Isotr0py 3affc7766d Enable VAE tiling for Qwen-Image-2.1 (#5) ...

* init

Signed-off-by: Isotr0py <mozf@inferact.ai>

* fix

Signed-off-by: Isotr0py <mozf@inferact.ai>

* test(diffusion): add Qwen-Image-2.1 unit tests

Signed-off-by: Isotr0py <mozf@inferact.ai>

* fix(diffusion): Qwen-Image-2.1 VAE spatial ratio, tiling e2e verification

- AutoencoderKLQwenImage21: derive spatial_compression_ratio from the
  architecture (2 ** len(temperal_downsample) = 16x) instead of trusting
  the checkpoint's scale_factor_spatial=8, which corrupted tiled
  decode/encode geometry (output cropped to half size) and tiling gating.
- pipeline: _read_vae_scale_factor prefers the architecture-derived value
  for the same reason.
- models/qwen_image_21/__init__: lazy exports to break the circular import
  distributed VAE -> package __init__ -> pipeline -> distributed VAE.
- Add VAE unit tests: ratio derivation, tiled vs untiled decode/encode
  consistency, distributed tile_split/exec/merge == tiled_decode.
- e2e matrix (GB200, seed 42, 20 steps) all green; see e2e_out/qwen21_vae/.
- examples: qwen_image_21_edit.py gains --vae-use-tiling.
- docs: mark VAE patch-parallel verified for Qwen-Image-2.1; recipe notes
  the default-tile banding limitation at 2048px+.

* clean tests

Signed-off-by: Isotr0py <mozf@inferact.ai>

* fix cudnn determination

Signed-off-by: Isotr0py <mozf@inferact.ai>

---------

Signed-off-by: Isotr0py <mozf@inferact.ai>

2026-09-18 10:02:50 +08:00

..

attention

[Rebase] Rebase to vllm 0.26.0 (#5443)

2026-07-27 23:35:49 +08:00

benchmarks

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

clients

[Core] Graduate MiniCPM-o 4.5 and PersonaPlex full-duplex serving out of experimental (#6196)

2026-09-05 08:20:21 +00:00

config

[Refactor] P0.2: Migrate API server helpers out of api_server (#5453)

2026-09-10 15:03:21 +08:00

core

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

deploy

[Model] Optimize MOSS-TTS Local batched execution and streaming codec (#7202)

2026-09-09 13:02:12 -07:00

diffusion

Enable VAE tiling for Qwen-Image-2.1 (#5)

2026-09-18 10:02:50 +08:00

distributed

[Core] Split Omni connector model runner mixin (#6903)

2026-09-09 13:49:43 +08:00

engine

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

entrypoints

[Refactor] P0.2: Migrate API server helpers out of api_server (#5453)

2026-09-10 15:03:21 +08:00

experimental

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

host_weight_runtime

[Bugfix] Skip HWR store-size scans when no limit is configured (#7131)

2026-09-08 22:54:49 +08:00

inputs

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

lora

[Feature] Diffusion LoRA Adapter Support (PEFT compatible) for vLLM alignment (#758)

2026-01-27 10:22:40 +08:00

metrics

[Model] Optimize MOSS-TTS Local batched execution and streaming codec (#7202)

2026-09-09 13:02:12 -07:00

model_executor

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

model_extras

[Model] Add native MAGI-2 Preview diffusion support (#5918)

2026-09-05 15:32:30 +08:00

outputs

[Model] Optimize MOSS-TTS Local batched execution and streaming codec (#7202)

2026-09-09 13:02:12 -07:00

platforms

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

plugins

[Quantization] Migrate GGUF diffusion model support to OOT plugin (#4769)

2026-07-11 18:48:43 +00:00

profiler

[Enhancement]modify profiling.md (#3051)

2026-04-24 16:30:24 +08:00

quantization

[Model] FP8 Support For Boogu-Image (#6925)

2026-09-06 04:48:38 +00:00

reasoning

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

sample

update dirs

2025-09-11 15:43:43 +08:00

tokenizers

[Fix] Restore MammothModa2 support (#6694)

2026-08-28 08:32:32 +08:00

transformers_utils

[Bugfix][MiniCPM-o] Align structured chat content with native omni rendering (#7344)

2026-09-10 11:38:11 +08:00

utils

[Bugfix] Build the forced-aligner prompt without a chat template (word timestamps one bin late) (#7240)

2026-09-08 19:27:38 +00:00

worker

[Rebase] Rebase to vLLM 0.29.0 (#7230)

2026-09-10 12:08:28 +08:00

__init__.py

[Rebase] Rebase to vllm 0.26.0 (#5443)

2026-07-27 23:35:49 +08:00

data_entry_keys.py

[Perf][NemotronVoiceChat] Fast defaults for all three deploy profiles: 2.8x offline, realtime full-duplex audio (#6354)

2026-09-03 18:49:19 -07:00

errors.py

[refactor] Refactor guardrail error handling - add 400 error code (#4297)

2026-06-11 21:09:24 +00:00

logger.py

[BugFix] Failed to print DEBUG level log (#307)

2025-12-15 07:13:36 +00:00

patch.py

[Rebase] Rebase to vllm 0.27.0 (#5976)

2026-08-11 15:23:46 +00:00

request.py

[Rebase] Rebase to vllm 0.28.0 (#6606)

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

version.py

[Misc] Warn When vLLM / vLLM-Omni Have Mismatched Versions (#2691)

2026-04-17 20:15:16 +08:00