Simple command line tool for text to image generation using OpenAI's CLIP and Siren (Implicit neural representation network). Technique was originally created by https://twitter.com/advadnoun
MIT License
4.37k
stars
327
forks
source link
Fixed CLIP's limit in README.md and cli.py docs | Should be 77 tokens, not characters #134
CLIP can actually manage to handle a much larger input than just 77 characters, the context length refers to the amount of tokens it can accept. 77 tokens is roughly equivalent to... ~60 words? It depends on the content, and how CLIP decides to tokenize it.
You'll know when you've gone over that 77 token limit, since CLIP raises an error anyway.
CLIP can actually manage to handle a much larger input than just 77 characters, the context length refers to the amount of tokens it can accept. 77 tokens is roughly equivalent to... ~60 words? It depends on the content, and how CLIP decides to tokenize it.
You'll know when you've gone over that 77 token limit, since CLIP raises an error anyway.