arXiv cs.AI - 2026-08-25

585 items collected.

1. KVBoost: Chunk-Level Key-Value Cache Reuse with Deviation-Guided Recomputation for Efficient Large Language Model Inference

Author: Srihari Unnikrishnan
Published: 8/25/2026, 4:00:00 AM
Categories: cs.AI, cs.DC

arXiv:2608.21362v1 Announce Type: new Abstract: Transformer-based large language models (LLMs) incur high prefill latency because key-value (KV) tensors must be recomputed for each request. Existing prefix-caching systems reduce this cost but require prompts to share a leading contiguous prefix, lim...

arXiv cs.AI - 2026-08-25 ​

1. KVBoost: Chunk-Level Key-Value Cache Reuse with Deviation-Guided Recomputation for Efficient Large Language Model Inference ​

2. AIREP: A Protocol for Per-Decision Evidence in AI Runtime Governance ​

3. Reviewing Model Collapse and Countermeasures ​

4. AI Learning and Conceptual Transfer in the Game of Hidden Rules ​

5. LitReview Arena: Evaluating Literature Review Agents with Battle-Style Peer Review Platform ​

6. SchemaRouter: Field-Aware Tool Routing for Efficient Heterogeneous Agentic RAG ​

7. RIACT: A Responsible AI System for Personalized Study Habit Tracking and Early Burnout Signal Detection in University Students ​

8. There Is No Neutral Harness: Modern LLM Leaderboards Are Manufactured by Config-Fragile Items ​

9. Spyre-Accelerated Retrieval-Augmented Generation on IBM LinuxONE: A Cloud-Native Architecture for Secure, High-Throughput Enterprise AI Inference ​

10. Hate Speech Classification In Roman Urdu: A Comparative Study On Parameter Efficient Fine-Tuning And Prompt Engineering ​

11. The Abstention Protocol: RCA for Clos Fabrics ​

12. Retrieval-grounded robot program generation and simulation-based correction via Model Context Protocol ​

13. Composable Trust Infrastructure for Manufacturing Knowledge Graphs: Cross-System Provenance, Temporal Reasoning, and Decision Traceability ​

14. Evaluating Multimodal Narrative Understanding of Popular Hollywood Films ​

15. Agentic AI for Safety-critical Multi-drone Systems: Challenges and Opportunities ​

16. Software Frameworks for Explainable AI in Time Series Classification: A Systematic Review ​

17. Enhanced Artificial Neural Networks Using QHAdamW in Air Quality Forecasting ​

18. Let Credit Follow Computation: Architecture-Aware Credit Transport for Large Language Model Reinforcement Learning ​

19. Quantifying geographic domain shift to decouple the geospatial transferability of human mobility flow generation models ​

20. A Reproducible, License-Aware Distillation Recipe for CPUDeployable Safety Classification ​

21. Robust Lightweight Deep Learning Models for Oral Cancer Screening ​

22. Data-Driven Dynamic Algorithm Dispatch with Large Language Models ​

23. K-Bench: measuring model performance on real scientific agent requests ​

24. Generate in the Chart, Not on the Boundary: Function-Symbol Grounding for Hard Constraints in LTN-GANs ​

25. Semantic Compression Trees: Multi-Resolution Knowledge Retrieval via Hierarchical Semantic Residuals ​

26. SAEM: Stage-Aware Expert Management for Memory-Efficient MoE Inference in Chain-of-Thought Reasoning ​

27. Measuring Activation Control in Large Language Models ​

28. From Mastery Profile to Simulated Response: Stochastic Student Knowledge Graphs (SSKG) for Faithful LLM Student Simulation ​

29. Context as an Environment: Programmatic Context Management for Long-Horizon Agents ​

30. From Association to Causation: Improving Retrieval Precision of Retrieval-Augmented Generation via Causal Relations and an Attention Mechanism ​

31. ATHENA: Knowledge-guided agentic neural architecture search for AutoFormer-based electronic health record modeling ​

32. Ask or Answer: A Decision Framework for Multi-Turn Health Misinformation Intervention ​

33. ECHO: A Cognitively Inspired, Auditable Memory Plane for Long-Horizon Agents ​

34. What Does CLIP Learn for Regional Geolocalization? Probing Visual Cues and Scene Configuration After Adaptation ​

35. Physics-Knowledge-Guided Hybrid Neural Learning for Arctic Sea Ice Concentration Evolution and Short-Range Prediction ​

