Diagnosing and Mitigating Context Rot in Long-horizon Search
- Published
- Source
- arXiv
- Paper number
- 524
- Field
- Information Retrieval
- arXiv ID
- 2606.29718
Key points
- Context rot is defined as the phenomenon in which accumulated context causes the model to give up or produce an uncertain answer.
- It performs a systematic diagnosis using four open-source models, including Qwen3.5-397B and GLM-4.7, across three benchmarks.
- It evaluates seven management methods, including context compaction, trimming, and isolation, in terms of performance, cost, and their effect on rot.
- Rot-aware filtering, which removes uncertain or give-up trajectories, improves average performance by 2.6 to 4.9 percent.
- The combination of compaction and trimming offers the best balance between cost and rot mitigation.
- Context isolation through subagents is highly model dependent and only works well on stronger models.
Paper links
External research summaries. These are not HDATF publications or measured product results.