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
11,747 results for “keyword:"models"” in ICLR 2027. 2,924 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]LLMs as Linguistic Chameleons: Decoupling Semantics and Structure for Privacy-Preserving Communication▼1.2k1.67
[PDF]MaskAhead: Unifying KV Cache Selection and Eviction in Block Diffusion Language Models▼1.2k1.67
[PDF]YANAMI: A Large-Scale Multimodal Dataset and Benchmark For Multi-turn Stateful Interaction▲5083.00
[PDF]What Transfers from a VLM Teacher? Comparing Supervision Signals for Visual Document Retrieval▲5083.00
[PDF]Conditional Diffusion without ERM Oracles: From Finite Neural Training to Sampling Guarantees▼5.9k2.33
[PDF]The Platonic gene: convergent representations across biological models learn cell plausibility▼4.3k2.00
[PDF]Retrofitting LLM-Based Pathway Enrichment Analysis with Dempster-Shafer Theory of Evidence▼3.7k2.25
[PDF]AWARD: Answer-Level Weighted Aggregation of Retrieval-Free Diagnostics for Hallucination Detection▼1351.33
[PDF]Model Collapse Is Avoidable Even as Real Data Vanishes: A Sharp Criterion for Iterative Gaussian Estimation▲3772.67
[PDF]LLM-as-its-judge: A compute paradigm for self-aware and self-directing reasoning machines▼4.3k2.00
[PDF]Prepare in Proportion: Deletion-Hazard-Aware Training for Unlearning-Ready Federated Language Models▲3772.67
[PDF]TotalGuardEval: Benchmark for Evaluating Guardrail Models on Underrepresented Risk Categories▼5.9k2.33
[PDF]What Multilingual Models Share: The Functional and Neural Significance of Representational Convergence▲6.1k2.00
Withdrawn or desk rejected (2,924)
[PDF]Mind Your Marginals: What Governs Fidelity in Linearized Hierarchical Optimal Transport?Desk rejected1.00
[PDF]A Window Is Not a Door: Mapping the Limits and Conditions of Knowledge Editing in Language ModelsDesk rejected2.33
[PDF]Propagation Is Not Decoration: When Physical Diffraction Carries Generative ComputationWithdrawn1.00
[PDF]When Is a Policy Violation a Model Failure? Policy Observability and Failure Attribution in Agent-Safety EvaluationDesk rejected