Add AnyFlow Any-Step Video Diffusion Pipelines (Bidirectional + FAR Causal)#13745

Open

Enderfga wants to merge 17 commits into

huggingface:mainfrom

Enderfga:add-anyflow-pipeline

Commits on May 6, 2026

[Pipelines] AnyFlow: scaffold pipelines/anyflow + register all top-level imports
Enderfga
committed
[Schedulers] AnyFlow: add FlowMapEulerDiscreteScheduler
Enderfga
committed
[Models] AnyFlow: add AnyFlowTransformer3DModel
Enderfga
committed
[Pipelines] AnyFlow: add AnyFlowPipeline and AnyFlowCausalPipeline
Enderfga
committed
[Docs] AnyFlow: add main pipeline documentation page
Enderfga
committed
[Auto/Scripts] AnyFlow: register AutoPipelineForText2Video + add conversion script
Enderfga
committed
[Quality] AnyFlow: ruff-format + regenerated dummy stubs
Enderfga
committed
[AnyFlow] address review feedback: bug fixes + DMD wording + EN/ZH tutorials
Enderfga
committed
[AnyFlow] rename Causal->FAR + explicit forward signature + dataclass output
Enderfga
committed
[AnyFlow] wire callback_on_step_end through inference_range + add chunk_partition to FAR fast-test fixture
Enderfga
committed
[AnyFlow] Phase 2: split transformer + drop chunk_partition from config + rename helpers
Enderfga
committed
[AnyFlow] Phase 3: convention compliance against .ai/AGENTS.md + .ai/models.md
Enderfga
committed
[AnyFlow] FAR fast-test compat: rope 0-dim guard + flex_attention CPU/head-dim fallbacks + KV-cache dtype + num_timesteps
Enderfga
committed