/ZH tutorial:
drop the fictitious `task_type` / `image` / `video` arguments and document
the real API: pass `context_sequence={"raw": tensor}` (or `{"latent": ...}`)
to switch between T2V (None) / I2V (1-frame) / TV2V (4n+1-frame) modes.
* Pipeline class docstrings + main doc: explicitly describe AnyFlow's
two-stage LoRA distillation including DMD reverse-divergence supervision
with Flow-Map backward simulation in stage 2 (was previously implicit).
* training_rollout: add detailed docstring explaining its role as the
3-segment Flow-Map backward simulation entry point used during DMD training.
* Long-form tutorial doc `using-diffusers/anyflow.md` (EN, 239 LOC) and
Chinese mirror `docs/source/zh/using-diffusers/anyflow.md` (224 LOC) added
and registered in both `_toctree.yml` files.
Tests:
* Skip `test_attention_slicing_forward_pass` in both pipeline test classes
with a clear rationale (custom attention processor does not support slicing).
* All 21 standalone tests still pass (12 scheduler + 9 transformer).
Quality gates:
* `ruff check` clean across all AnyFlow files.
* `ruff format --check` reports 6 files already formatted.
* `python utils/check_copies.py` reports no diff.
Out of scope for this commit (deferred until reviewer feedback):
* Splitting AnyFlowTransformer3DModel into bidi + causal subclasses
* Unifying _forward_inference / _forward_cache return types
* Migrating model tests from plain unittest to BaseModelTesterConfig + mixins
* HF model card / config.json metadata updates on the nvidia/* repos
(push to Hub manually before opening the PR)
* [AnyFlow] rename Causal->FAR + explicit forward signature + dataclass output
Round 2 of review feedback. Three groups of changes; transformer state-dict
keys, module hierarchy, and tensor flow are unchanged so the H200 bit-exact
validation remains valid.
A. Pipeline rename (mechanical, no behavior change):
* Class: AnyFlowCausalPipeline -> AnyFlowFARPipeline (Causal in diffusers
usually means an attention mask; AnyFlow's variant is FAR autoregressive,
so the FAR name is more specific and matches the paper).
* File: pipeline_anyflow_causal.py -> pipeline_anyflow_far.py (git mv).
* Test file: test_anyflow_causal.py -> test_anyflow_far.py (git mv).
* All references updated in src/, tests/, docs/, scripts/, plus stale
anyflowcausalpipeline anchor links in tutorial markdown.
B. Pipeline test bug fixes (closes 19 fast-test failures reported by
precision-validation reviewer):
* pipeline_anyflow.py / pipeline_anyflow_far.py: __call__ now sets
self._num_timesteps = num_inference_steps before the rollout, so the
PipelineTesterMixin callback tests can read pipe.num_timesteps.
* tests/pipelines/anyflow/test_anyflow_far.py: drop the fictitious
task_type="t2v" kwarg that crashed every causal fast test (the FAR
pipeline selects mode via context_sequence, not a task_type arg).
C. Transformer architecture cleanups (review-driven, no tensor changes):
* Replace forward(*args, **kwargs) dispatcher with an explicit signature
listing every supported kwarg (hidden_states, timestep, r_timestep,
encoder_hidden_states, encoder_hidden_states_image, chunk_partition,
clean_hidden_states, clean_timestep, kv_cache, kv_cache_flag, is_causal,
attention_kwargs, return_dict). Helps IDE / type-checker / torch.compile
tracing.
* Drop SimpleNamespace returns. Add AnyFlowFARTransformerOutput
(BaseOutput dataclass with sample + kv_cache fields) for the two causal
paths that need to also propagate kv_cache (_forward_inference and the
newly return_dict-aware _forward_cache). _forward_train and
_forward_bidirection now consistently return Transformer2DModelOutput.
Pipeline call sites already use return_dict=False with positional
unpacking, so the fix is transparent there.
Out of scope (deferred until canonical-org HF metadata sync):
* Splitting AnyFlowTransformer3DModel into a bidi class plus an
AnyFlowFARTransformer3DModel subclass — touches register_to_config keys
and would require updating model_index.json on every released checkpoint.
* Promoting chunk_partition from register_to_config to a forward-time
argument (same reason).
* Renaming training_rollout to _denoise — would break callers in the
FAR-Dev on-policy trainer that produced the released checkpoints.
Local fast tests: 21/21 still pass (12 scheduler + 9 transformer).
ruff check, ruff format, and check_copies.py are all clean.
* [AnyFlow] wire callback_on_step_end through inference_range + add chunk_partition to FAR fast-test fixture
Two root causes for the 19 remaining PipelineTesterMixin failures, identified
by the H200 reviewer:
1. callback_on_step_end was accepted by __call__ but never invoked. Both
pipelines pass it through to training_rollout (and FAR additionally through
inference()), and inference_range now fires it after scheduler.step in
the standard inference branch:
if callback_on_step_end is not None:
callback_kwargs = {k: locals()[k] for k in callback_on_step_end_tensor_inputs}
callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)
latents = callback_outputs.pop("latents", latents)
prompt_embeds = ...
negative_prompt_embeds = ...
`nonlocal prompt_embeds, negative_prompt_embeds` lets the callback rewrite
the closure-captured embeddings, matching upstream WanPipeline semantics.
The 3-segment grad_timestep training rollout does not invoke the callback;
it is intentionally training-only.
2. tests/pipelines/anyflow/test_anyflow_far.py::get_dummy_components built
the dummy transformer without a `chunk_partition`, leaving it None on the
model config and crashing the pipeline at `sum(self.transformer.config.chunk_partition)`.
Set `chunk_partition=[1, 1, 1]` in the fixture (3 chunks of 1 latent frame
each, matching the test's num_frames=9 -> 3 latent frames).
Local fast tests: 21/21 still pass.
ruff check, ruff format, and check_copies.py are all clean.
* [AnyFlow] Phase 2: split transformer + drop chunk_partition from config + rename helpers
Major architectural refactor that aligns the integration with diffusers conventions
ahead of the canonical-org Hub upload. State-dict keys, module hierarchy, and
tensor flow are unchanged so the H200 bit-exact validation remains valid; only
the on-disk transformer/config.json fields move.
Changes:
1. **Sibling transformer classes** replace the flag-driven single class:
* AnyFlowTransformer3DModel — bidirectional only. Drops compressed_patch_size /
full_chunk_limit / init_far_model / init_flowmap_model / chunk_partition
kwargs (always-on for AnyFlow distilled checkpoints).
* AnyFlowFARTransformer3DModel — adds far_patch_embedding + the 3 FAR forward
paths (train / cache-prefill / autoregressive inference).
* AnyFlowTimeTextImageEmbedding (the legacy single-time embedder used only by
the old setup_flowmap_model bootstrap) is removed; both classes now build
AnyFlowDualTimestepTextImageEmbedding directly in __init__.
* setup_flowmap_model / setup_far_model methods are removed; weight warm-start
for far_patch_embedding (trilinear interpolation from patch_embedding) moves
into AnyFlowFARTransformer3DModel.__init__.
2. **chunk_partition** is no longer a model config field. The FAR pipeline owns
the schedule:
* AnyFlowFARPipeline.default_chunk_partition = [1, 3, 3, 3, 3, 3, 3, 2]
matches the released 81-frame NVIDIA checkpoints.
* AnyFlowFARPipeline.__call__ / _denoise_rollout accept a chunk_partition
argument that overrides the default for non-default num_frames.
3. **training_rollout -> _denoise_rollout** rename across both pipelines and all
English / Chinese docs that referenced it. Signals the method is internal to
the pipeline driver, not a public training API.
4. **Conversion script + tests + docs + registries**:
* scripts/convert_anyflow_to_diffusers.py: VARIANTS dict picks the right
transformer class per variant; init_far_model / init_flowmap_model /
chunk_partition kwargs are removed from the from_pretrained call.
* Transformer test file split into AnyFlowTransformer3DModelTest and
AnyFlowFARTransformer3DModelTest classes.
* Pipeline test fixtures use the right class and pass chunk_partition via
get_dummy_inputs (3-frame schedule [1, 1, 1] for the 9-frame test).
* New docs page docs/source/en/api/models/anyflow_far_transformer3d.md;
anyflow_transformer3d.md rewritten for the bidi-only class.
* AnyFlowFARTransformer3DModel registered in src/diffusers/__init__.py,
src/diffusers/models/__init__.py, models/transformers/__init__.py and the
dummy_pt_objects.py stubs.
* docs/source/en/_toctree.yml: new entry for the FAR transformer page.
5. **Cleanups**:
* Pipeline __call__ no longer passes is_causal=False to the bidi forward (the
bidi class doesn't accept it).
* Pipeline class docstrings drop stale references to init_*_model flags.
Local tests: 22/22 pass (12 scheduler + 10 transformer covering both classes).
ruff check / format / check_copies clean.
Hub artifacts (model_index.json, transformer/config.json, scheduler config) need
to be regenerated for the released checkpoints; the HF update guide will be
delivered separately.
