4 ms·
The found function is in here: https://github.com/google-deepmind/funsearch/blob/main/cap_set/cap_set.ipynb https://github.com/google-deepmind/funsearch/blob/ma
by nybsjytm 3y ago
The found function is in here: https://github.com/google-deepmind/funsearch/blob/main/cap_set/cap_set.ipynb https://github.com/google-deepmind/funsearch/blob/main/cap_s.... I'm not very familiar with genetic algorithms but it's pretty hard for me to imagine that they couldn't come up with this. I'd be surprised if too many people (if any) have tried.
On the other hand, as observed in Appendix A.2 of this paper, the non-LLM genetic approach would have to be engineered by hand more than the LLM approach.
- jcgrillo 3y agoI guess what I'm really trying to get at is this seems like a huge missed opportunity to show their LLM is actually doing something cool. Like if it somehow significantly outperforms established genetic programming update strategies that should be pretty easy to demonstrate given their experimental setup, but no attempt was made. That's kinda bizarre... like, in what sense is this an advancement of the field?
- resource0x 3y agoThey need to show something to keep the excitement around Google's upcoming AI alive. Expect a slew of (non-)discoveries to continue unabated going forward.