36. HIRA: A Human-in-the-Loop Retrieval-Augmented Cascade for Document Classification in Regulated Industries ​

37. Hints, Critics, and Teachers: Prior Injection for Sparse-Reward RL in Vision-Language Math Reasoning ​

38. VisAdj: Learning Adjacency Matrices from Node-Link Images ​

arXiv cs.AI - 2026-08-25

1. KVBoost: Chunk-Level Key-Value Cache Reuse with Deviation-Guided Recomputation for Efficient Large Language Model Inference

2. AIREP: A Protocol for Per-Decision Evidence in AI Runtime Governance

3. Reviewing Model Collapse and Countermeasures

4. AI Learning and Conceptual Transfer in the Game of Hidden Rules

5. LitReview Arena: Evaluating Literature Review Agents with Battle-Style Peer Review Platform

6. SchemaRouter: Field-Aware Tool Routing for Efficient Heterogeneous Agentic RAG

7. RIACT: A Responsible AI System for Personalized Study Habit Tracking and Early Burnout Signal Detection in University Students

8. There Is No Neutral Harness: Modern LLM Leaderboards Are Manufactured by Config-Fragile Items

9. Spyre-Accelerated Retrieval-Augmented Generation on IBM LinuxONE: A Cloud-Native Architecture for Secure, High-Throughput Enterprise AI Inference

10. Hate Speech Classification In Roman Urdu: A Comparative Study On Parameter Efficient Fine-Tuning And Prompt Engineering

11. The Abstention Protocol: RCA for Clos Fabrics

12. Retrieval-grounded robot program generation and simulation-based correction via Model Context Protocol

13. Composable Trust Infrastructure for Manufacturing Knowledge Graphs: Cross-System Provenance, Temporal Reasoning, and Decision Traceability

14. Evaluating Multimodal Narrative Understanding of Popular Hollywood Films

15. Agentic AI for Safety-critical Multi-drone Systems: Challenges and Opportunities

16. Software Frameworks for Explainable AI in Time Series Classification: A Systematic Review

17. Enhanced Artificial Neural Networks Using QHAdamW in Air Quality Forecasting

18. Let Credit Follow Computation: Architecture-Aware Credit Transport for Large Language Model Reinforcement Learning

19. Quantifying geographic domain shift to decouple the geospatial transferability of human mobility flow generation models

20. A Reproducible, License-Aware Distillation Recipe for CPUDeployable Safety Classification

21. Robust Lightweight Deep Learning Models for Oral Cancer Screening

22. Data-Driven Dynamic Algorithm Dispatch with Large Language Models

23. K-Bench: measuring model performance on real scientific agent requests

24. Generate in the Chart, Not on the Boundary: Function-Symbol Grounding for Hard Constraints in LTN-GANs

25. Semantic Compression Trees: Multi-Resolution Knowledge Retrieval via Hierarchical Semantic Residuals

26. SAEM: Stage-Aware Expert Management for Memory-Efficient MoE Inference in Chain-of-Thought Reasoning

27. Measuring Activation Control in Large Language Models

28. From Mastery Profile to Simulated Response: Stochastic Student Knowledge Graphs (SSKG) for Faithful LLM Student Simulation

29. Context as an Environment: Programmatic Context Management for Long-Horizon Agents

30. From Association to Causation: Improving Retrieval Precision of Retrieval-Augmented Generation via Causal Relations and an Attention Mechanism

31. ATHENA: Knowledge-guided agentic neural architecture search for AutoFormer-based electronic health record modeling

32. Ask or Answer: A Decision Framework for Multi-Turn Health Misinformation Intervention

33. ECHO: A Cognitively Inspired, Auditable Memory Plane for Long-Horizon Agents

34. What Does CLIP Learn for Regional Geolocalization? Probing Visual Cues and Scene Configuration After Adaptation

35. Physics-Knowledge-Guided Hybrid Neural Learning for Arctic Sea Ice Concentration Evolution and Short-Range Prediction

36. HIRA: A Human-in-the-Loop Retrieval-Augmented Cascade for Document Classification in Regulated Industries

37. Hints, Critics, and Teachers: Prior Injection for Sparse-Reward RL in Vision-Language Math Reasoning

38. VisAdj: Learning Adjacency Matrices from Node-Link Images