* [AnyFlow] Phase 3: convention compliance against .ai/AGENTS.md + .ai/models.md
Hard violations (per official diffusers guidelines):
* drop einops dependency — replace 25+ rearrange() calls with native
permute/reshape/unflatten in transformer + both pipelines
* device-gate torch.float64 — apply_rotary_emb and AnyFlowRotaryPosEmbed now
fall back to float32 / complex64 on MPS / NPU; freqs are lazily rebuilt
per-device via _build_freqs (matches transformer_wan / transformer_flux
pattern)
* migrate attention to dispatch_attention_fn — replace direct
F.scaled_dot_product_attention calls with dispatch_attention_fn (works
with sage / flash / native backends); introduce AnyFlowAttention(
AttentionModuleMixin) with _default_processor_cls / _available_processors;
rename processors to AnyFlowAttnProcessor / AnyFlowCrossAttnProcessor and
declare _attention_backend / _parallel_config class attrs
* drop dead config fields — qk_norm and added_kv_proj_dim are pruned from
both transformer __init__ signatures and AnyFlowTransformerBlock;
AnyFlowAttention is hardcoded to rms-norm-across-heads (the only scheme
the released checkpoints use) and has no add_k_proj path (T2V only)
* add _repeated_blocks = ["AnyFlowTransformerBlock"] to both transformer
classes for compile_repeated_blocks() support (matches Wan)
* annotate prepare_latents with `# Copied from diffusers.pipelines.wan.
pipeline_wan.WanPipeline.prepare_latents`; the pipeline-side rearrange
to (B, T, C, H, W) layout is moved to the call site
State-dict keys are preserved (legacy Attention had identical to_q / to_k /
to_v / to_out / norm_q / norm_k naming), so existing AnyFlow checkpoints load
bit-exactly into the new AnyFlowAttention class.
The HF Hub config-update guide is updated correspondingly: transformer/
config.json now drops qk_norm and added_kv_proj_dim alongside the previous
init_far_model / init_flowmap_model / chunk_partition removals.
22 fast CPU tests still pass; ruff format / ruff check / check_copies all
clean.
* [AnyFlow] FAR fast-test compat: rope 0-dim guard + flex_attention CPU/head-dim fallbacks + KV-cache dtype + num_timesteps
Phase 3 migrated bidi + cross-attention to dispatch_attention_fn but the FAR
causal path still calls flex_attention directly, which has hard requirements
(CPU compile, head_dim >= 16) that fail on PipelineTesterMixin's tiny dummy
components. Real ckpts (head_dim=128, CUDA) never hit these branches; bit-exact
numerical equivalence with FAR-Dev preserved on all 4 released ckpts (forward
0.00e+00, backward kernel-nondet only, ratio 1.000).
Code fixes:
1. AnyFlowRotaryPosEmbed._forward_compressed_frame / _forward_full_frame now
short-circuit to an empty tensor when num_frames / height / width is 0.
PipelineTesterMixin's dummy VAE has scale_factor_spatial=8, so a 16x16 raw
spatial input becomes a 2x2 latent which then floors to 0 against
compressed_patch_size=(1, 4, 4); the original
`freqs[:0].view(0, k, 1, -1)` reshape was ambiguous in that regime.
2. flex_attention dispatch: split the module-load
`torch.compile(flex_attention, dynamic=True)` into `_flex_attention_eager`
(always available) plus `_flex_attention_compiled`, with a tiny wrapper
that picks compiled for CUDA tensors and eager for CPU. Avoids
torch._inductor C++ codegen failures that broke fast tests after
`pipe.to("cpu")`. CUDA performance unchanged (L10 benchmark: 0.0% delta on
bidi 1.3B fwd, 0.0% delta on FAR causal 1.3B fwd).
3. AnyFlowAttnProcessor (FAR causal branch): when head_dim < 16
(flex_attention's hard minimum) zero-pad q/k/v's last dim to 16 and pass
`scale=1/sqrt(original_head_dim)` to flex_attention. Padded value rows
contribute 0, so trimming the output back is mathematically equivalent.
Released ckpts use head_dim=128 so the branch is never taken in production.
4. pipeline_anyflow_far.encode_kv_cache: replace the hardcoded
`latents.to(torch.bfloat16)` with `self.transformer.dtype`. The hardcoded
bf16 crashed conv3d on dummy fp32 components ("Input type (BFloat16) and
bias type (float) should be the same"); real bf16 ckpts are unaffected.
5. pipeline_anyflow_far._denoise_rollout sets
`self._num_timesteps = (len(chunk_partition) - num_context_chunks) * num_inference_steps`
before the chunk loop, so PipelineTesterMixin.test_callback_cfg's
`pipe.num_timesteps`-based assertion matches the actual number of callback
fires (chunks * NFE) instead of the previous hardcoded num_inference_steps.
Tests:
* test_callback_inputs cannot pass without changing FAR's chunk-wise output
semantics — it zeroes latents on the final step and asserts the *entire*
output buffer is zero, but only the active chunk's slice is overwritten in
a chunk-wise rollout. Marked `@unittest.skip` with a detailed rationale;
callback functionality itself is still covered by test_callback_cfg.
* Full pytest run on tests/pipelines/anyflow/ +
tests/models/transformers/test_models_transformer_anyflow.py +
tests/schedulers/test_scheduler_flow_map_euler_discrete.py: 81 passed,
0 failed, 11 skipped.
Quality gates:
* `ruff check` and `ruff format --check` clean across all AnyFlow files.
* `python utils/check_copies.py` clean.
* `python utils/check_dummies.py` clean.
* [AnyFlow] docs/code: paper-release tidy-up
User-facing alignment with the official HF Hub model card and the day-of-announcement
materials at https://huggingface.co/collections/nvidia/anyflow.
* Fill in the arXiv identifier 2605.13724 (5 paper links + 2 BibTeX entries).
* Rename TV2V → V2V across docs + pipeline_anyflow{,_far}.py so the diffusers
copy uses the same Video-to-Video terminology as the official model card.
* Add the [nvidia/anyflow](https://huggingface.co/collections/nvidia/anyflow)
HF collection link to the three tutorial intros.
* Drop the temporary "guyuchao/* staging" tip from the EN tutorial / API page
/ ZH tutorial — the nvidia/AnyFlow-*-Diffusers repos are now live.
* Wire up NVlabs/AnyFlow (training code) and nvlabs.github.io/AnyFlow (project
page) in place of the prior <github-org> / <project-page-url> placeholders.
* Cite the authors (Yuchao Gu, Guian Fang et al.) and NUS ShowLab × NVIDIA
affiliation in the main tutorial, API pipeline page, and both transformer
model pages; BibTeX uses the standard `and others` to elide the full list
until the next pass.
Working tree, CI gates, and tests after the change:
ruff format --check ✓
ruff check ✓
python utils/check_copies.py ✓
python utils/check_dummies.py ✓
pytest tests/models + tests/schedulers (22 fast) ✓
No production code logic changes — only docstring wording inside pipeline
files (TV2V → V2V).
* [AnyFlow] docs: drop in official BibTeX (full author list)
Replace the placeholder ``@article{gu2026anyflow, author = {Gu, Yuchao and
Fang, Guian and others}, ...}`` block in both the English and Chinese
tutorials with the canonical ``@misc{gu2026anyflowanystepvideodiffusion,
...}`` form from arxiv.org/abs/2605.13724, which lists all seven authors:
Yuchao Gu, Guian Fang, Yuxin Jiang, Weijia Mao, Song Han, Han Cai,
Mike Zheng Shou.
Docs-only.
* [AnyFlow] align with diffusers conventions + drop training-only code
Scheduler
- FlowMapEulerDiscreteScheduler.step now returns a
FlowMapEulerDiscreteSchedulerOutput dataclass (or tuple with return_dict=False)
and uses the conventional positional order (model_output, timestep, sample,
r_timestep).
- Drop training-only helpers: adaptive_weighting, set_train_weight,
get_train_weight, linear_timesteps_weights, and the weight_type config field.
- Add scale_model_input no-op for API parity; raise ValueError on missing
r_timestep.
Transformer
- Remove gate_track debug write inside
AnyFlowDualTimestepTextImageEmbedding.forward_timestep.
- Compile flex_attention lazily on first CUDA call instead of at import time.
- Replace assert with ValueError in build_block_mask.
- Resolve <arxiv-id> placeholders to 2605.13724.
Pipelines (AnyFlowPipeline + AnyFlowFARPipeline)
- Add EXAMPLE_DOC_STRING + @replace_example_docstring and full __call__
docstrings covering every argument.
- Move use_mean_velocity from __init__ to __call__ so save/load round-trips.
- Drop _denoise_rollout's grad_timestep branch (DMD on-policy training rollout),
the inner inference_range closure, and the redundant negative-prompt concat.
- Replace asserts with ValueError; wire show_progress to tqdm; rename inference
-> _inference; remove dead current_timestep property.
- Update scheduler.step call sites to the new signature.
- Trim class docstrings to inference-only language.
Pipeline output
- Add Apache 2.0 license header; switch to relative import.
Auto pipeline / conversion script
- Register AnyFlowFARPipeline in AUTO_IMAGE2VIDEO_PIPELINES_MAPPING and
AUTO_VIDEO2VIDEO_PIPELINES_MAPPING.
- Document the weights_only=False requirement in the conversion script.
Tests
- Scheduler tests use the new step signature and verify the Output dataclass
contract.
- Drop the four obsolete training-weight tests; drop weight_type kwarg from
pipeline test fixtures; remove internal milestone names from TODO comments.
Docs
- Resolve <arxiv-id> in the scheduler docs page.
- Trim DMD / on-policy distillation language in EN/ZH tutorials and the
pipelines page; the paper abstract quote is preserved verbatim.
* [AnyFlow] split FAR causal transformer into transformer_anyflow_far.py
Per @dg845's review on #13745: extract FAR causal modules into a dedicated
sibling file so each transformer variant reads in isolation. Shared submodules
are duplicated via `# Copied from` so `make fix-copies` keeps both in sync.
