How to download checkpoint sdxlUnstableDiffusers_v8HeavensWrathVAE in infer_style.py

[huiyang865]

No description provided.

[haofanwang]

It's on Civitai, you can convert it to diffusers via their scripts.

[huiyang865]

Thanks for replying, where is their script, I can't seem to find it in their github (https://github.com/civitai/civitai).

[vessenes]

This took some digging. the specified model is a safetensors sdxl model; once downloaded, I used https://github.com/huggingface/transformers/scripts/convert_original_stable_diffusion_to_diffusers.py to dump out into the (long filename) as a directory. The script then wanted a single binary file, so I merged them. This worked, although there is also an error loading the tile directory - it tries to load from "./" and it should be "./checkpoints/".

[long-yy]

It's on Civitai, you can convert it to diffusers via their scripts.

Can you provide this weight? Or can we use stabilizability/table diffusion-xl-base-1.0 instead?

[HuanBor]

Run the code here https://github.com/huggingface/diffusers/blob/main/scripts/convert_original_stable_diffusion_to_diffusers.py

with the command of python convert_original_stable_diffusion_to_diffusers.py --checkpoint_path path/to/sdxlUnstableDiffusers_v8HeavensWrathVAE.safetensors --dump_path /InstantStyle-Plus/checkpoints/sdxlUnstableDiffusers_v8HeavensWrathVAE --image_size 512 --prediction_type epsilon --from_safetensors --device cuda:0 --half

[zhangtianjia]

Thanks for your wonderful work. Could you provide the downloadable weights for sdxlUnstableDiffusers_v8HeavensWrathVAE.
When I run the code, it occurs the error message:
Error no file named model.safetensors found in directory ./checkpoints/sdxlUnstableDiffusers_v8HeavensWrathVAE/text_encoder.
I cannot find model.safetensors in the text_encoder directory.

[cavalier501]

Thanks for your wonderful work. Could you provide the downloadable weights for sdxlUnstableDiffusers_v8HeavensWrathVAE. When I run the code, it occurs the error message: Error no file named model.safetensors found in directory ./checkpoints/sdxlUnstableDiffusers_v8HeavensWrathVAE/text_encoder. I cannot find model.safetensors in the text_encoder directory.

the weight in https://huggingface.co/wangqixun/YamerMIX_v8 seems to work. Just change path "./checkpoints/sdxlUnstableDiffusers_v8HeavensWrathVAE" to "xxx/YamerMIX_v8"

[ttrinh55]

When change path "./checkpoints/sdxlUnstableDiffusers_v8HeavensWrathVAE" to "xxx/YamerMIX_v8" , I got this error: ValueError: You are trying to load the model files of the variant=fp16, but no such modeling files are available. :((

[para133]

When change path "./checkpoints/sdxlUnstableDiffusers_v8HeavensWrathVAE" to "xxx/YamerMIX_v8" , I got this error: ValueError: You are trying to load the model files of the variant=fp16, but no such modeling files are available. :((

I have converted through convert_original_stable_diffusion_to_diffusers.py, but still meet same problem ,have you solved it?

[ttrinh55]

I solved this problem by removing variant=fp16 in infer_style.py. This works with both converted one and YamerMIX_v8. If we need to use variant=fp16, I think one more step (convert model to fp16) is required.

[para133]

I solved this problem by removing variant=fp16 in infer_style.py. This works with both converted one and YamerMIX_v8. If we need to use variant=fp16, I think one more step (convert model to fp16) is required.

It's pity ,that not works for me.Can you tell me your version of diffusers? I think it may be relevant.

[ttrinh55]

I am using 0.31.0

[para133]

I am using 0.31.0

ummm,we are the same, but thanks for your fast reply.

[anujsinha72094]

Did anybody found the solution?