//
sign in
Post
by @danabra.mov
PostEmbed
by @danabra.mov
Record
by @jimpick.com
Record
by @atsui.org
+ new component
Post
🧡[3/n] πŸ“‰ Even Gradients Are Sparse in RL πŸ“‰ 🧠 In PRIME, 72% of parameters never receive any gradient β€” ever! ↔️ Some do, but their gradients cancel out over time. 🎯 It’s not just sparse updates, even the gradients are sparse
May 21, 2025
Sagnik Mukherjee