- `transformer_anyflow.py`: bidi-only. `AnyFlowAttnProcessor` no longer carries
the flex/KV-cache branch (was: dispatch in one branch, bare flex_attention in
the other); `AnyFlowRotaryPosEmbed` drops the compressed-frame helpers and
the `is_causal` arg; `AnyFlowDualTimestepTextImageEmbedding` drops its causal
branch. `AnyFlowTransformerBlock` keeps a single class with a new
`is_causal: bool = False` ctor flag that selects the self-attn processor —
the forward path is identical in both modes, only the processor differs.
- `transformer_anyflow_far.py`: new. Contains `AnyFlowFARTransformerOutput`,
`AnyFlowCausalAttnProcessor` (routed through `dispatch_attention_fn(backend=
"flex")` with a clear ValueError when a non-flex backend is configured; the
BlockMask is consumed only by the flex backend in `_native_flex_attention`),
`AnyFlowDualTimestepTextImageEmbeddingCausal`, `AnyFlowCausalRotaryPosEmbed`,
`AnyFlowFARTransformer3DModel`, and `# Copied from` clones of the shared
shared `AnyFlowAttention`/`AnyFlowCrossAttnProcessor`/`AnyFlowImageEmbedding`/
`AnyFlowTransformerBlock`/`AnyFlowAttnProcessor` modules.
Verified bit-exact against the pre-refactor branch on H200 (float32):
- bidi: L2 = 0.000e+00, max|Δ| = 0.000e+00
- FAR : L2 = 4.772e-06, max|Δ| = 3.576e-07
The FAR delta is fp32 accumulation noise from the dispatch path permuting
(B,L,H,D) ↔ (B,H,L,D) around the same `flex_attention` kernel.
Addresses review comments at transformer_anyflow.py:215, :261, :450, :622,
:671, :958.
* [AnyFlow] pipeline cleanup: video_processor, encode_video, inline rollout, kwarg rename
Per @dg845's review on #13745, applied to both bidi `AnyFlowPipeline` and
causal `AnyFlowFARPipeline`:
- Use `self.video_processor.preprocess_video(...)` instead of the manual
`* 2 - 1` normalize.
- Merge `vae_encode` + `encode_latents` + `_normalize_latents` into a single
`encode_video` method, mirroring `WanImageToVideoPipeline.encode_image`'s
flat structure.
- Inline `_denoise_rollout` into `AnyFlowPipeline.__call__`. For the FAR
pipeline, inline both `_denoise_rollout` and `_inference` as a nested loop
(outer over chunks, inner over denoising steps), mirroring
`WanAnimatePipeline.__call__`. `encode_kv_cache` is intentionally kept as a
method — it is one transformer call with a different `kv_cache_flag` mode
(cache-write), and inlining it would interleave two distinct forward
semantics in the same loop body and lose readability.
- Rename `context_sequence` → `video` (pixel-space) + `video_latents`
(pre-encoded), matching `WanVideoToVideoPipeline`. For the FAR pipeline,
the old `{"raw"/"latent"}` dict form is replaced by the two kwargs.
Mutually-exclusive validation raises `ValueError`.
Addresses review comments at pipeline_anyflow.py:358, :372, :393, :473 and
pipeline_anyflow_far.py:395, :489, :675.
* [AnyFlow] scheduler: N-length timesteps + step defaults r_timestep
Per @dg845's review on #13745:
- `set_timesteps(N)` now produces `N` timesteps backed by an internal
`sigmas[N+1]` linspace, matching `FlowMatchEulerDiscreteScheduler.set_
timesteps`. The final sigma (== 0) is the implicit r-endpoint of the last
step; the pipeline rollouts iterate `for i, t in enumerate(timesteps)`
without the old `[:-1]` slicing.
- `step(r_timestep=None)` now defaults to the next timestep on the schedule
(resolved via fp-tolerant `argmin` over `sigmas[:-1]`), instead of raising.
Any-step sampling is preserved when `r_timestep` is explicit. The raise
stays only for the case where the caller passes a `timestep` value that
isn't on the schedule and provides no `r_timestep` — there's no sensible
default in that case.
- Build sigmas in float64 on CPU then move to the target device, with a
float32 downcast for MPS / NPU (float64 isn't supported on those backends).
Pipeline rollout loops updated to compute `r = sigmas[i + 1] * num_train_
timesteps` for the model's `r_timestep` input and pass `r_timestep=None` to
`scheduler.step` (which resolves it from the schedule internally).
Addresses review comments at scheduling_flow_map_euler_discrete.py:107 and
:148.
* [AnyFlow] tests: regenerate via generate_model_tests.py; split bidi/FAR files
Per @dg845's review on #13745: replaced the hand-rolled transformer tests
with the standard mixin-based suite produced by `utils/generate_model_tests
.py`, and split the FAR causal model tests into their own file to mirror the
transformer file split.
- `tests/models/transformers/test_models_transformer_anyflow.py`: regenerated
bidi suite. Pulls in `ModelTesterMixin`, `MemoryTesterMixin`,
`TrainingTesterMixin`, `AttentionTesterMixin`, `TorchCompileTesterMixin` via
`BaseModelTesterConfig`, with `get_init_dict()` / `get_dummy_inputs()`
filled in for the small bidi config used in CI.
- `tests/models/transformers/test_models_transformer_anyflow_far.py`: new.
Same mixin set (TorchCompile is intentionally skipped — FAR's
`_build_causal_mask` uses `flex_attention.create_block_mask(_compile=False)`
which conflicts with the standard compile tester's assumptions; the bidi
file covers compile, FAR is bit-exact-validated end-to-end on H200 via the
pipeline replay). Also carries an `AnyFlowCausalAttnProcessor` smoke test
that exercises the backend gate (non-flex backends must raise) and asserts
the `AnyFlowFARTransformerOutput` dataclass exposes the expected fields.
Addresses review comments at test_models_transformer_anyflow.py:71 and :128.
* [AnyFlow] docs: update for video / video_latents kwarg rename
Following the pipeline kwarg refactor in e9d50b2, sweep the user-facing docs
to reflect the new API:
- `docs/source/en/api/pipelines/anyflow.md`: T2V / I2V / V2V code examples now
use `video=` instead of `context_sequence={"raw": ...}`. The "Generation
with AnyFlow (FAR Causal)" intro describes the new mutually-exclusive
`video` / `video_latents` selector.
- `docs/source/en/using-diffusers/anyflow.md`: the scenario selector table,
the "Image-to-video and video-to-video" walkthrough, and the closing note
about pre-encoded latents are all updated. `vae_encode` references are
replaced with `encode_video`.
* [AnyFlow] tests: skip FAR training tests on CPU (flex backward); align scheduler tests with N-length timesteps
- TestAnyFlowFARTransformer3DTraining: skip test_training / test_training_with_ema /
test_gradient_checkpointing_equivalence on CPU. FAR causal self-attn uses
torch.nn.attention.flex_attention whose backward kernel is GPU-only.
- test_scheduler_flow_map_euler_discrete: assert timesteps is N-length (not N+1) and
the sigma=0 r-endpoint lives in self.sigmas[-1]; test_step_one_shot_sampling now
exercises r_timestep=None (resolved from sigmas) since N=1 has no timesteps[1].
* [AnyFlow] docs: complete forward() Args: sections for check_forward_call_docstrings
main #13758 added utils/check_forward_call_docstrings.py which requires every signature
arg to appear as its own `name (...):` entry under Args:. Expand the bidi and FAR
transformer forward docstrings to list each parameter individually.
* [AnyFlow] apply 5/21 review suggestions (A: 1-click)
FAR transformer:
- AnyFlowCausalAttnProcessor: default _attention_backend = 'flex' (was None);
remove None from _SUPPORTED_BACKENDS. None previously fell through to SDPA
which silently ignored the BlockMask; failing loudly is the right default.
- dispatch_attention_fn call: read self._attention_backend instead of hardcoded
'flex', so '_native_flex' selection works.
- _build_freqs / _forward_full_frame: add '# Copied from' to bidi RoPE.
Pipelines:
- bidi + FAR docstrings: video shape (B, C, T, H, W) -> (B, T, C, H, W) to
match VideoProcessor.preprocess_video.
- FAR EXAMPLE_DOC_STRING: single-frame I2V tensor wrap uses unsqueeze(1) for the
T axis instead of unsqueeze(2).
- FAR encode_video: drop duplicated @torch.no_grad() decorator.
Tests:
- test_anyflow / test_anyflow_far: lift the test_save_load_optional_components
skip (the test actually passes).
- FAR processor smoke test: assert default backend is 'flex' (was 'None').
* [AnyFlow] apply 5/21 review suggestions (B: refactors)
Pipelines:
- check_inputs accepts video / video_latents and raises early on:
(a) mutual exclusion (was checked late in __call__);
(b) FAR's (num_frames - 1) % 4 == 0 constraint.
__call__ no longer carries duplicate validation.
- FAR pipeline: drop the show_progress kwarg and replace the single tqdm with
nested progress bars in the LLaDA-2 pattern — outer 'Chunks' (position=0)
and per-chunk inner 'Inference Steps' (position=1, leave=False) — both
picking up DiffusionPipeline._progress_bar_config (so set_progress_bar_config
controls them, including disable=None).
Scheduler:
- step() resolves source and target sigmas by indexing self.sigmas via the new
index_for_timestep(), instead of dividing the input timesteps by
num_train_timesteps. This keeps the math correct for any future schedule
whose timestep/sigma relationship is non-linear. For an off-schedule
r_timestep the code falls back to r / num_train_timesteps, so explicit
any-step sampling outside the schedule still works (and t off-schedule with
r=None still raises a clear ValueError, as before).
