Weak-Driven Learning: How Weak Agents make Strong Agents Stronger
This addresses a saturation bottleneck in post-training for large language models, which is an incremental but practical improvement for AI developers.
The paper tackles the problem of diminishing returns in post-training optimization for large language models by proposing WMSS, a method that leverages weak checkpoints to guide further training. Experiments on mathematical reasoning and code generation datasets show performance improvements with no additional inference cost.
As post-training optimization becomes central to improving large language models, we observe a persistent saturation bottleneck: once models grow highly confident, further training yields diminishing returns. While existing methods continue to reinforce target predictions, we find that informative supervision signals remain latent in models' own historical weak states. Motivated by this observation, we propose WMSS (Weak Agents Can Make Strong Agents Stronger), a post-training paradigm that leverages weak checkpoints to guide continued optimization. By identifying recoverable learning gaps via entropy dynamics and reinforcing them through compensatory learning, WMSS enables strong agents to improve beyond conventional post-training saturation. Experiments on mathematical reasoning and code generation datasets show that agents trained with our approach achieve effective performance improvements, while incurring zero additional inference cost.