Self-Compacting Language Model Agents

Apple Machine Learning Research · 3 days ago · 13

Self-compacting language model agents were proposed to prevent long agent traces from accumulating stale context that pushes earlier content out of the context window. The method, SELFCOMPACT, reports up to 18.1 point gains on math and uses 30–70% fewer tokens per question. It changes compaction from fixed token-interval summarization to a model-invoked summarization tool plus a lightweight rubric that decides when to summarize or suppress during trajectories.