search

Stability-AI / StableCascade

Official Code for Stable Cascade
MIT License
6.44k stars 518 forks source link

text to image inference too slow #108

Open ryhhtn opened 4 months ago

ryhhtn commented 4 months ago

The following message is displayed before the calculation of stage_b or stage_c starts, and there is a time lag of several minutes before the calculation starts

Please let me know if there is a way to solve these problems.

os: ubuntu 22.04 python: 3.10.12

I have installed the other libraries according to the requirements.txt file.

huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...
To disable this warning, you can either:
        - Avoid using `tokenizers` before the fork if possible
        - Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)

100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 20/20 [02:52<00:00,  8.63s/it]
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10/10 [06:51<00:00, 41.18s/it]
TwoPrismsOneGround commented 4 months ago

I am having the same issue, using the sample code from the text_to_image.jpynb example. I get to cell 4 and then get this:

image

However, upon restarting the kernel that died (WSL2, also running ubuntu 22.04) I now get this:

image

EDIT:

I fixed it by removing the "optional: compile stage c and stage b" cell and re running from the beginning. I was able to generate images after doing this.

FurkanGozukara commented 3 months ago

well we have much faster app with amazing optimizations

https://github.com/Stability-AI/StableCascade/issues/125