Sobhan Mohammadpour retweeted
Check out Toner! A general rule that preconditions the optimizer's update with each layer's input activations. Applied to Muon and AdamW as MuonT and AdamT, it speeds up LM pretraining, improving the critical batch size by ~2脳 over Muon.
Sobhan Mohammadpour retweeted
Simple deep RL was thought to fail in imperfect-info games like poker.
A new ICLR 2026 paper shows that with proper tuning, generic methods like PPO match or beat specialized approaches like fictitious play and counterfactual regret minimization.
arxiv.org/abs/2502.08938
1/2
Sobhan Mohammadpour retweeted
It is becoming increasingly apparent that Carney is a modern Castlereagh.








