can u plz make it to ckpt that can used in auto1
Really gooood job,but i wanna kown how to make whole of it to SD ckpt,that i can use it in the auto1 webui.......tks
I use the
"https://github.com/huggingface/diffusers/tree/main/scripts/convert_diffusers_to_original_stable_diffusion.py"
such like this
"python convert_diffusers_to_original_stable_diffusion.py --model_path \Taiyi-Stable-Diffusion-1B-Chinese-v0.1 --checkpoint_path \Taiyi-Stable-Diffusion-1B-Chinese-v0.1\model.ckpt"
And I get a ckpt, but when I using it in my auto111 webui,I get a error like this:
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for LatentDiffusion:
size mismatch for cond_stage_model.transformer.text_model.embeddings.position_ids: copying a param with shape torch.Size([1, 512]) from checkpoint, the shape in current model is torch.Size([1, 77]).
I can know a little reason about it,but I dont's know how to deal it...
It looks like the text encoder seq length is not match, I will try it.
I use the
"https://github.com/huggingface/diffusers/tree/main/scripts/convert_diffusers_to_original_stable_diffusion.py"
such like this
"python convert_diffusers_to_original_stable_diffusion.py --model_path \Taiyi-Stable-Diffusion-1B-Chinese-v0.1 --checkpoint_path \Taiyi-Stable-Diffusion-1B-Chinese-v0.1\model.ckpt"
And I get a ckpt, but when I using it in my auto111 webui,I get a error like this:raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for LatentDiffusion:
size mismatch for cond_stage_model.transformer.text_model.embeddings.position_ids: copying a param with shape torch.Size([1, 512]) from checkpoint, the shape in current model is torch.Size([1, 77]).I can know a little reason about it,but I dont's know how to deal it...
The text model is a roberta, not the same as CLIPTextModel. So you can not convert it directly.
Maybe a fast way to deal with it is to change the way of model loading of the WebUI. We will try it later, It would be good it you can try it and contribute to the open source community.
@wf-genius any good news for WebUI,thanks
同问：如何应用在AUTOMATIC1111 webui上？谢谢
@jkyeah @lastpanda @Alfred-Duncan you can refer to https://github.com/IDEA-CCNL/Fengshenbang-LM/issues/186
太好了，非常感谢