Live data from Hacker News

Show HN: This Word Does Not Exist

thisworddoesnotexist.com

1–10 of 359 posts

Re: Show HN: This Word Does Not Exist

#3

This is awesome! What kind of bias would we expect to see in the training set?

For example if it consumes lots of data that uses the pronoun "he" for doctor, it's much more likely to spit out "he" in a medical context.

Since the model lacks world knowledge for new words, sometimes it also ascribes words to a specific cultural origin / group that is completely wrong.

Re: Show HN: This Word Does Not Exist

#7

This is pretty cool but I am getting some strange results sometimes: https://imgur.com/a/rv0wHN1

That's not ideal and probably picking up some of the original training set from GPT-2 (this model is bolted on top of it)! How about DUOLINGOLOGY instead https://bit.ly/3fPGP8q

I'm using a blacklist to reject "real" words but it's surprisingly hard to build for rare words. I'm up to ~600K items after parsing Wikipedia tokens and it still doesn't capture everything.

Re: Show HN: This Word Does Not Exist

#9

This is pretty cool but I am getting some strange results sometimes: https://imgur.com/a/rv0wHN1

That's not ideal and probably picking up some of the original training set from GPT-2 (this model is bolted on top of it)! How about DUOLINGOLOGY instead https://bit.ly/3fPGP8q I'm using a blacklist to reject "real" words but it's surprisingly hard to build for rare words. I'm up to ~600K items after parsing Wikipedia tokens and it still doesn't capture everything.

That's understandable! Another one I got was "whitepaper".

In any case, it's all pretty impressive! Defintely something one can use for creative writing or fantasy world building.

Post reply on HN