Global AI News Aggregator
About
By
–
o3 is a lot of matmuls trained with gradient descent.
→ View original post on X — @soumithchintala