Gold-Guided Programmatic Distillation for Financial Reasoning over Hybrid Tables and Text
On the map: LLM reasoning → Reasoning for agents, planning and code
Abstract (arXiv)
Financial question answering over hybrid tabular and textual data may require multi-source reasoning and precise numerical computation. While large language models (LLMs) can generate intermediate reasoning steps, natural-language rationales remain prone to arithmetic errors, making them an unreliable supervision source for distillation. Building on programmatic distillation, we develop an approach that transfers reliable numerical reasoning from a large teacher model to a compact student using execution-verified Python programs instead of free-form textual rationales. It leverages gold derivations to guide teacher-side program synthesis and retains only programs that execute correctly and produce the gold answer, ensuring high-quality supervision. We further introduce an iterative recovery stage that revisits teacher-failed examples, enabling the student to recover and incorporate newly verified programs into training. Experiments on TAT-QA show that our framework is highly effective for hybrid financial reasoning. Our best 7B student achieves 87.00 EM / 87.18 F1 on the test set, substantially outperforming the 72B teacher (78.46 EM) as well as traditional and strong LLM-based baselines, including TAGOP and TAT-LLM. These results demonstrate that execution-verified programmatic distillation provides an effective and extensible framework for training smaller models to perform reliable numerical reasoning.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 13 of its 16 references that are in the map. First posted on arXiv 16 Jul 2026.
Builds on
13 of 16 references.
References · 13
- 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
- 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,115
- 2023Qwen Technical Reportcited by 3,273
- 2022Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Taskscited by 1,057
- 2022STaR: Bootstrapping Reasoning With Reasoningcited by 895
- 2023DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelinescited by 803
- 2023Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizescited by 710
- 2022PAL: Program-aided Language Modelscited by 635
- 2023Scaling Relationship on Learning Mathematical Reasoning with Large Language Modelscited by 417
- 2021TAT-QA: A Question Answering Benchmark on a Hybrid of Tabular and Textual Content in Financecited by 402
- 2022Teaching Small Language Models to Reasoncited by 279
- 2023PaD: Program-aided Distillation Can Teach Small Models Reasoning Better than Chain-of-thought Fine-tuningcited by 29
1 more
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025FinanceReasoning: Benchmarking Financial Numerical Reasoning More Credible, Comprehensive and Challenging
- 2026ASDA: Automated Skill Distillation and Adaptation for Financial Reasoning
- 2026Data-Centric Post-Training for Financial Reasoning: Mining, Distillation, and Verifiable Learning
- 2024FinLLMs: A Framework for Financial Reasoning Dataset Generation with Large Language Models
- 2023BizBench: A Quantitative Reasoning Benchmark for Business and Finance
- 2023Beyond Classification: Financial Reasoning in State-of-the-Art Language Models
- 2024Fine-tuning Smaller Language Models for Question Answering over Financial Documents
- 2025Evaluating Large Language Models for Financial Reasoning: A CFA-Based Benchmark Study