logoalt Hacker News

ferguess_kyesterday at 6:41 PM2 repliesview on HN

Can you please explain "the transfer function is negative"?

I'm wondering whether one has tested with the same model but on two situations:

1) Bring it to superhuman level in game A and then present game B, which is similar to A, to it.

2) Present B to it without presenting A.

If 1) is not significantly better than 2) then maybe it is not carrying much "knowledge", or maybe we simply did not program it correctly.


Replies

Zanfatoday at 5:25 AM

According to Carmack's recent talk [0], SOTA models that have been trained on game A don't perform better or train faster on game B. Even worse, training on game B negatively affects performance in game A when returning to it.

[0] https://www.youtube.com/watch?v=3pdlTMdo7pY

show 1 reply
toughyesterday at 6:43 PM

I think the problem is we train models to pattern match, not to learn or reason about world models

show 6 replies