GitLab

* feature: qwen-image support cpu offload (block) and refactor transfomrer model

* bugfix

* update seko file names

* update

* [Update] update seko-models configs

* Add seko scripts

---------
Co-authored-by: gushiqiao <975033167@qq.com>

* [Feat] Support vae bf16 and add 4090 best config

* [Feat] Support vae bf16 and add 4090 best config

* [Feat] Support vae bf16 and add 4090 best config

* [Feat] Support vae bf16 and add 4090 best config

* update dtype

---------
Co-authored-by: helloyongyang <yongyang1030@163.com>

* update docs

* update docs

* [Feat] audio driven for wan2.2-r2v-5b

* refactor: reduce vae args dim &&  timestep init

* update audio pre_infer

* update runner

* update scripts

* [Fix] move ref timestep padding to Scheduler

* fix wan i2v

---------
Co-authored-by: helloyongyang <yongyang1030@163.com>

* [Fix] Fix audio model offload bug.

* Fix

* Fix

* Fix

* Fix

* Fix

* Fix

* bugfix:adapt to  5B dit model, derive attention_head_dim from config[dim]

* [Fix] Wan2.2 Vae Encode refactor: drop args parameter and use self.cpu_offload

* [Fix] fix bug

* Fix bug

* Update run_wan_vace.sh

* feature: support qwen-image-edit(i2i)

* bugfix

* [Paper] add vace paper

* [Paper] add vace paper

* Delete docs/PAPERS_ZH_CN/source/papers/vace.md

* Update index.rst

[fea] Update patch vae

* Add test cases

* Add offload script

---------
Co-authored-by: gushiqiao <gushiqiao@sensetime.com>

* Add magcache

* fix

* lint

* fix

* update script

* Support cfg infer

* support cfg parallel

* update cfgsp/t2v config

* support magcache calibration

* calibration

Add flf2v model

* feature: support qwen-image t2i

* pass ci

---------
Co-authored-by: wushuo1 <wushuo1@sensetime.com>