PontifexCipher

5 ms·

1. 1.▲
   by PontifexCipher 2mo ago

   No, my point is that it's a post on a personal blog not a peer reviewed paper.
2. 2.▲
   by PontifexCipher 2mo ago

   https://en.wikipedia.org/wiki/Blog
3. 3.▲
   by PontifexCipher 1y ago

   This is not just PR and is very interesting. However, in my view, (and from a quick read of the paper) this is actually a very classical method in applied math work: - Build a complex intractable mathematical model (here, Navier-Stokes) - A
4. 4.▲
   by PontifexCipher 1y ago

   No examples of before/after? Maybe I missed something.
5. 5.▲
   by PontifexCipher 1y ago

   What does this add to the conversation?
6. 6.▲
   by PontifexCipher 2y ago

   Some info that may be missing: - v2/v3 (not r1) seem to be cloned from o1/4o output, and perform worse (this cost the oft-repeated 5ish mm USD) - r1 is specifically a reasoning step (using RL) _on top of_ v2/v3 and performs s
7. 7.▲
   by PontifexCipher 2y ago

   Two months without a commit could still be quite active and useful software, especially for a personal project. Where would you draw the line?