Search papers, labs, and topics across Lattice.
1
0
3
4
Latent reasoning can now leverage outcome-reward reinforcement learning, achieving better performance with less computational overhead than traditional methods.
