GitLab

[feat]: support qwen-image-edit-2509

This new script facilitates the merging of R2V and distillation models
via LoRA, and converts the merged model and audio adapter to multiple
precision formats (FP32, BF16, FP8). It includes detailed logging and
error handling for each step of the conversion pipeline.

Co-authored-by: gaclove <peng.gaoc@gmail.com>

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

[feat]: refactor to adapt new config system and align precision for qwen-image & qwen-image-edit (#383)

[feat]: refactor to adapt new config system and align precision for
qwen-image & qwen-image-edit

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

Co-authored-by: XHPlus <xhplus@163.com>

Co-authored-by: root <root@pt-731f15f5c62740f495519352dc90d9c4-worker-0.pt-731f15f5c62740f495519352dc90d9c4.ns-devsft-3460edd0.svc.cluster.local>

…config

主要用于步数蒸馏实验
- 支持在 highnoise / lownoise 文件夹下指定特定的checkpoint用于加载
- 删除冗余的 Wan22MoeDistill 类
- 修改同时发起多个请求时，请求中没有 seed 参数导致的 bug

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

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

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

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

1、frontend vue+vite
2、share task & template
3、x264 rtc stream push