Researchers propose DynaKV, a post-training framework that dynamically allocates compression rates to individual tokens based on semantic importance. The method achieves 94% baseline performance while reducing KV cache to just 6% of original size on LongBench benchmarks.
Researchers propose a novel framework called SharedLLM that extends language model context windows from 8K to 128K tokens without costly continual pre-training. The method uses two stacked short-context models—one as a compressor, one as a decoder—with specialized tree-based information retrieval, achieving 2-3x inference speedups while maintaining competitive performance.
Researchers introduce MemSifter, a framework that offloads memory retrieval to smaller proxy models instead of burdening the primary LLM. The approach uses outcome-driven reinforcement learning to optimize retrieval accuracy while minimizing computational overhead during inference.
Researchers introduce AMA-Bench, a benchmark for evaluating long-horizon memory in LLM-based autonomous agents using real-world trajectories and synthetic scaling. Existing memory systems underperform due to lack of causality and reliance on lossy similarity-based retrieval. The proposed AMA-Agent system with causality graphs and tool-augmented retrieval achieves 57.22% accuracy, outperforming baselines by 11.16 percentage points.