Hacker News new | ask | show | jobs
by ares623 240 days ago
Wouldn’t the training or whatever make that unicode sequence effectively a smiley face?
2 comments
jncfhnb 240 days ago
Yes, but the same face gets represented by many unique strings. Strings which may more may not be tokenized into a single clean “smiley face” token.
link
scotty79 240 days ago
Don't ask ChatGPT about seahorse emoji.
link
astrange 240 days ago
That's caused by the sampler and chatbot UI not being part of the LLM. It doesn't get to see its own output before it's sent out.
link
tensor 240 days ago
Don't ask humans either, apparently.
link