[9] [ arXiv:2608.28421 ](https://arxiv.org/abs/2608.28421 "Abstract") [[pdf](https://arxiv.org/pdf/2608.28421 "Download PDF"), [html](https://arxiv.org/html/2608.28421v1 "View HTML"), [other](https://

Abstract

[9] [ arXiv:2608.28421 ](https://arxiv.org/abs/2608.28421 "Abstract") [[pdf](https://arxiv.org/pdf/2608.28421 "Download PDF"), [html](https://arxiv.org/html/2608.28421v1 "View HTML"), [other](https://arxiv.org/format/2608.28421 "Other formats")] Title: Program Learning with Verifiable Rewards: Symbolic Backpropagation for Post-Training LLMs [Vishvesh Bhat](https://arxiv.org/search/cs?searchtype=author&query=Bhat,+V) Subjects: Artificial Intelligence (cs.AI)