Numerical equivalence: for the shipped linspace+shift schedule the two
formulations are bit-identical (verified: max abs diff = 0.0 over an N=8,
shift=5 schedule).
* [AnyFlow] apply Claude bot review (5/21): 8 findings beyond dg845's list
Finding #1 — attention_kwargs plumbing:
Both transformers now decorate forward() with @apply_lora_scale('attention_kwargs')
(matches Wan); pipelines forward attention_kwargs to the transformer + encode_kv_cache,
and the unused parameter is dropped from the inner _forward_train / _forward_cache /
_forward_inference signatures. Pipeline docstrings updated to the standard wording.
Finding #2 — naming:
Rename far_cfg -> layout_cfg in the bidi transformer (the bidi path is not FAR; the
FAR transformer keeps far_cfg, which is accurate there).
Finding #3 — scheduler state machine:
Add _step_index, _begin_index, step_index property, begin_index property,
set_begin_index(), _init_step_index(). step() lazily initializes and advances the
counter so downstream callbacks / composable schedulers can observe rollout progress.
Sigma resolution remains a pure function of (timestep, r_timestep) — calling step()
twice with identical args still returns identical prev_sample (idempotent).
Finding #4 — redundant @torch.no_grad():
Drop the redundant decorators on bidi pipeline's encode_video and FAR pipeline's
encode_kv_cache (callers are already in __call__'s no-grad scope).
Finding #5 — dead code:
Remove the unreachable temb.ndim == 2 else branch from the bidi transformer's
output-norm path (condition_embedder.forward always returns a 3D temb).
Finding #6 — private rename:
forward_far_patchify[_inference] -> _forward_far_patchify[_inference] (only called
internally by _forward_train / _forward_cache / _forward_inference).
Finding #7 — pipeline comment numbering:
Bidi + FAR pipelines renumber steps so the # 4. slot is no longer skipped.
Finding #8 — mask-mod comment numbering:
_build_causal_mask numbered comments now run 1) 2) 3) ... (was 1) 3) 4) ...).
Tests:
- New test_step_index_advances + test_set_begin_index_anchors_step_index in the
scheduler test file exercise the new state machine.
- All existing pipeline / transformer / scheduler tests still pass (85 passed,
85 skipped on CPU).
Bit-exact: 8-step rollout vs the previous formulation, max abs diff = 0.0 (the new
sigma-lookup is byte-identical to t/num_train_timesteps on this schedule).
* [AnyFlow] scheduler: honour off-schedule any-step in _init_step_index; drop dead _resolve_next_timestep
Audit caught two issues in the previous scheduler commit:
1. The new state machine raised in _init_step_index whenever the first timestep
wasn't on the active schedule, contradicting the documented contract that
step() falls back to t/num_train_timesteps for off-schedule any-step
sampling. The fall-back numerics were intact but they were unreachable —
the init check fired first.
Fix: _init_step_index now initializes _step_index to 0 when the timestep is
off-schedule (still a valid observable counter for callbacks). step()'s
sigma resolution is untouched, so on-schedule rollouts stay bit-exact and
off-schedule any-step sampling actually runs again. Regression test:
test_step_off_schedule_anystep_supported.
2. _resolve_next_timestep had no remaining callers after the step() rewrite
inlined the same lookup. Removed (private helper, no external API).
* [AnyFlow] docs: align user guides with video shape + kwarg fixes
- en api/pipelines/anyflow.md: video shape (B, C, T, H, W) -> (B, T, C, H, W);
example tensor wrap uses unsqueeze(0).unsqueeze(1) and permute(0, 3, 1, 2)
to match VideoProcessor.preprocess_video's 5D contract.
- zh using-diffusers/anyflow.md: same shape fixes; also flip the I2V / V2V
examples from the obsolete context_sequence={...} dict to the current
video= / video_latents= kwargs; helper to_video_tensor returns (1, T, C, H, W);
add a note about mutual exclusion.
* [AnyFlow] tests: drop @slow integration test scaffolds for initial PR
.ai/skills/model-integration/SKILL.md is explicit: 'No integration / slow
tests in the initial PR — don't add anything gated on @slow / RUN_SLOW=1
yet.' Our two integration test classes were shape-only assertions with TODOs
for a future numeric reference, so dropping them loses no actual coverage —
the relevant rollouts are covered by H200 bit-exact replay outside the
pytest suite. Can land a follow-up PR after merge with proper numeric
reference slices once the maintainer is comfortable enabling slow tests.
* Apply style fixes
* [AnyFlow] apply 5/22 dg845 review: comment cleanups + custom sigmas/timesteps schedule
dg845 third pass — 7 of 9 comments applied; the 8th (custom sigmas/timesteps support)
matches FlowMatchEulerDiscreteScheduler conventions; the 9th (_build_causal_mask
refactor) is explicitly marked non-blocking and deferred to a follow-up that also
re-enables TorchCompileTesterMixin.
Comment cleanups:
- transformer_anyflow.py:704 temb output-norm comment: drop redundant 'no ndim==2 branch'.
- pipeline_anyflow.py:550 denoise loop comment: '# 6. Denoising loop'.
- pipeline_anyflow_far.py:684 denoise loop comment: '# 8. Denoising loop (outer over
chunks, inner over timesteps).'.
- pipeline_anyflow_far.py:702 drop trailing inline comment on `timesteps = scheduler.timesteps`.
- scheduling_flow_map_euler_discrete.py: clearer wording on the off-schedule `r_timestep`
error.
Custom schedule support:
- FlowMapEulerDiscreteScheduler.set_timesteps gains `sigmas` and `timesteps` kwargs
mirroring FlowMatchEulerDiscreteScheduler. Default behaviour is unchanged
(linspace + shift); the validation + length-N → length-N+1 terminal-0 append are
shared with the default path so on-schedule rollouts stay bit-exact.
- AnyFlowPipeline.__call__ and AnyFlowFARPipeline.__call__ accept `sigmas` and
`timesteps` kwargs, override num_inference_steps from their length, and forward
to set_timesteps (matches LTX2Pipeline pattern).
- New scheduler tests: test_set_timesteps_custom_sigmas and
test_set_timesteps_custom_timesteps cover both override paths.
Dtype skip on save/load:
- TestAnyFlowTransformer3D and TestAnyFlowFARTransformer3D now skip
test_from_save_pretrained_dtype_inference (parametrized over fp16/bf16), mirroring
WanTransformer3DModel's skip — the test's tolerance requirements are too high for
meaningful signal under AnyFlow's flow-map mixed-precision sampling.
* [AnyFlow] docs: apply hf-doc-builder line wrap (max_len 119)
CI doc-builder style check flagged 3 files with docstring lines >119 chars.
Ran 'doc-builder style src/diffusers docs/source --max_len 119' to autoformat;
content unchanged, line wrapping only.
* [AnyFlow] apply 5/22 follow-up review: new_zeros terminal sigma + cleanup
dg845 blocking suggestion (r3287274209):
- scheduling_flow_map_euler_discrete.py:185 — use `working_sigmas.new_zeros(1)`
instead of `torch.zeros(1, dtype=...)` so the appended terminal sigma inherits
both device and dtype from working_sigmas. The current working_sigmas always
starts on CPU so the device mismatch is latent, but new_zeros is the correct
defensive pattern and matches how the published FAR test fixtures run on CUDA.
Claude bot final-review follow-ups:
- transformer_anyflow_far.py: drop three stale `# step 3: generate attention mask`
comments left over from the original numbered-step structure (bot #6).
- pipeline_anyflow_far.py: annotate `encode_video` with
`# Copied from diffusers.pipelines.anyflow.pipeline_anyflow.AnyFlowPipeline.encode_video`
and align docstring + inline comment so `make fix-copies` keeps them in sync (bot #3).
Skipped (not real / judgment-call):
- bot #2 (private rename of `_forward_far_patchify*`) — already done in 84605d5;
bot was looking at a stale snapshot.
- bot #4 (check_inputs `# Copied from`) — FAR's check_inputs has an extra
`(num_frames - 1) % 4 == 0` constraint that doesn't map onto the bidi version,
so a clean `# Copied from` link would require restructuring. Bot called it a
consistency nit; leaving as-is.
- bot #5 (`encode_kv_cache` → `_encode_kv_cache`) — bot itself flagged this as
judgment-call territory; the helper is a coherent operation that advanced
inference callers may want to invoke directly.
---------
Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
|
||
|
Sayak Paul
|
6b0f61cc39 |
[docs] fix ace step checkpoint id. (#13787)
* fix ace step checkpoint id. * style |
||
|
Linoy Tsaban
|
7aa746c4b0 | [LTX 2.3] update docs (#13788) | ||
Steven Liu
|
9a2923d501 |
[docs] remove pipeline examples section (#13771)
* docs * links |
||
|
Sayak Paul
|
2375f70f67 |
[core] minimum torch version is 2.6 (#13725)
minimum torch version is 2.6 |
||
|
Sayak Paul
|
62ec337e30 |
[agents] add a section on tests in the ai skill and integration guides. (#13752)
* add a section on tests in the ai skill and integration guides. * up |
||
Dhruv Nair
|
b9b7df3627 |
Update contribution guidelines (#13753)
* update * update * update * update |
||
Wai Ting Cheung
|
fc77592427 |
feat: Add Motif-Video model and pipelines (#13551)
* feat: add Motif Video T2V and I2V pipelines with AdaptiveProjectedGuidance support Add complete Motif Video implementation to diffusers: New Models: - Add MotifVideoTransformer3DModel with T5Gemma2Encoder for multimodal conditioning - Supports text-to-video and image-to-video generation with vision tower integration New Pipelines: - Add MotifVideoPipeline for text-to-video generation - Default resolution: 736x1280, 121 frames, 25 fps - Supports classifier-free guidance and AdaptiveProjectedGuidance - Add MotifVideoImage2VideoPipeline for image-to-video generation - First frame conditioning with vision encoder - Same defaults as T2V pipeline Enhanced Guidance: - Update AdaptiveProjectedGuidance with normalization_dims parameter - Support "spatial" normalization for 5D tensors (per-frame spatial normalization) - Support custom dimension lists for flexible normalization - Update AdaptiveProjectedMixGuidance with same parameter Documentation & Tests: - Add comprehensive API documentation for transformer and pipelines - Add test suites for both T2V and I2V pipelines - Register all new components in __init__ files - Add dummy objects for torch and transformers backends Total: 18 files changed, 3416 insertions(+), 2 deletions(-) * Remove linear quadratic * Remove musicldm * Update docstring * Address vision_encoder comment * Add copy source in I2V pippeline * Refactor _get_prompt_embeds Co-authored-by: Beomgyu Kim <beomgyu.kim@motiftech.io> * Fix a typo * Refactor MotifVideo transformer to use diffusers Attention conventions - Use default Attention class with custom MotifVideoAttnProcessor2_0 - Inline cross-attention in transformer blocks - Use dispatch_attention_fn for backend support - Inherit AttentionMixin for attn_processors/set_attn_processor - Move TransformerBlockRegistry to _helpers.py - Add _repeated_blocks for regional compilation * Use base classes for scheduler and guider * Implement MotifVideoAttention * Update style and quality * Fix a typo * Fix a typo * Fix a typo * Update year * Address rope dtype * Update docstring and remove frame_rate * Address unused sigmas * Add available processors * Address copy from comment * Remove torch.no_grad() * Remove use_attention_mask * Address inline cross-attention * Address compute dtype * Remove unused variables * Merge main APG into this branch and update documentation * Refactor cross attention processor * Remove unused timestep * Inline create_attention_mask * Make guider required * Address encode_prompt comment * Address preprocess_video comment * Use T5Gemma2Encoder in test cases * Address None feature_extractor * Address output type * Renable skipped tests * Update style and quality * Generate standard transformer test case * Add model test case * Remove guider in documentation * Implement cross_attn layer * Remove prepare_negative_prompt * Address latent is None * Clean up feature_extractor * Fix prepare_latents * Remove transformers assertion * Fix style and quality * Fix python utils/check_copies.py --fix_and_overwrite python utils/check_dummies.py --fix_and_overwrite outputs * Add dropout rate to text config * Skip tests requiring guidance_scale * Fix encode_prompt in test cases * Fix test_cpu_offload_forward_pass_twice * Update tests/pipelines/motif_video/test_motif_video.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update tests/pipelines/motif_video/test_motif_video.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update tests/pipelines/motif_video/test_motif_video.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update tests/pipelines/motif_video/test_motif_video_image2video.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Address test_attention_slicing_forward_pass comment * Update tests/pipelines/motif_video/test_motif_video_image2video.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update tests/pipelines/motif_video/test_motif_video_image2video.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Update tests/pipelines/motif_video/test_motif_video_image2video.py Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * Skip I2V test cases * Fix style and quality * Add docs to toctree * Fix docs location in toctree and add link in overview * Inline gradient checkpointing * Add _keep_in_fp32_modules for timestep_embedder * Address num_decoder_layers comment * Address guider is not None comment * Remove _keep_in_fp32_modules * Address parameter_dtype comment --------- Co-authored-by: Ken Cheung <ken.cheung@motiftech.io> Co-authored-by: Beomgyu Kim <beomgyu.kim@motiftech.io> Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> Co-authored-by: YiYi Xu <yixu310@gmail.com> |
||
彼彼
|
0ceddf7dca |
[docs] add docs for JoyAI-Image-Edit (#13726)
add docs |
||
|
Sayak Paul
|
e5cf820fc3 |
[docs] add magcache to caching api listing (#13714)
add magcache to caching api listing |
||
Ting-Yun Chang
|
4ca863323d |
Add LoRA support for Cosmos Predict 2.5 and fix pipeline to match official Cosmos repo (#13664)
* support lora for cosmos 2.5
* Fix inconsistencies with cosmos official repo in VAE encoding, text encoder attention implementation, and timestep scaling
* Support f_min and f_max in linear_scheduler warmup
* Add requirements and dataset preprocessing scripts to run examples
* Add LoRA training scripts
* Add LoRA eval scripts
* add assets for blogpost
* Fix(scheduler): device mismatch from upstream
|
||
|
Sayak Paul
|
5bd51bd189 |
Update attention_backends.md to update FA3 minimum support to Ampere (#13283)
* Update attention_backends.md * Update docs/source/en/optimization/attention_backends.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> --------- Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> |
||
Gong Junmin
|
1a8a17b71b |
Add ACE-Step pipeline for text-to-music generation (#13095)
* Add ACE-Step pipeline for text-to-music generation
Rebased on origin/main from the original pr-13095 branch (3 commits squashed).
- AceStepDiTModel: Diffusion Transformer with RoPE, GQA, sliding window,
AdaLN timestep conditioning, and cross-attention.
- AceStepConditionEncoder: fuses text / lyric / timbre into a single
cross-attention sequence.
- AceStepPipeline: text2music / cover / repaint / extract / lego / complete.
- Conversion script for the original checkpoint layout.
- Docs + tests.
* Fix ACE-Step pipeline audio quality and auto-detect turbo/base/sft variants
The PR's original inference produced low-quality audio on turbo because the
pipeline (a) mangled the SFT prompt format, (b) applied classifier-free guidance
with the wrong unconditional embedding (empty-string encoded vs. the learned
`null_condition_emb`), and (c) hardcoded turbo defaults even when loading a
base/SFT checkpoint.
Changes:
* Converter preserves `null_condition_emb` (stored under the condition encoder)
and propagates `is_turbo`/`model_version` into the transformer config so the
pipeline can route per-variant defaults.
* `AceStepConditionEncoder` registers `null_condition_emb` as a learned
parameter matching the original module.
* Pipeline auto-detects variant via `is_turbo`/`model_version` and picks
defaults that match `acestep/inference.py`:
* turbo: steps=8, shift=3.0, guidance_scale=1.0 (no CFG)
* base/SFT: steps=27, shift=1.0, guidance_scale=7.0
* Base/SFT timestep schedule uses the linear+shift transform from
`acestep/models/base/modeling_acestep_v15_base.py`; turbo still uses the
hardcoded 8-step `SHIFT_TIMESTEPS` table.
* CFG reuses the learned `null_condition_emb` and batches the
conditional+unconditional forwards into a single transformer call.
* `SFT_GEN_PROMPT` matches the newline layout in `acestep/constants.py` so the
text encoder sees the same prompt distribution it was trained on.
DiT parity vs. the original ACE-Step 1.5 turbo DiT is bit-identical
(max_abs=0.0 in fp32 eager/SDPA across 4 seed/shape cases) — see
scripts/dit_parity_test.py.
* Add ACE-Step parity test scripts
Two developer-facing parity harnesses live under scripts/:
* dit_parity_test.py — loads the same converted turbo weights into the
original AceStepDiTModel and the diffusers AceStepDiTModel, drives
identical (hidden_states, timestep, timestep_r, encoder_hidden_states,
context_latents) inputs, and asserts max-abs-diff ≤ 1e-5 in fp32
eager/SDPA. Currently passes bit-identical (max_abs=0) across four
shape/seed cases including batched + odd-length paths.
* audio_parity_jieyue.py — full end-to-end audio parity. Given the same
JSON example, runs both the original ACE-Step 1.5 pipeline and the
diffusers AceStepPipeline at matched seed/precision (bf16 + FA2 by
default) and saves side-by-side .wav files for listening verification.
Supports text2music / cover / repaint × turbo / base / sft via a
--matrix mode that writes 18 wavs named
{variant}_{task}_{official,diffusers}.wav.
* Route SFT parity to acestep-v15-sft checkpoint
On jieyue the release tree has a dedicated SFT checkpoint at
checkpoints/acestep-v15-sft with its own modeling_acestep_v15_base.py
shipped under acestep/models/sft/. Point the SFT row of the parity matrix
at that checkpoint / module so we're testing the actual SFT weights, not
the plain base ones.
* audio_parity_jieyue: fix doubled 'acestep-' in cache path; --converted-root flag
Previously the converted-pipeline cache dir was
`/tmp/acestep-<variant>-diffusers` but <variant> already starts with
"acestep-", giving `/tmp/acestep-acestep-v15-turbo-diffusers`. Drop the
prefix.
On jieyue the overlay rootfs (including /tmp) only has a few GB free; a
full turbo conversion needs ~5 GB per variant. Add --converted-root (env
ACESTEP_CONVERTED_ROOT) so the cache can live on vepfs.
* audio_parity_jieyue: two-phase matrix bootstraps cover/repaint from text2music
The ACE-Step release bundle on jieyue doesn't ship sample .wav/.mp3
files, so matrix mode had no default --src-audio and would skip
cover/repaint entirely. Run text2music first for every variant, then
reuse the TURBO official text2music output as the shared source for the
cover/repaint rows. Users can still override with --src-audio.
