Papers for

small ai model developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Small language model agents learn better with harness-aware distillation

Harness-Aware Distillation for Small Language Model Agents

Abstract: Language model agents are deployed with a harness, the software around the model that manages its context, tools, and feedback. When such an agent is distilled into a smaller one, the harness stays in place, so the student mainly needs the teacher-specific abilities that the harness cannot provide, such as acting correctly on harness information. Standard distillation, however, imitates the teacher's full outputs and treats the harness as part of the input. We propose Harness-Aware Distillation (HAD), which focuses distillation on what the teacher adds beyond the harness. HAD complements on-policy distillation with two components: an action preference that contrasts the same teacher's actions with and without the harness information, scored after the student's own reasoning, and a validity check that drops preference pairs whose preferred action contradicts the harness records. We show that the contrast gives the student information that imitating the teacher alone cannot provide, and HAD needs no task rewards, success labels, or future information. Across multiple long-horizon agent benchmarks and models, HAD outperforms on-policy distillation baselines with the same fixed harness. Our analysis shows that HAD enters fewer unproductive loops and recovers from errors more often than the baselines, and suggests that it adaptively keeps learnable feedback in its weights while reading state information from the harness.

Fri 2 OctArtificial Intelligence
The gist
Large language models often work inside special software called a harness that helps manage their context and tools. When making smaller versions of these models, the usual way is to copy everything the big model does, including the harness’s effects. The authors of this paper found a better way to teach the small model by focusing only on what the big model adds beyond the harness. This method, called Harness-Aware Distillation, helps the small model learn to act correctly with the harness’s help without needing extra rewards or success signals. Their tests show this approach helps small models avoid mistakes and handle errors better.
Open → 2610.02858v1