GitLab

* add step & cfg distillation wan model

* bug fixed

* Implement distributed inference API server with FastAPI

- Added a new file `api_server_dist.py` to handle video generation tasks using distributed inference.
- Introduced endpoints for task submission, status checking, and result retrieval.
- Implemented image downloading and task management with error handling.
- Enhanced `infer.py` to ensure proper initialization of distributed processes.
- Created a shell script `start_api_with_dist_inference.sh` for easy server startup with environment setup.

This commit establishes a robust framework for managing video generation tasks in a distributed manner.

* Enhance file download endpoint with path traversal protection and error handling

* style: format

* refactor: Enhance video generation functionality with task interruption support

* feat: Add image upload and video generation endpoint with unique task handling

- Introduced a new endpoint `/v1/local/video/generate_form` for video generation that accepts image uploads.
- Implemented unique filename generation for uploaded images to prevent conflicts.
- Enhanced directory management for input and output paths.
- Improved file download response with detailed status and size information.
- Added error handling for distributed inference processes and graceful shutdown procedures.

* split dit server from default runner

* split dit server from default runner

* update loading functions

* simplify loader functions and runner functions

* simplify code && split dit service

* simplify code && split dit service

* support split server for cogvideox

* clear code.

Co-authored-by: “de1star” <“843414674@qq.com”>

* add load_transformer methods for split server

* add service utils

* [feature] support split servers

* [bugs fixed] fixed bugs for cpu offload.

* [rename] rename causal_model -> causvid_model

* [feature] add prompt enhancer

* [feature] add prompt enhancer

* [rename] rename causal_model -> causvid_model

* [feature]: add support for Causvid inference

* [feature]: support long video generation for autoregressive wan2.1-t2v

* [feature]: support long video generation for autoregressive wan2.1-t2v

* Update wan_t2v_causal.json

* Update run_wan_t2v_causal.sh

* update readme for Wan2.1-T2V-CausVid

---------
Co-authored-by: Yang Yong(雍洋) <yongyang1030@163.com>

* Support hunyuan offload and teacache.

* Fix

* Fix

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

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

* ring attn & comm

* ring-attn

* ring attn

* ring attn & unit test

* ring attn

* pre-commit reformat

* Update run_wan_i2v.sh

* Update attn.py

---------
Co-authored-by: “huangxinchi” <“huangxinchi@sensetime.com”>
Co-authored-by: “de1star” <“843414674@qq.com”>
Co-authored-by: gushiqiao <gushiqiao@sensetime.com>
Co-authored-by: Yang Yong(雍洋) <yongyang1030@163.com>

* [minor]: optimize dockerfile for fewer layer

* [feature]: add pre-commit lint, update readme for contribution guidance

* [minor]: fix run shell privileges

* [auto]: first lint without rule F, fix rule E

* [minor]: fix docker file error

* 修改了main.py, t5的model, wan的model、三个weights文件和三个infer文件, 并且在common的conv3d算子中注册新算子

* 修改了Conv3dWeightForceBF16算子，更新了wan的pre_weights中对此算子的使用

* 修复了import中的bug

* 修复了WanPreWeights, WanTransformerWeights没有self.config的bug

* 修复了WanPreWeights, WanTransformerWeights没有self.config的bug

* 修复了config的bug，目前在使用cpu_offload的时候，vae阶段有tensor不在同一device的bug

* 修复了vae阶段迁移的bug

* 修复了scale在mean和inv_std迁移后仍需重新赋值的bug

* parallel attention

* Update run_hunyuan_t2v.sh

* Update main.py

---------
Co-authored-by: “de1star” <“843414674@qq.com”>
Co-authored-by: Yang Yong <yongyang1030@163.com>