Harness-Aware Distillation for Small Language Model Agents

The paper introduces Harness-Aware Distillation (HAD), a method for training smaller language model agents that preserves the surrounding harness—software managing context, tools, and feedback—while focusing distillation on the teacher’s contributions beyond the harness. HAD combines an action preference that contrasts teacher actions with and without harness information, and a validity check that filters out contradictory preference pairs. Experiments on long-horizon agent benchmarks show that HAD outperforms standard on‑policy distillation, reducing unproductive loops and improving error recovery without requiring task rewards or future information.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

• llms
• agents
• efficiency
• benchmarks