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
97 results for “keyword:"llm post training"” in ICLR 2027. 15 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]Not All Tokens Should Update the Same Parameters: Prefix-Conditioned Gradient Routing for On-Policy Distillation▼1.2k3.00
[PDF]Confidence-Guided Policy Optimization: Thompson Sampling and UCB Trees for Group-Relative Code Refinement▼5.9k2.50
[PDF]CyberNeuro: A Privacy-Preserving Agentic Workbench for Cohort-Scale Brain Imaging and Clinical Data Analysis▼1.3k1.75
[PDF]Entropy Regularization: A Free Correction to Cross-Entropy for Verified Demonstrations▼5.8k2.33
[PDF]Learning to Stop without Learning to Stop: Self-Supervised Confidence Training Improves Reasoning Efficiency▲3952.67
[PDF]LatentGym: A Testbed for Cross-task Experiential Learning with Contollable Latent Structure▼1.2k3.00
[PDF]Diagnosing Recoverability: Suppression-Like vs. Overwrite-Like Capability Loss in LLM Post-Training▲3952.67
[PDF]Can domain synergy mislead? Measurement confounds in multi-domain RLVR interaction estimation▼1.2k1.67
[PDF]RecoAtlas: A Benchmark and RL Environment for LLM Recommendation Agents Optimizing Set-Level Utility▼4.3k2.00
Withdrawn or desk rejected (15)
[PDF]From Gradient Surgery to Bottleneck Regularization: A Unified Framework for Combining Verifiable-Reward RL and On-Policy Self-DistillationDesk rejected
[PDF]ExpertGuard: Transcending the Stability-Plasticity Dilemma in LLM Reinforcement Learning via Expert-Vector Guarded Policy UpdatesDesk rejected