23 papers this week in Efficiency & systems12 active threadsbusiest: Alignment Training Dynamics

Efficiency & systems research

Agents

Stopping agents when further gains are unlikely saves token costs without hurting quality

The authors introduce GainCtrl, a method that uses a lightweight CPU regressor to predict the remaining score gain if an LLM agent continues its current attempt. By stopping attempts when the predicted gain is low, the controller reduces token consumption while preserving the quality of the best result across attempts. Across three tool-using environments and three different LLM actors, GainCtrl consistently achieves better token–quality trade-offs than full execution and a prior early-stopping baseline.

4 days ago