Beyond Binary Verifiable Rewards: Behavioral Self-Rewarding for SWE Agents

Key points

• We investigate whether the current policy can judge its own trajectories and use these judgments to improve task performance without a separately trained reward model.1
• To this end, we introduce Behavioral Self-Rewarding (BSR), a reward-refinement method that reuses the policy to compare observable debugging behavior among trajectories from the same task with the same verifier outcome.1
• BSR aggregates pairwise preferences into task-local scores and converts them into bounded auxiliary rewards, introducing reward contrast while preserving the verifier-defined pass/fail ordering.1
• Experiments on SWE-bench Verified and SWE-bench-Live show consistent improvements across the evaluated Qwen3-14B and Qwen3-32B configurations, supporting using the policy's own comparative judgments as a complementary learning signal for training software engineering agents.1

Sources (1)

• [1]Beyond Binary Verifiable Rewards: Behavioral Self-Rewarding for SWE Agents
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 06:21 PM

  Reinforcement learning with verifiable rewards (RLVR) provides reliable supervision for software engineering agents through executable tests, but binary rewards fail to distinguish trajectories with substantially different debugging quality.

  We investigate whether the current policy can judge its own trajectories and use these judgments to improve task performance without a separately trained reward model.

Extractive summary: sentences quoted from the sources.