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
770 results for “keyword:"post train"” in ICLR 2027. 139 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]RACE: Learning Multimodal Search through Referential Abstraction and Compositional Evidence▲3.0k2.33
[PDF]Not All Reconstruction Errors Matter: Score-Guided Block-wise Reconstruction for Low-Bit LLM Quantization▲5.4k3.00
[PDF]FieldForge: A Pretraining and Post-Training Framework for Flow-Based Crystal Generation▼4.3k2.00
[PDF]Rotation Is Token-Blind: A Structural Criterion for Post-Training Quantization of Feed-Forward Geometry Transformers▲5.4k3.00
[PDF]TRPQuant: Token Rearrangement and Channel Pruning for Quantizing Large Vision-Language Models▼4.5k2.67
[PDF]Retrospection Makes a Better Learner: Retrospective On-Policy Self-Distillation for Reasoning Models▲6.1k2.00
[PDF]ScaleSweep: Accurate NVFP4 Post-Training Quantization of LLMs via Block Scale Initialization▼1.2k1.67
[PDF]No Man Is an Island: Overlapping Pruning for Near-Unstructured Quality at 2:4 Efficiency▲2623.67
[PDF]When Do We Need On-Policy Distillation? Distilling on Offline Student Rollouts Is Often Better▼4.5k2.67
[PDF]Off-Diagonal Hessian Sensitivity for Post-Training Quantization of Large Language Models▼4.3k2.00
[PDF]Adaptive Mutual Distillation for Balanced Multi-Task Post-Training of Large Language Models▲3.8k2.50
[PDF]PulseQuant: Propagation-Guided Subspace Correction for 4-Bit Video Diffusion Transformers▲5103.00
[PDF]Grounded without Drifting from Pretraining: Rollout-Free Target Optimization with Implicit Distillation for RAG▲5.4k3.00
[PDF]Does Assigned Identity Change How Language Models Use History? Separating Self-Binding from Focal-Agent Effects▼1.2k1.67
[PDF]What Is Lost in Post-Training? Default Collapse and the Loss of In-Context Steerability Across Diverse Perspectives▼4.3k2.00
[PDF]What about gravity in video generation? Post-Training Newton's Laws with Verifiable Rewards▼16k1.33
Withdrawn or desk rejected (139)
[PDF]Everything in Moderation: Interior Per-Domain Coverage Optima and Gap-Preserving Fixed-Budget Alignment in Multi-Domain Mid-TrainingDesk rejected
[PDF]Why Benign Post-Training Induces Safety Gate Collapse: Changes in Harmful Activation Geometry Relative to the Inherited Refusal GateWithdrawn1.67
[PDF]Hierarchical Safety-Oriented Dense RL with Refined Credit Assignment for Tokenized Multi-Agent Traffic SimulationWithdrawn1.75
[PDF]Decoupling KL and Trajectories: A Unified Perspective for SFT, DAgger, Offline RL, and OPD in LLM DistillationDesk rejected1.50