SelfCompact enables language models to autonomously decide when and how to compact accumulated context during reasoning. By combining a model-invoked summarization tool with a lightweight rubric that guides compaction based on trajectory structure, it achieves effective adaptive compaction without fine-tuning. Results show it matches or exceeds fixed-interval methods on math and agentic search benchmarks, improving baselines by up to 18.1 points on math and 5-9 points on search, at 30-70% lower token cost.
arxiv
arXiv cs.CL
·
2d ago
·
research
SelfCompact: Self-Driving Context Compaction for Language Models
from English
Benchmarks
| Benchmark | Model | Score |
|---|---|---|
| AgentBench | selfcompact | 5pts |
| MATH-500 | selfcompact | 18.1pts |