
Neel Ajjarapu
@neelajj
product @openai
ID: 2739589096
11-08-2014 02:31:58
69 Tweet
2,2K Followers
325 Following

Deep Research is genuinely useful - depending on your application - but crucially (as anticipated by Rebooting AI in 2019, and by Judea Pearl) facts and temporal reasoning remain problematic for current neural network-based approaches that lean heavily on statistics rather than







It's been an absolute privilege to get to work with Michelle Pokrass on the 4.1 model series, and especially on nano which i co-led (shoutout Suvansh Sanjeev on post training!) 4.1 nano is a ridiculously small model, which made training this model challenging in ways many of our









