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
22 results for “keyword:"top k"” in ICLR 2027. 5 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]Stable Accuracy, Unstable Allocations:Retraining Reproducibility in Selective Prediction▼1.2k1.67
[PDF]Ranking Is Not Support Selection: Exact Marginals and Certified Headroom for Sparse Attention in LLMs▼1.2k1.67
[PDF]Dictionary Size and Dead Latents Handling Jointly Set the Recovery-Optimal Sparsity of Top-K Sparse Autoencoders▼1361.33
[PDF]Posterior Top-K Selection under Noisy Reviews: Equivalence and Gain Relative to Posterior Mean▼4.3k2.00
[PDF]Geometric Capacity of Top- and Sliding-Window Attention: A Theoretical Analysis via Tropical Geometry▼4.3k2.00
[PDF]Not All Tokens Deserve Equal Resolution: Score-Adaptive Indexing for Sparse Attention in Long-Context LLMs▲5103.00
Withdrawn or desk rejected (5)
[PDF]Learning What to Distill: Bilevel Top-K Token Selection for Self-Distillation in Large Language ModelsDesk rejected