Hacker News new | past | comments | ask | show | jobs | submit login

Just curious, have you tried a more uncommon paragraph? Could it be the case that the model simply learned the poem due to it being in the training set?




That fragment isn't any poem I know of. Google shows one result for an exact search of the opening, which is this post.


I found more of it in here: https://www.oranlooney.com/post/playfair/

However, I can't determine where this is originally from...


it's… from men in black iirc, so if it knows movie scripts, it'd know this.


Here is the result of asking for word segmentation with the text of your comment and the text of this comment, minus the link.

https://chat.openai.com/share/b17ecc0b-570c-4e20-9556-23bfa1...


Seems easy enough to do a more rigorous test. Just find a large set of novel text, write a program to segment it by sentence as well as uppercasing and removing spaces/punctuation.

Then run it through the GPT-4 API and compare the output to the original.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: