Search papers, labs, and topics across Lattice.
4
0
7
Multi-objective bandits can make finding the best arm *easier* than single-objective settings, even identifying larger optimal sets with fewer samples.
Offline Q-learning can achieve $\widetilde{O}(1/蔚)$ sample complexity in finite-horizon KL-regularized MDPs, even without explicit pessimism, by leveraging KL regularization.
SPOT redefines on-policy distillation by ensuring that probing decisions directly enhance downstream reasoning performance, not just teacher alignment.
