Wenjia LiandWeiQing Chen d9980c75e8 Perf/minimax h3 dit swiglu rope cache npu (#6410) ...

Signed-off-by: Wenjia Li <wjialish@gmail.com>
Co-authored-by: WeiQing Chen <40507679+david6666666@users.noreply.github.com>

2026-08-29 09:14:52 +00:00

..

attention

[Feat] Minimax H3 support ref and tgt video sparse attention on NPU (#6518)

2026-08-29 11:10:53 +08:00

cache

Perf/minimax h3 dit swiglu rope cache npu (#6410)

2026-08-29 09:14:52 +00:00

diffusion_kv

[Diffusion] Add native paged KV backend for diffusion workers (#6102)

2026-08-22 21:45:45 +08:00

distributed

[Core] Restore topology-aware diffusion worker titles (#6722)

2026-08-29 08:45:57 +08:00

executor

[Core] Add UniProc diffusion executor for single-GPU (#6308)

2026-08-22 21:58:21 +08:00

hooks

[Refactor][Misc] Remove unused code in diffusion hooks/model_loader (#5270)

2026-07-22 09:45:17 +08:00

layers

Perf/minimax h3 dit swiglu rope cache npu (#6410)

2026-08-29 09:14:52 +00:00

lora

[Feature] Support MiniMax-H3 Turbo LoRA with DLO (#6550)

2026-08-26 16:15:47 +08:00

model_loader

[Diffusion] Register HWR mmap for direct DLO H2D (#6591)

2026-08-25 23:30:11 +08:00

models

Perf/minimax h3 dit swiglu rope cache npu (#6410)

2026-08-29 09:14:52 +00:00

offloader

[Diffusion] Register HWR mmap for direct DLO H2D (#6591)

2026-08-25 23:30:11 +08:00

postprocess

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

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

profiler

[Bugfix] Report stepwise Qwen-Image profiler metrics (#4517)

2026-07-03 04:32:22 +00:00

quantization

[Bugfix] Fix HSDP compatibility with the new online FP8 linear method (#5677)

2026-08-14 15:45:55 +08:00

sched

text and image metric (#6150)

2026-08-26 02:54:45 +00:00

utils

[Model][Frontend] MiniMax-H3: Parallelize MP4 response frame conversion (#6499)

2026-08-28 21:22:00 -06:00

worker

[Core] Restore topology-aware diffusion worker titles (#6722)

2026-08-29 08:45:57 +08:00

__init__.py

[Diffusion] Qwen image support (#82)

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

compile.py

fix: keep DLO hooks outside regional compilation (#6073)

2026-08-22 21:44:03 +08:00

config.py

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

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

data.py

[Feature] Support text encoder Disaggregation for H3 (#5885)

2026-08-28 14:27:05 +08:00

diffusion_engine.py

text and image metric (#6150)

2026-08-26 02:54:45 +00:00

envs.py

[Refactor] Extract has_flash_attn_pkg in OmniPlatform (#3068)

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

forward_context.py

[Feature][MiniMax-H3] Support diffusion continuous batching (#5810)

2026-08-24 21:08:53 +08:00

inline_stage_diffusion_client.py

[Feature] Support text encoder Disaggregation for H3 (#5885)

2026-08-28 14:27:05 +08:00

io_support.py

[Feature] Support text encoder Disaggregation for H3 (#5885)

2026-08-28 14:27:05 +08:00

ipc.py

[CI]Remove DreamID-Omni and MagiHuman support (#6357)

2026-08-26 01:31:12 +00:00

model_metadata.py

[Feature] Support text encoder Disaggregation for H3 (#5885)

2026-08-28 14:27:05 +08:00

output_formatter.py

[Feature] Support text encoder Disaggregation for H3 (#5885)

2026-08-28 14:27:05 +08:00

prompt_update.py

[Frontend] Add midway prompt update for streaming video generation (#4652)

2026-07-27 14:07:43 +08:00

registry.py

[CI]Remove DreamID-Omni and MagiHuman support (#6357)

2026-08-26 01:31:12 +00:00

request.py

text and image metric (#6150)

2026-08-26 02:54:45 +00:00

stage_diffusion_client.py

[Feature] Support text encoder Disaggregation for H3 (#5885)

2026-08-28 14:27:05 +08:00

stage_diffusion_proc.py

[Rebase] Rebase to vllm 0.28.0 (#6606)

2026-08-27 06:12:55 +00:00

vllm_config.py

[Diffusion] Add native paged KV backend for diffusion workers (#6102)

2026-08-22 21:45:45 +08:00