SpecCompact: Accelerating LLM Agent Context Compaction with Slot-Parallel Semantic Speculative Verification
cs.CL
Submitted: 2026-11-04
Updated: 2026-11-04
Code: https://github.com/NousResearch/hermes-agent
Terminology
Sources
- LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding
- Accelerating Large Language Model Decoding with Speculative Sampling
- Slipstream: Trajectory-Grounded Compaction Validation for Long-Horizon Agents
- Parallel Context Compaction for Long-Horizon LLM Agent Serving
- RULER: What's the Real Context Size of Your Long-Context Language Models?
- LoRA: Low-Rank Adaptation of Large Language Models
- DREAM-R: Multimodal Speculative Reasoning with RL-Based Refined Drafting, Precise Verification, and Fully Parallel Execution
- LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Models
- ACON: Optimizing Context Compression for Long-horizon LLM Agents
- Efficient Memory Management for Large Language Model Serving with PagedAttention
- Fast Inference from Transformers via Speculative Decoding
- SnapKV: LLM Knows What You are Looking for Before Generation
- LiveClawBench: Benchmarking LLM Agents on Complex, Real-World Assistant Tasks
- Decoupled Weight Decay Regularization
- Scaling LLM Multi-turn RL with End-to-end Summarization-based Context Management
- MemGPT: Towards LLMs as Operating Systems
- SpecReason: Fast and Accurate Inference-Time Compute via Speculative Reasoning
- $\tau$-Knowledge: Evaluating Conversational Agents over Unstructured Knowledge
- Scaling Long-Horizon LLM Agent via Context-Folding
- Gemma 4 Technical Report
Related papers
- ALARM: Audio-Language Alignment for Reasoning Models
- ThinkSum: Probabilistic reasoning over sets using large language models
- Coherence boosting: When your pretrained language model is not paying enough attention
- From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents
- Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization
- The Art of Asking: Multilingual Prompt Optimization for Synthetic Data