MaciejBalaNVandHongsheng Liu 6dc4638a7e Fixed Cosmos3 regression (#4627) ...

Signed-off-by: Maciej Bala <mbala@nvidia.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>

2026-06-24 09:03:34 +08:00

..

attention

[XPU] Add sage_attn backend (#3785)

2026-06-09 19:28:18 +08:00

cache

[Cache Refactor 1/N] Simplify CacheDiT Integration (#2527)

2026-06-20 08:30:44 +08:00

distributed

[Bugfix] Fix HSDP + FP8 online quantization compatibility (#4494)

2026-06-19 15:06:27 +08:00

executor

[Bugfix] Surface all-rank diffusion RPC failures (#4403)

2026-06-13 15:42:16 +08:00

hooks

[Feat] Support MagCache (#1287)

2026-05-29 22:26:23 +08:00

layers

[Enhancement] Introduce High-Performance MoT (Mixture-of-Tokens) Kernels: Triton Implementation (#3960)

2026-06-16 03:08:02 +00:00

lora

[Feature] LoRA support for SenseNova-U1 (#3971)

2026-06-09 14:36:35 +00:00

model_loader

[BugFix] Use ModuleDiscovery for HSDP DiT sharding to support nested pipelines (#3456)

2026-06-23 02:54:14 +00:00

models

Fixed Cosmos3 regression (#4627)

2026-06-24 09:03:34 +08:00

offloader

[Refactor] Migrate existing pipelines to use SupportsComponentDiscovery for offload discovery (#3076)

2026-06-17 21:33:06 +08:00

postprocess

[Bugfix] Fix RIFE device selection for CPU-transported videos (#2876)

2026-04-17 19:19:34 +08:00

profiler

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00

quantization

[Bugfix] Fix HSDP + FP8 online quantization compatibility (#4494)

2026-06-19 15:06:27 +08:00

sched

[Diffusion] Unify diffusion request identity on request_id (#3744)

2026-05-28 14:56:56 +08:00

utils

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00

worker

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00

__init__.py

[Diffusion] Qwen image support (#82)

2025-11-28 14:22:11 +08:00

compile.py

[Perf]Enable regional compile for Cosmos3 HSDP blocks (#4485)

2026-06-18 16:54:05 +08:00

config.py

[Diffusion][Attention] Support per-role attention backend via CLI (#2681)

2026-05-10 11:59:05 +08:00

data.py

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00

diffusion_engine.py

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00

envs.py

[Refactor] Extract has_flash_attn_pkg in OmniPlatform (#3068)

2026-04-27 04:07:34 +00:00

forward_context.py

[Model] Add Ming-flash-omni-2.0 Image Generation (Diffusion) Stage (#2875)

2026-05-27 23:21:20 +08:00

inline_stage_diffusion_client.py

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00

io_support.py

[Refactor]Base class for dit pipelines with unified parameter declaration[1/N] (#4225)

2026-06-18 17:14:18 +08:00

ipc.py

[Bugfix] Surface all-rank diffusion RPC failures (#4403)

2026-06-13 15:42:16 +08:00

model_metadata.py

[Feature] HunyuanImage-3.0 IT2I: multi-image input + prompt API cleanup (#3444)

2026-05-14 16:58:40 +08:00

output_formatter.py

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00

registry.py

[Diffusion] add GR00T-N1.7 pipeline with OpenPI serving (#3798)

2026-06-20 11:45:26 +08:00

request.py

[Diffusion] Unify diffusion request identity on request_id (#3744)

2026-05-28 14:56:56 +08:00

stage_diffusion_client.py

[BugFix]Fix diffusion profile RPC None results (#4372)

2026-06-21 17:14:21 +08:00

stage_diffusion_proc.py

[feat] Streaming diffusion video generation output (#3737)

2026-06-23 09:57:53 +00:00