INFUSER: Influence-Guided Self-Evolution Improves Reasoning
2026-06-08 • Machine Learning
Machine LearningArtificial IntelligenceComputation and LanguageComputer Science and Game Theory
AI summaryⓘ
The authors created INFUSER, a system where two parts—a Generator and a Solver—work together to improve a language model's reasoning skills without much outside help. The Generator makes questions and answers from a bunch of documents, while the Solver learns from these to get better. They use a special scoring method to make sure the Generator creates questions that actually help the Solver improve, not just hard ones. INFUSER showed significant improvements on challenging benchmarks and proved flexible across different setups. The authors also introduced a new training technique called DuGRPO to better handle the noisy feedback during training.
self-evolutionlanguage modelco-trainingGeneratorSolverinfluence scoreGRPODuGRPOadaptive curriculuminstruction finetuning
Authors
Siyu Chen, Miao Lu, Beining Wu, Heejune Sheen, Fengzhuo Zhang, Shuangning Li, Zhiyuan Li, Jose Blanchet, Tianhao Wang, Zhuoran Yang
Abstract
Self-evolution offers a scalable path to stronger reasoning: a pretrained language model improves itself with only minimal external supervision. Yet existing methods either depend on extensively curated or teacher-generated training data, or, when the generator runs unsupervised, reward it by a difficulty heuristic that need not improve the solver. We introduce INFUSER, an iterative co-training framework with two co-evolving roles: a Generator that drafts questions and reference golden answers from a pool of unstructured, automatically collected documents, and a Solver that improves by training on them. The solver is trained with standard correctness rewards against the generator-provided answers, while the generator is rewarded by an optimizer-aware influence score that measures whether each proposed question would actually improve the solver on the target distribution. Because this continuous, noisy influence score is poorly served by standard GRPO, we propose DuGRPO, a dual-normalized variant of GRPO, for generator training. Together, these turn the document pool into an adaptive curriculum that favors questions useful to the current solver, not just hard ones. On Qwen3-8B-Base, INFUSER outperforms strong self-evolution baselines with over 20% relative improvement on Olympiad and SuperGPQA benchmarks, and an 8B INFUSER co-evolving generator outperforms a frozen 32B thinking generator on math and coding. Ablations confirm each design choice is necessary, and two extensions, applying INFUSER to an instruction-finetuned anchor and augmenting it with rule-verifiable RLVR data, further demonstrate the flexibility and generalizability of the framework. Code is available at https://github.com/FFishy-git/INFUSER.