NeuroProlog: Multi-Task Fine-Tuning for Neurosymbolic Mathematical Reasoning via the Cocktail Effect
arXiv:2603.02504v1 Announce Type: new Abstract: Large Language Models (LLMs) achieve strong performance on natural language tasks but remain unreliable in mathematical reasoning, frequently generating fluent …
Pratibha Zunjare, Michael Hsiao
조회수 21회