everyone else using word2vec: king – man + woman = queen me using word2vec: fish + music = bass fish + friend = chum fish + hair = mullet fish + struggle = flounder oink - pig + bro = wassup yeti – snow + economics = homo economicus https://graceavery.com/word2vec-fish-music-bass/ …
-
Show this thread
-
Replying to @gracecondition @St_Rev
Gordon Mohr ꧁ 👁 👁꧂ Retweeted Radim Řehůřek
some similar fun is 'transorthogonal linguistics' (words on path in word2vec) from
@metasemantic:https://twitter.com/radimrehurek/status/638920830806716416 …Gordon Mohr ꧁ 👁 👁꧂ added,
4 replies 2 retweets 14 likes -
It me! Even though there has been a lot of talk recently about embedding models not respecting the analogy game wrt. gender, this experiment convinced me that there is *significant* structure in this space, and gender is one of the strongest directions.
1 reply 0 retweets 0 likes -
Replying to @metasemantic @gojomo and
I've always wondered if these significant directions could be pulled out automatically, esp. if they are non linear. That is, is there structure on the all-to-all distance matrix and does this recover the biases? (Bonus Q: does it have temporal dependence?)
1 reply 0 retweets 0 likes
Topological data analysis seems promising.https://www.youtube.com/watch?v=h0bnG1Wavag …
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
. Banned in Sweden. SubGenius, Zhuangist, white-hat troll. Defrocked mathematician. Brain problems.