DICE aggregates independently encoded document chunks into a single vector to reduce evidence dilution in long-document dense retrieval, reporting gains on LongEmbed especially beyond 4k tokens.
L oo GLE : Can Long-Context Language Models Understand Long Contexts?
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
years
2026 7verdicts
UNVERDICTED 7representative citing papers
LLM novel summaries emphasize endings more than human ones, measured by aligning summary sentences to referenced chapters.
SeKV introduces resolution-adaptive semantic KV caching with GPU-CPU hierarchy and selective zoom-in reconstruction, achieving 5.9% average improvement over semantic baselines and 53.3% GPU memory reduction at 128K context.
NLAC architecture translates natural language requests to access policies via LLMs, with embedding-based subgraph selection enabling up to 98.7% accuracy on large networks per NLACBench evaluations.
MemOp is a closed-loop memory augmentation framework for SE agents that defines memory utility via downstream task impact and reports gains of up to 5.25% success rate, 4.63% resolve efficiency, and 9.79% cost reduction.
Transformers are limited to a linearly growing number of accessible output sequences with prompt length, with exponential decay in accessible proportion beyond a critical point, even under unbounded context.
MixRea benchmark reveals LLMs achieve at most 42.8% consistency on explicit-implicit reasoning tasks, with PRCP prompting proposed to recover overlooked relations.
citing papers explorer
-
Lost in a Single Vector: Improving Long-Document Retrieval with Chunk Evidence Aggregation
DICE aggregates independently encoded document chunks into a single vector to reduce evidence dilution in long-document dense retrieval, reporting gains on LongEmbed especially beyond 4k tokens.
-
Attention Flows: Tracing LLM Conceptual Engagement via Story Summaries
LLM novel summaries emphasize endings more than human ones, measured by aligning summary sentences to referenced chapters.
-
SeKV: Resolution-Adaptive KV Cache with Hierarchical Semantic Memory for Long-Context LLM Inference
SeKV introduces resolution-adaptive semantic KV caching with GPU-CPU hierarchy and selective zoom-in reconstruction, achieving 5.9% average improvement over semantic baselines and 53.3% GPU memory reduction at 128K context.
-
Natural Language Access Control (NLAC): From Help Desk Requests to Structured Policies
NLAC architecture translates natural language requests to access policies via LLMs, with embedding-based subgraph selection enabling up to 98.7% accuracy on large networks per NLACBench evaluations.
-
Enhancing Software Engineering Through Closed-Loop Memory Optimization
MemOp is a closed-loop memory augmentation framework for SE agents that defines memory utility via downstream task impact and reports gains of up to 5.25% success rate, 4.63% resolve efficiency, and 9.79% cost reduction.
-
How Many Different Outputs Can a Transformer Generate?
Transformers are limited to a linearly growing number of accessible output sequences with prompt length, with exponential decay in accessible proportion beyond a critical point, even under unbounded context.
-
MixRea: Benchmarking Explicit-Implicit Reasoning in Large Language Models
MixRea benchmark reveals LLMs achieve at most 42.8% consistency on explicit-implicit reasoning tasks, with PRCP prompting proposed to recover overlooked relations.