* audio_parity_jieyue: seed the diffusers generator on the pipeline device
The ORIGINAL ACE-Step pipeline seeds on the execution device
(`torch.Generator(device=device).manual_seed(seed)`), i.e. the CUDA RNG
stream when running on GPU. Previously the parity harness seeded the
diffusers side with a CPU generator, so even though the seed integer
matched, the two sides drew different noise from the outset and the
final outputs were essentially uncorrelated. Use the execution-device
generator on both sides for a fair comparison.
* Fix ACE-Step pipeline: switch to APG guidance + peak normalization
Two issues found after the first jieyue audio parity run:
1. The original base/SFT pipeline uses APG (Adaptive Projected Guidance,
acestep/models/common/apg_guidance.py) with a stateful momentum
buffer and norm/projection steps — NOT vanilla CFG. Using vanilla CFG
produced uncorrelated outputs vs. the reference (pearson ~0.0 on
20 s samples); this PR ports `_apg_forward` + `_APGMomentumBuffer`
and plugs them into the denoising loop when `guidance_scale > 1`.
Momentum is instantiated once per pipeline call (persists across
denoising steps) to match the reference semantics.
2. The post-VAE "anti-clipping normalization" in this pipeline was
`audio /= std * 5` with a `std<1 -> std=1` guard. The original
post-processing in
acestep/core/generation/handler/generate_music_decode.py is simple
peak normalization: `if audio.abs().max() > 1: audio /= peak`. The
std-based proxy both (a) let clips with peak < 1 leak through
unchanged (over-quiet) and (b) failed to bring clipping peaks to
exactly 1 in a bunch of base/SFT cases (observed max=1.000, std=0.200
repeatedly in the first parity run). Switch to peak normalization on
both sides.
Tested via scripts/audio_parity_jieyue.py on A800; re-run pending to
confirm the base/SFT correlation improvements.
* Fix ACE-Step chunk mask values to match the original pipeline
The DiT receives `context_latents = concat(src_latents, chunk_mask)` on the
channel dim, and was trained with chunk_mask values drawn from the three
sentinels documented in acestep/inference.py:
2.0 -> model-decided (default for text2music / cover / full-generation)
1.0 -> keep this latent frame from src_latents (repaint preserved region)
0.0 -> explicitly repaint this frame (only inside the repaint window)
Previously _build_chunk_mask returned all-1.0 for text2music (and cover /
lego), and an inverted 0/1 mask for repaint (1 inside the window, 0 outside).
Either case puts context_latents out of distribution. Switch text2music /
cover to the 2.0 sentinel and flip the repaint mask so it's 1.0 outside /
0.0 inside. Update the repaint src_latents zero-out to multiply by the new
mask (was `1 - chunk_mask`) so the zero region still lines up with the
repaint window.
* Add direct invoker for ACE-Step generate_music (ground truth)
Our earlier audio_parity_jieyue.py reconstructs the original pipeline by
calling AceStepConditionGenerationModel.generate_audio() directly, which
silently skips a lot of the real handler plumbing (conditioning masks,
silence-latent tiling, cover/repaint pre-processing, etc.). That made the
'official' wavs we saved sound wrong — flat, drone-like, not music.
This new script calls acestep.inference.generate_music end-to-end through
the real AceStepHandler, with LM + CoT explicitly disabled so we still have
a deterministic comparison. Use it to generate the ground-truth 'official'
wav for a given JSON example, then separately run the diffusers pipeline
with the same inputs and diff the two.
* run_official_generate_music: call initialize_service to bind a DiT variant
AceStepHandler() is a shell — you have to call handler.initialize_service(
project_root=..., config_path=..., device=..., use_flash_attention=..., ...)
before generate_music will work. Mirror what cli.py does at the equivalent
spot (around cli.py:1400).
* Fix silence-reference for ACE-Step timbre encoder
The root cause for the flat / drone-like outputs I was seeing (including
in my 'official' reconstruction): when no reference_audio is provided the
pipeline was feeding literal zeros to the timbre encoder. The real
handler feeds a slice of the learned `silence_latent` tensor.
The handler also transposes silence_latent on load (see
acestep/core/generation/handler/init_service_loader.py:214:
self.silence_latent = torch.load(...).transpose(1, 2)
) converting [1, 64, 15000] -> [1, 15000, 64] so that
`silence_latent[:, :750, :]` yields the expected [1, 750, 64] shape.
Changes:
* Converter: load silence_latent.pt, transpose to [1, T, C], bake into
the condition_encoder safetensors under key `silence_latent`.
(Also keeps the raw .pt file at the pipeline root for debugging.)
* AceStepConditionEncoder: register `silence_latent` as a persistent
buffer so from_pretrained loads it alongside the trained weights.
* Pipeline: when reference_audio is None, slice
`condition_encoder.silence_latent[:, :timbre_fix_frame, :]` and
broadcast across the batch instead of zeros. Emits a loud warning
(and falls back to zeros) if the buffer is all-zero — that means the
checkpoint was produced by an older converter and should be rebuilt.
* audio_parity_jieyue.py: the reference path now matches the handler's
silence-latent slicing.
Without this fix, every variant/task combo produced drone-like audio
even when my numeric DiT-forward parity claimed they were identical.
* Fix three more ACE-Step pipeline bugs I found by dumping real inputs
Instrumented the live generate_audio call in the real ACE-Step handler and
observed the exact tensors it sees — my diffusers pipeline was wrong in
three independent ways:
1. src_latents for text2music should be silence_latent tiled to
latent_length, NOT zeros. The handler fills no-target cases from
silence_latent_tiled (observed std=0.96). Zeros are OOD for the DiT
context_latents concat and produce drone-like outputs.
2. chunk_mask values cap at 1.0 (not 2.0). The handler starts with a
bool tensor (True inside the generate span, False outside); the
chunk_mask_modes=auto -> 2.0 override does NOT take effect because
the underlying tensor is bool, so setting entry = 2.0 casts to True.
After the later .to(dtype) float cast, the DiT sees 1.0/0.0 — exactly
what I observed in the captured tensor (unique values = [True]).
3. Default shift is 1.0 for ALL variants, including turbo. I was
defaulting turbo to shift=3.0 which picks a different SHIFT_TIMESTEPS
table (the 8-step schedule is keyed by shift, not variant).
Also:
* Added _silence_latent_tiled() helper that slices / tiles the learned
silence_latent (now loaded as a buffer on the condition encoder) to
the requested latent length.
* Repaint path now substitutes silence_latent (not raw zeros) inside
the repaint window — matches conditioning_masks.py.
* audio_parity_jieyue.py mirrors the same src/chunk/shift choices on
its 'original' leg for apples-to-apples parity once the buggy
reconstruction is removed from the picture.
* Add peak+loudness post-normalization to AceStepPipeline
The real pipeline normalizes audio in two stages (see
acestep/audio_utils.py:72 normalize_audio + generate_music_decode.py):
1. if peak > 1: audio /= peak (anti-clip)
2. audio *= target_amp / peak (target_amp = 10 ** (-1/20) ~ 0.891)
Step 2 is loudness normalization to -1 dBFS. Without it diffusers outputs
had peak=1.0 vs the real 0.891 — same music content (pearson was ~0.86
already), just 1.12x louder. Add step 2 after the existing anti-clip step.
* Match acestep/inference.py inference_steps=8 for ALL variants
GenerationParams.inference_steps default is 8 — turbo AND base/SFT. I had
base/SFT defaulting to 27 here, so every base/SFT parity run was comparing
a 27-step diffusers trajectory against an 8-step real trajectory. Different
number of denoising steps means different audio even at fixed seed.
This likely explains the lower base/SFT correlation in my earlier jieyue
runs (turbo was 0.86, base/SFT were 0.32-0.34). Aligning step counts
should bring base/SFT closer to turbo parity.
* Address PR #13095 review: rename classes + reuse diffusers primitives
Response to dg845's PR comments batch 1+2. DiT parity harness still bit-identical
(max_abs=0 on fp32 / SDPA across 4 shape cases).
Transformer file:
* Rename AceStepDiTModel -> AceStepTransformer1DModel (alias kept).
* Rename AceStepDiTLayer -> AceStepTransformerBlock (alias kept).
* Inherit AttentionMixin + CacheMixin on the DiT model.
* Swap in diffusers.models.normalization.RMSNorm for the hand-rolled
AceStepRMSNorm (weight-key-compatible).
* Swap the hand-rolled rotary embedding + apply_rotary for diffusers'
get_1d_rotary_pos_embed + apply_rotary_emb (use_real_unbind_dim=-2 to
match the cat-half convention ACE-Step inherits from Qwen3).
* Use get_timestep_embedding with flip_sin_to_cos=True — keeps the
(cos, sin) ordering of the original sinusoidal. State-dict-compatible.
* Drop max_position_embeddings arg from DiT config (RoPE computes freqs
per call based on seq_len); converter drops it.
* Gradient-checkpoint call now takes just the layer module (matches the
Flux2 idiom).
Pipeline modeling file (pipelines/ace_step/modeling_ace_step.py):
* Moved _pack_sequences + AceStepEncoderLayer here — they aren't used
by the DiT, so they shouldn't live in the transformer file.
* AceStepLyricEncoder + AceStepTimbreEncoder set
_supports_gradient_checkpointing = True and wrap encoder-layer calls
through the checkpointing func when enabled.
* Use diffusers RMSNorm + the RoPE helper from the transformer file
(shared single implementation).
