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
37 results for “keyword:"attention sinks"” in ICLR 2027. 8 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]Attention Sinks as Knowledge-Conflict Arbiters: Training-Free Gating-Preserving Routing for Contextual Faithfulness▲3972.67
[PDF]One Token Owns the Calibration Hessian: Why Layer-wise Quantization Can Silently Break Instruction Following▼1.2k3.00
[PDF]Sink Before You Speak: Connecting Language Generation, Attention Patterns, and Representational Geometry▼4.3k2.00
[PDF]Attention Sinks in Transformer-based Language Models: Emergence Mechanisms and Benign Effects▲5.4k3.00
[PDF]ASEMbed: Training-Free Embedding Extraction for Order-Robust Retrieval via Internal LLM Signals▼5.8k2.33
[PDF]The Life Cycle of a Massive Activation: Stochastic Birth, Weight-Decay-Driven Growth, and Competitive Consolidation▼4.3k2.00
[PDF]The Price of a Merge Is Paid in Depth: A Tight, Label-Free Bound on Token Merging in Vision Transformers▼3.7k2.25
[PDF]Before Attention Sinks Form: Hidden-State Geometry Reveals and Modulates Their Formation▼1.2k1.67
[PDF]The Sink Tax: Auditing Memory Tokens and Assigning Their Roles in Long-Context Language Models▼4.3k2.00
[PDF]Sink or Swim: Diagnosing and Resolving the Relational Invariance Bottleneck in Causal Attention▼1.2k1.67
[PDF]Mind the Spike: Mechanisms and Brittleness of Visual Massive Activations in Large Vision–Language Models▼4.5k2.67
[PDF]Weight-Dictionary Decomposition: Reading a Transformer's Residual Stream with Its Own Writes▼1361.33
[PDF]Structural Text Recipients Route Same-Prompt Visual Realization in Diffusion Transformers▼1.2k1.67
[PDF]Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateaus▲3972.67
Withdrawn or desk rejected (8)
[PDF]From Frequency to Function: Component-Specific RoPE Extrapolation for Long-Context Language ModelsDesk rejected3.00
[PDF]Topological Hallucination Detectors on Decoder Attention Reduce to First-Order StatisticsDesk rejected
[PDF]SAGA: Spatially-Aware Gated Attention for Separating Outlier Structure from Functional Effect in Vision TransformersWithdrawn1.67
[PDF]Memory Outweighs Control: A Causal Audit of Cached History in Autoregressive Video GenerationWithdrawn1.00