Gong Chao comments

Results 14 comments of


                                            Gong Chao

i found it hard to implement this optimizer on yolov5.looking forward to s.b. could do me a FAVOR. THX

hey, you can check here https://github.com/davda54/sam/issues/7

Prompt details about evaluating on EditBench

And there seems no refined EditBench in the link you provide. Looking forward to your reply!

Prompt details about evaluating on EditBench

Thanks for your reply! I am confused why not use prompt_mask, the local description for inpainting area? It seems that using local description is much more common in real inpainting...

Finetune SDXL

Sorry to bother everyone. The loss non-convergence problem confuses me a lot. Does anyone solve it? @huiyang865 @windson87 @RunpuWei @dydxdt @onewangqianqian

testing the editing performance directly using stage 1's ckpt

Thanks for your reply! I see that the BIM's great role. There are 2 traing stage for SmartEdit, stage 1 for MLLM aligning with the CLIP text encoder and state...

testing the editing performance directly using stage 1's ckpt

Gotcha！

Web Demo 502 Bad Gateway

Also, [openxlab demo](https://openxlab.org.cn/apps/detail/SunzeY/Alpha-CLIP_with_LLM) is broken too.

Doesn't converge when I train with my own data

Me too. It confuses me a lot. Have you solved it? @zf-666 Or could the authors help us please? @juxuan27

Why not shuffling training dataset

> > Hi, such an excellect work! I am reading your code and find you don't use shuffle when training [here](https://github.com/TencentARC/BrushNet/blob/101dc3d78ffb9e6acc5d89a656b32d635babc65f/examples/brushnet/train_brushnet.py#L1125). I remember that not using shuffle will damage the...