Converter (scripts/convert_ace_step_to_diffusers.py):
* model_index.json now carries AceStepTransformer1DModel.
* Drop max_position_embeddings / use_sliding_window from the emitted
configs.
No numerical regressions: scripts/dit_parity_test.py PASSES with
max_abs=0.0 on fp32/SDPA across short, long, batched, and
padding-path shape variants.
* Address PR #13095 review: pipeline polish + converter HF-hub support
Response to dg845 review comments on the pipeline side. DiT parity still
bit-identical (max_abs=0 across 4 shape cases).
Pipeline (pipelines/ace_step/pipeline_ace_step.py):
* Add `sample_rate` + `latents_per_second` properties sourced from the
VAE config so the pipeline no longer hardcodes 48000 / 25 / 1920.
Propagates through prepare_latents, chunk_mask window math, and the
audio-duration round-trip.
* Add `do_classifier_free_guidance` property (matches LTX2 et al.).
* Add `check_inputs(...)` called from `__call__` before allocating noise.
Validates prompt type, lyrics type, task_type, step count, guidance
scale, shift, cfg interval bounds and repaint window ordering.
* Add `callback_on_step_end` + `callback_on_step_end_tensor_inputs` —
the modern callback form. The legacy `callback` / `callback_steps`
pair is kept for back-compat. Setting `pipe._interrupt = True` inside
the callback stops the loop early.
* Expose `encode_audio(audio)` as a public helper that wraps the tiled
VAE encode + (B, T, D) transpose the pipeline performs internally.
Converter (scripts/convert_ace_step_to_diffusers.py):
* Accept a Hugging Face Hub repo id for `--checkpoint_dir`; resolves it
via `huggingface_hub.snapshot_download` when the argument isn't a
local path.
Exports:
* Register `AceStepTransformer1DModel` in the top-level __init__,
models/__init__, models/transformers/__init__, and dummy_pt_objects so
`from diffusers import AceStepTransformer1DModel` works and the
pipeline loader resolves the new class name from model_index.json.
Deferred for a follow-up (commented inline in the PR): full
`Attention + AttnProcessor + dispatch_attention_fn` refactor and
`FlowMatchEulerDiscreteScheduler` migration — both would benefit from a
dedicated parity re-run and review.
* Fix stale ACE-Step 1.0-era docs / class names in the 1.5 integration
Docs and docstrings still carried a mix of 1.0 paper title, non-existent
`ACE-Step/ACE-Step-v1-5-turbo` hub id, `shift=3.0` turbo default, and
the old `AceStepDiTModel` class name. Cleaned up to match the actual
1.5 release:
* pipelines/ace_step.md: correct citation title ("ACE-Step 1.5: Pushing
the Boundaries of Open-Source Music Generation"), correct repo
(`ace-step/ACE-Step-1.5`), new variants table with real HF ids
(`Ace-Step1.5` / `acestep-v15-base` / `acestep-v15-sft`) and their
per-variant step/CFG defaults, drop the wrong `shift=3.0` tip.
* models/ace_step_transformer.md: page renamed to
`AceStepTransformer1DModel` with a short 1.5-specific description;
`AceStepDiTModel` noted as a backwards-compat alias.
* pipeline_ace_step.py: import, docstring, `Args`, and `__init__`
annotation reference `AceStepTransformer1DModel`; example model id
now `ACE-Step/Ace-Step1.5`; `_variant_defaults` docstring and the
`__call__` variant-fallback comment no longer claim `shift=3.0` /
`27 steps` — real defaults are 8 steps / shift=1.0 across all
variants, guidance=1.0 (turbo) vs 7.0 (base+sft).
* Address PR #13095 review: VAE tiling on AutoencoderOobleck + Timesteps class
Two more deferred review threads from dg845 addressed:
* Move tiled encode/decode onto AutoencoderOobleck
(https://github.com/huggingface/diffusers/pull/13095#discussion_r2785513647).
AutoencoderOobleck now carries `use_tiling` + `tile_sample_min_length` /
`tile_sample_overlap` / `tile_latent_min_length` / `tile_latent_overlap`
attributes and private `_tiled_encode` / `_tiled_decode` methods; the
existing `encode` / `_decode` dispatch to them when tiling is enabled and
the input exceeds the threshold. `AutoencoderMixin.enable_tiling()` is
already inherited.
AceStepPipeline's private `_tiled_encode` / `_tiled_decode` and the
`use_tiled_decode` `__call__` arg are gone; `__init__` now calls
`self.vae.enable_tiling()` so the long-audio memory behaviour is preserved
by default. Users can opt out with `pipe.vae.disable_tiling()`.
Note: the VAE-side tiling concatenates encoder features (h) and samples
the posterior once, instead of the old per-tile `.sample()` calls. This
is the standard diffusers pattern; numerically differs only in the
structure of the noise across tile boundaries.
* Use the Timesteps nn.Module for the sinusoid
(https://github.com/huggingface/diffusers/pull/13095#discussion_r2785420234).
`AceStepTimestepEmbedding` wraps `Timesteps(in_channels, flip_sin_to_cos=
True, downscale_freq_shift=0)` instead of calling `get_timestep_embedding`
directly — reviewer asked for the Module form.
* Address PR #13095 review: refactor AceStepAttention to Attention + AttnProcessor
Splits the monolithic AceStepAttention into the diffusers standard
Attention + AttnProcessor layout:
- AceStepAttention (torch.nn.Module, AttentionModuleMixin) holds the
to_q/to_k/to_v/to_out projections and norm_q/norm_k RMSNorms.
- AceStepAttnProcessor2_0 runs the attention dispatch through
dispatch_attention_fn so users can pick flash / sage / native backends
via model.set_attention_backend(...) or the attention_backend context
manager.
GQA (Q has 16 heads / K,V have 8) is preserved by passing enable_gqa=True
to dispatch_attention_fn instead of repeat_interleave; fusion is disabled
(_supports_qkv_fusion = False) because Q and K,V have different output
sizes.
The converter is updated to rename the six attention sub-keys
(q_proj -> to_q, k_proj -> to_k, v_proj -> to_v, o_proj -> to_out.0,
q_norm -> norm_q, k_norm -> norm_k) on both the DiT decoder path and the
condition encoder path, since AceStepLyricEncoder / AceStepTimbreEncoder
share the same AceStepAttention class.
Addresses review comments r2785433213 and r2785450463.
* Address PR #13095 review: migrate to FlowMatchEulerDiscreteScheduler
Replace the hand-rolled flow-matching Euler loop with
`FlowMatchEulerDiscreteScheduler`. ACE-Step still computes its own shifted /
turbo sigma schedule via `_get_timestep_schedule`, but now passes it to
`scheduler.set_timesteps(sigmas=...)` and delegates the ODE step to
`scheduler.step()`. The scheduler is configured with `num_train_timesteps=1`
and `shift=1.0` so `scheduler.timesteps` stays in `[0, 1]` (the convention the
DiT was trained on) and the scheduler doesn't re-shift already-shifted sigmas.
The scheduler's appended terminal `sigma=0` reproduces the old loop's
final-step "project to x0" case exactly: `prev = x + (0 - t_curr) * v`.
Parity on jieyue (seed=42, bf16 + flash-attn, turbo text2music, 8 steps):
waveform Pearson = 0.999999
spectral Pearson = 1.000000
max |diff| = 2.5e-3 (fp32 step-math vs previous bf16 step-math)
fp32 Euler-loop A/B against the hand-rolled path: max |diff| = 3.6e-7.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* Address PR #13095 review: move DiT tests + drop stale test kwargs
- Move the DiT transformer tests out of the pipeline test file into a new
tests/models/transformers/test_models_transformer_ace_step.py that follows
the standard BaseModelTesterConfig + ModelTesterMixin scaffold (matches
test_models_transformer_longcat_audio_dit.py).
- Drop `max_position_embeddings` from the remaining AceStepDiTModel and
AceStepConditionEncoder test fixtures — neither constructor accepts that
argument anymore.
- Drop `use_sliding_window` from the same fixtures — also no longer a
constructor argument (the actual `sliding_window` int kwarg is kept).
- Wire `FlowMatchEulerDiscreteScheduler(num_train_timesteps=1, shift=1.0)`
into `get_dummy_components()` now that the pipeline requires it.
Resolves https://github.com/huggingface/diffusers/pull/13095#discussion_r3115653554,
r3115664850, r3115673059, r3115676580, r3115680700.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* Address PR #13095 review from dg845 (2026-04-23)
Fixes 5 review threads + style:
1. Converter now builds `AceStepPipeline` in memory and calls
`save_pretrained`. Previously the hand-written `model_index.json` was
missing the `scheduler` entry — fresh converter output couldn't be loaded
by `AceStepPipeline.from_pretrained` (r3127767785). This also makes the
converter robust to future `__init__` signature changes.
2. `latent_length` uses `math.ceil(...)` instead of `int(...)` so non-integer
products (e.g. `latents_per_second=2.0, audio_duration=0.4 → 0.8`) round up
to `1` instead of truncating to `0` and crashing shape checks (r3127790939).
3. Add `_callback_tensor_inputs = ["latents"]` on `AceStepPipeline` so the
standard diffusers callback tests pick up the right tensor (r3127795954).
4. `AceStepConditionEncoder.silence_latent` no longer hard-codes the channel
dim to 64. The placeholder buffer now uses the `timbre_hidden_dim`
constructor argument, so smaller test configs with `timbre_hidden_dim != 64`
load without shape errors (r3127812932).
