Conor Durkan
@conormdurkan.bsky.social
Generative modeling person
https://conordurkan.com
https://conordurkan.com
First book of the year and it was a doozy. I think I have a soft spot for aerospace engineering--I liked this almost as much as 'Carrying the Fire'.
January 7, 2025 at 11:38 PM
First book of the year and it was a doozy. I think I have a soft spot for aerospace engineering--I liked this almost as much as 'Carrying the Fire'.
I like the Bayesian framing of reward-based post-training (i.e. reward-maximization with a KL penalty).
(Figure from 'RL with KL penalties is better viewed as Bayesian inference', link below along with other useful references)
(Figure from 'RL with KL penalties is better viewed as Bayesian inference', link below along with other useful references)
January 3, 2025 at 5:02 PM
I like the Bayesian framing of reward-based post-training (i.e. reward-maximization with a KL penalty).
(Figure from 'RL with KL penalties is better viewed as Bayesian inference', link below along with other useful references)
(Figure from 'RL with KL penalties is better viewed as Bayesian inference', link below along with other useful references)