NeurIPS 2026 reviews are up: reviews, scores, rebuttals and meta-reviews for 7,877 papers. See the highest rated
acceptodds: the papers of ICLR 2027
4,863 results for “keyword:"lm"” in ICLR 2027. 938 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]BRIEF: Bounded-Context Regeneration of Long Documents with Verifiable Structural Topology▼1.2k1.67
[PDF]ControlBench: Did the Approval Actually Happen? Process-Grounded Auditing of Human Control over LLM Agents▼4.3k2.00
[PDF]What Off-Policy Corrections Keep: Gradient Direction and Realized Gains from Stale LLM Rollouts▲5.4k3.00
[PDF]ROWGATE: DETERMINISTIC ROW-LEVEL ACCESS CONTROL FOR LLM QUERY AGENTS OVER FEDERATED SOURCES▼1.2k1.67
[PDF]AffectLedger: Preserving Affective Evidence and Reconstructing Longitudinal Context for LLM Agents▼5.8k2.33
[PDF]Can Post-Training on Agentic Chain-of-Thought Autonomously Screen Protein Binders For Experimental Success?▼1.2k1.67
[PDF]ProjOPD: Conflict Projection for On-Policy Distillation in Agentic Reinforcement Learning▼1361.33
[PDF]Interpretable Feature Discovery from Unstructured Data Using LLMs and Statistical Choice Models▼5.8k2.33
Withdrawn or desk rejected (938)
[PDF]When Is a Policy Violation a Model Failure? Policy Observability and Failure Attribution in Agent-Safety EvaluationDesk rejected
[PDF]-Bench: Benchmarking Proactive Speech Interaction with Full-Duplex Conversational AIDesk rejected