leejet

@ranjithum The VAE in SDXL encounters NaN issues under FP16, but unfortunately, the ggml_conv_2d only operates under FP16. Hence, a parameter is needed to specify the VAE that has fixed...

> Works for me, but colors are weirdly off with SD XL plus fp16 fix:   Try changing the image size to 1024x1024. SDXL is not suitable for generating...

Thank you for your contribution. But as a C-style api header file, stable-diffusion.h should be kept as simple as possible, using only one file, so I don't think these changes...

> Check your prompt lenght (max. 75 tokens) Prompt longer than 75 tokens are already supported.

> Does this version support CPU only? I execute it and closes immediately. @cbresciano Can you provide more details? Including what operating system to run on, detailed output of sd...

> @leejet can you help us with this? pls. I have updated ggml to the latest commit. You can pull the latest changes from the master branch and give it...

> @leejet can you add support for this: [ggerganov/ggml#647](https://github.com/ggerganov/ggml/issues/647) pls I am not familiar with Metal at the moment, so I may not be of much help.

It seems that your ggml is not up to date because the submodule URL has changed. Try executing the following command to update it. ``` git submodule sync git submodule...

The latest master code has already utilized `ggml_backend_alloc_ctx_tensors`.

The latest master branch is using this: https://github.com/leejet/ggml/tree/batch-inference. It seems that your ggml is not up to date because the submodule URL has changed. Try executing the following command to...