Accelerating Cross-Model LLM Inference via Attention Map Reuse
cs.LG
Submitted: 2026-10-10
Updated: 2026-10-10
Code: https://github.com/PKU-DAIR/MapFlow
Sources
- SEED: Domain-Specific Data Curation With Large Language Models
- Latent Space Communication via K-V Cache Alignment
- Cache-to-Cache: Direct Semantic Communication Between Large Language Models
- {\lambda}-Tune: Harnessing Large Language Models for Automated Database System Tuning
- RULER: What's the Real Context Size of Your Long-Context Language Models?
- ChatDB: Augmenting LLMs with Databases as Their Symbolic Memory
- RouterBench: A Benchmark for Multi-LLM Routing System
- Automated Design of Agentic Systems
- LlamaTune: Sample-Efficient DBMS Configuration Tuning
- Q-KVComm: Efficient Multi-Agent Communication Via Adaptive KV Cache Compression
- FlexPrefill: A Context-Aware Sparse Attention Mechanism for Efficient Long-Sequence Inference
- Personal LLM Agents: Insights and Survey about the Capability, Efficiency and Security
- SnapKV: LLM Knows What You are Looking for Before Generation
- Ministral 3
- DroidSpeak: KV Cache Sharing for Cross-LLM Communication and Multi-LLM Serving
- Coral: Cost-Efficient Multi-LLM Serving over Heterogeneous Cloud GPUs
- Qwen3 Technical Report
- LLMs as On-demand Customizable Service
- Multi-Segment Attention: Enabling Efficient KV-Cache Management for Faster Large Language Model Serving
- Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks