proomt

Search

Search posts, papers, and topics

All posts

Hugging Face Daily PapersJie Zhao, Ziyu Jiang, Suhang Zheng1 min readpaperadvanced

One to More, More to One: Category-Aware Iterative Expert Training for Software Engineering Agents

Read on arxiv.org

Summary

The authors propose a category‑aware expert‑training loop (RRE) and a multi‑teacher on‑policy distillation (MOPD) to improve software‑engineering agents across heterogeneous task categories. On Pro‑618 and SWE‑bench Multilingual the final student model reaches 58.04% and 59.00% resolution, 5.39 and 2.78 points above the base.

  • Category‑specific RL improves average success but leaves uneven per‑category progress, prompting explicit expert consolidation.
  • Refresh‑Repair‑Expand (RRE) cycles refresh mastery, repair via SFT on verified trajectories, and expand task selection.
  • Label‑routed multi‑teacher on‑policy distillation (MOPD) merges experts into a single deployable model with ReLU‑gated reward extrapolation.
  • No external solution trajectories are required; the framework relies solely on self‑generated experience.

Teams building LLM‑based software‑engineering agents need methods to balance performance across diverse task categories without external supervision.

7/10
llmmachine learningreinforcement learningagentssoftware engineering

Related reading

  1. Generalized Agent Iteration: One Formal Framework for Iterative Policy Improvement and Recursive Self-Improvement

    The paper introduces Generalized Agent Iteration (GAI), a formal framework that unifies classical iterative policy improvement (GPI) and recursive self‑improvement (RSI). GAI treats an agent as a set of modifiable components and models learning as a loop of evaluation and improvement. Two binary “dials”—whether the improvement mechanism is internal to the agent and whether the evaluation standard…

    Hugging Face Daily Papersarxiv.org1 minpaper

  2. Training Specialist Models without Reasoning Trajectories for Domain Expert Distillation

    The paper shows that when specialist LLMs are trained only on QA pairs (no explicit reasoning supervision), their optimization implicitly selects a latent distribution of reasoning trajectories. By treating the distilled student as an agnostic probe—since it inherits only the sampled trajectories—the authors empirically demonstrate a strong correlation (across 27 specialist‑student pairs) between…

    Hugging Face Daily Papersarxiv.org1 minpaper

  3. RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning

    RetireOPD introduces a self‑retiring on‑policy distillation framework for multi‑turn RL agents. A skill‑conditioned teacher is first trained with environment rewards, then a skill‑free student learns jointly via RL and token‑level distillation. The student automatically drops the teacher once its performance gap stops shrinking and it reaches a target success‑rate fraction, after which training c…

    Hugging Face Daily Papersarxiv.org1 minpaper

  4. Agentic Skill Decay

    Addy Osmani warns that AI agents can short‑circuit the hands‑on practice (“reps”) that builds deep expertise and judgment. He recommends deliberately inserting hypothesis‑forming, “why” questioning, diff inspection, failure prediction, and occasional manual coding into the workflow, especially for junior engineers. A 2026 Anthropic study showed junior developers using AI scored 17 % lower on a fo…

    Addy Osmaniaddyosmani.com16 min

  5. AI Skills with Matt Pocock

    Matt Pocock explains how he uses AI agents for software development, emphasizing "strategic programming" and guiding agents with "leading words" from classic engineering texts. He argues that this approach makes engineering fundamentals more critical than ever for creating agent-optimized codebases.

    The Pragmatic Engineerpragmaticengineer.com7 min