Epistemically Adaptive Language Agents
Experimental artifacts of research on a clipped policy-gradient objective that assigns turn-level credit using posterior-sensitive rewards.
Collection by
about 12 hours ago
Discover the best community collections!