Table-R1: Inference-Time Scaling for Table Reasoning

Zheyuan Yang; Lyuhao Chen; Arman Cohan; Yilun Zhao

返回文献探索

Paper · arXiv 2505.23621

Table-R1: Inference-Time Scaling for Table Reasoning

Zheyuan Yang, Lyuhao Chen, Arman Cohan, Yilun Zhao

93 upvotesMay 29, 2025arXiv 预印本

AI 摘要

Two post-training strategies, distillation and RLVR, enable inference-time scaling in table reasoning tasks, resulting in a model (Table-R1-Zero) that matches GPT-4.1's performance using fewer parameters and shows strong generalization.

distillationreinforcement learningverifiable rewardsRLVRreasoning tracesDeepSeek-R1LLMsTable-R1-SFTGRPOTable-R1-Zeroshort-form QAfact verificationfree-form QAinstruction tuningmodel architecture choicescross-task generalizationtable reasoning skills

Abstract

In this work, we present the first study to explore inference-time scaling on table reasoning tasks. We develop and evaluate two post-training strategies to enable inference-time scaling: distillation from frontier model reasoning traces and reinforcement learning with verifiable rewards (RLVR). For distillation, we introduce a large-scale dataset of reasoning traces generated by DeepSeek-R1, which we use to fine-tune LLMs into the Table-R1-SFT model. For RLVR, we propose task-specific verifiable reward functions and apply the GRPO algorithm to obtain the Table-R1-Zero model. We evaluate our Table-R1-series models across diverse table reasoning tasks, including short-form QA, fact verification, and free-form QA. Notably, the Table-R1-Zero model matches or exceeds the performance of GPT-4.1 and DeepSeek-R1, while using only a 7B-parameter LLM. It also demonstrates strong generalization to out-of-domain datasets. Extensive ablation and qualitative analyses reveal the benefits of instruction tuning, model architecture choices, and cross-task generalization, as well as emergence of essential table reasoning skills during RL training.