5. Revert `self.vae.enable_tiling()` from `AceStepPipeline.__init__`. Users can
call `pipe.vae.enable_tiling()` themselves for long-form generation; that
matches the opt-in convention used by the rest of diffusers (r3127777296).
6. `ruff check --fix` + `ruff format` over all ACE-Step sources (the style fix
dg845 asked for via `@bot /style`).
Also: converter now accepts sharded `model.safetensors.index.json` layouts
alongside the single-file `model.safetensors`, so the 5B XL turbo variant
converts without a pre-processing step.
Parity on jieyue (seed=42, bf16 + flash-attn, turbo text2music 160s, fresh
converter output loaded via `from_pretrained`):
waveform Pearson = 0.999954
spectral Pearson = 0.999977
max |a-b| bf16 = 4.3e-02 (dominated by the VAE tiling default flip)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* Address PR #13095 review from yiyixuxu (2026-04-23)
Code-level (22 threads):
1. Delete 3 dev/parity scripts (`scripts/audio_parity_jieyue.py`,
`scripts/dit_parity_test.py`, `scripts/run_official_generate_music.py`)
that shouldn't have been committed.
2. Rename `AutoencoderOobleck._encode_one` → `_encode` to match the convention
used by other diffusers VAEs.
3. Delete the hard-coded `SHIFT_TIMESTEPS` / `VALID_SHIFTS` table in
`pipeline_ace_step.py`: the per-shift turbo schedules are recovered
exactly by `linspace(1, 0, N+1)[:-1]` plus the flow-match shift formula
that the non-turbo branch already uses, so a single code path covers both.
4. Drop the backwards-compat `AceStepDiTModel` / `AceStepDiTLayer` aliases
and every reference (top-level `__init__`, `models/__init__`,
`transformers/__init__`, dummy objects, tests, docs toctree, model card).
`AceStepTransformer1DModel` is the only exported name now.
5. Remove the unused `attention_mask` / `encoder_attention_mask` args from
`AceStepTransformer1DModel.forward`; the model rebuilds its masks from
the sequence shape and never consumed them.
6. In the DiT forward and both encoders, pass `None` instead of an all-zero
`full_attn_mask` / `encoder_4d_mask` to non-sliding attention layers — SDPA
dispatches to a faster kernel when the mask is None.
7. Inline the shared `_run_encoder_layers` helper directly into
`AceStepLyricEncoder.forward` / `AceStepTimbreEncoder.forward` so layer
calls are visible at the forward boundary (diffusers style).
8. Move `is_turbo` / `sample_rate` / `latents_per_second` from `@property`s
that re-read module configs each call to cached attributes populated in
`__init__` (Flux2-style), with a default-ACE-Step fallback when
`self.vae` is offloaded. Drop the now-unused `SAMPLE_RATE = 48000`
module-level constant and the three property definitions.
9. Warn + coerce `guidance_scale` to 1.0 on turbo (guidance-distilled)
checkpoints, following `pipeline_flux2_klein`. Prevents over-guided
audio when users forward their base/sft CFG settings to a turbo pipe.
10. Remove the `logger.warning(...)` paths that triggered on
`silence_latent` missing/zero — those only fired for author-side
unconverted checkpoints and tests; end users always load converted
weights where the buffer is baked in.
11. Drop the redundant `with torch.no_grad():` wrappers inside
`encode_prompt` — the pipeline's `__call__` runs under `torch.no_grad`
already.
12. Strip "reviewer comment on PR #13095" attribution comments from three
docstrings (here and everywhere).
Parity on jieyue (seed=42, bf16 + flash-attn, XL turbo 160s text2music):
waveform Pearson = 0.9747
spectral Pearson = 0.9895
The shift comes from full-attention layers switching `attn_mask=0_tensor` →
`attn_mask=None`, which dispatches to a different SDPA kernel on bf16. The
two outputs are algebraically equivalent for fp32 eager; on bf16+FA the
delta is dominated by kernel-level ULPs, well within the sampler-noise
band (ear-check on the 160s example confirms no audible regression).
Still open — AudioTokenizer/Detokenizer (deferred) + APG guider follow-up
(dims differ from `diffusers.guiders.adaptive_projected_guidance`, not a
drop-in; worth a separate PR).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* Address ACE-Step audio token and APG review
* Fix ACE-Step docs CI
* Address ACE-Step pipeline cleanup review
* Fix ACE-Step flash attention sliding windows
* Add ACE-Step callback properties
* Address ACE-Step final review comments
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com>
|
||
Álvaro Somoza
|
303c1d8b04 |
[Ernie-Image] Add lora support (#13575)
add lora support Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> Co-authored-by: Dhruv Nair <dhruv.nair@gmail.com> |
||
|
Sayak Paul
|
a5bc04696b |
NucleusMoE docs (#13661)
up |
||
songh11
|
50cb2db4ad |
feat: support ring attention with arbitrary KV sequence lengths (#13545)
* feat: support ring attention with arbitrary KV sequence lengths * fix: align ring_anything with ulysses_anything (size gather + unshard) * docs: document ring_anything mode * fix: merge hook branches, add ring_anything comment + guard * docs: address ring_anything review comments * docs: update ring_anything guidance * docs: refine ring_anything guidance per review * fix: address ring_anything style check --------- Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> |
||
Alexander Ivanov
|
2173c554ea | [docs] fix typo in AutoencoderOobleck docs (#13642) (#13645) | ||
|
Sayak Paul
|
656da843e8 |
[docs] add a mention of torchao and other backends in speed memory docs. (#13499)
add a mention of torchao and other backends in speed memory docs. |
||
|
Lancer
|
947bc23ba4 |
[chore] Add diffusers-format example to LongCatAudioDiTPipeline (#13483)
* [chore] Add diffusers-format example and seed parameter to LongCatAudioDiTPipeline Signed-off-by: Lancer <maruixiang6688@gmail.com> * Apply style fixes * Apply suggestions from code review Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * Apply style fixes --------- Signed-off-by: Lancer <maruixiang6688@gmail.com> Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> |
||
Lancer
|
c41a3c3ed8 |
[Feat] Adds LongCat-AudioDiT pipeline (#13390)
* Add LongCat-AudioDiT pipeline Signed-off-by: Lancer <maruixiang6688@gmail.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * upd * Apply suggestions from code review Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * Apply style fixes * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * Apply style fixes * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> * Apply style fixes * upd Signed-off-by: Lancer <maruixiang6688@gmail.com> --------- Signed-off-by: Lancer <maruixiang6688@gmail.com> Co-authored-by: dg845 <58458699+dg845@users.noreply.github.com> Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> |
||
HsiaWinter
|
dc8d903217 |
Add ernie image (#13432)
* Add ERNIE-Image * Update doc * Update doc * Change from Custom-Attention to Diffusers Style Attention * Change from Custom-Attention to Diffusers Style Attention * 兼容SGLang * 优化PE模块的加载与offload策略 * 更新Doc文件与config配置相关内容 * Fix官方反馈的内容 * 根据官方建议优化代码 * Update code * update * update * Apply style fixes * update * update * Apply style fixes --------- Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> |
||
Xyc2016
|
251676dfda |
Fix grammar in LoRA documentation (#13423)
Fix grammar in LoRA documentation (LoRA's → LoRAs, trigger it → trigger them) |
||
Samuel Meddin
|
3e53a383e1 |
Fix typos and grammar errors in documentation (#13391)
- Fix 'allows to generate' -> 'allows you to generate' in controlling_generation.md - Fix 'it's refiner' -> 'its refiner' (possessive) in sdxl.md - Fix 'it's state' -> 'its state' (possessive) in reusing_seeds.md - Fix missing word 'you'll a function' -> 'you'll create a function' in sdxl.md |
||
|
YiYi Xu
|
cf6af6b4f8 |
[docs] add auto docstring and parameter templates documentation for m… (#13382)
* [docs] add auto docstring and parameter templates documentation for modular diffusers Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/modular_diffusers/auto_docstring.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/_toctree.yml Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * up --------- Co-authored-by: yiyi@huggingface.co <yiyi@ip-26-0-161-123.ec2.internal> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com> Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> |
||
|
Steven Liu
|
e365d749a1 |
[docs] deprecate pipelines (#13157)
* deprecate * fix * fix * fix * fix * remove deprecated .md files * update links * fix |
||
Pranav Thombre
|
7e463ea4cc |
[docs] Add NeMo Automodel training guide (#13306)
* [docs] Add NeMo Automodel training guide Signed-off-by: Pranav Prashant Thombre <pthombre@nvidia.com> * Update docs/source/en/training/nemo_automodel.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/training/nemo_automodel.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * adding contacts into the readme * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Apply suggestion from @stevhliu Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Address CR comments Signed-off-by: Pranav Prashant Thombre <pthombre@nvidia.com> * Update docs/source/en/training/nemo_automodel.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Update docs/source/en/training/nemo_automodel.md Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> --------- Signed-off-by: Pranav Prashant Thombre <pthombre@nvidia.com> Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> Co-authored-by: linnan wang <wangnan318@gmail.com> |
||
Howard Zhang
|
1fe2125802 |
remove str option for quantization config in torchao (#13291)
* remove str option for quantization config in torchao * Apply style fixes * minor fixes * Added AOBaseConfig docs to torchao.md * minor fixes for removing str option torchao * minor change to add back int and uint check * minor fixes * minor fixes to tests * Update tests/quantization/torchao/test_torchao.py Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> * Update docs/source/en/quantization/torchao.md Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> * Update tests/quantization/torchao/test_torchao.py Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> * version=2 update to test_torchao.py --------- Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: Sayak Paul <spsayakpaul@gmail.com> |