Lune

ACL Findings2026

Can Post-Training Transform LLMs into Causal Reasoners?

Junqi Chen, Sirui Chen, Chaochao Lu

Abstract

Causal inference is essential for decisionmaking but remains challenging for nonexperts. While large language models (LLMs) show promise in this domain, their precise causal estimation capabilities are still limited, and the impact of post-training on these abilities is insufficiently explored. This paper examines the extent to which post-training can enhance LLMs' capacity for causal inference. We introduce CauGym, a comprehensive dataset comprising seven core causal tasks for training and five diverse test sets. Using this dataset, we systematically evaluate five post-training approaches: SFT, DPO, KTO, PPO, and GRPO. Across five in-domain and four existing benchmarks, our experiments demonstrate that appropriate post-training enables smaller LLMs to perform causal inference competitively, often surpassing much larger models. Our 14Bparameter model achieves 93.5% accuracy on the CaLM benchmark, compared to 55.4% by OpenAI o3. Furthermore, the post-trained LLMs exhibit strong generalization and robustness under real-world conditions such as distribution shifts and noisy data. Collectively, these findings provide the first systematic evidence that targeted post-training can produce reliable and robust LLM-based causal reasoners. Our data and GRPO-model are available at https://github.com/OpenCausaLab/CauGym.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Unveiling Causal Reasoning in Large Language Models: Reality or Mirage? cites Is ChatGPT a Good Causal Reasoner? A Comprehensive Evaluation
  • Unveiling Causal Reasoning in Large Language Models: Reality or Mirage? cites Can Large Language Models Infer Causation from Correlation?
Builds onCited byLeft to right by year; the more cited, the higher.

Builds on5

  1. Using an LLM to Help With Code UnderstandingDaye Nam, Andrew Macvean, Vincent J. Hellendoorn, Bogdan Vasilescu et al.ICSE 2024 · 264 citations
  2. Can Large Language Models Infer Causation from Correlation?Zhijing Jin, Jiarui Liu, Zhiheng Lyu, Spencer Poff et al.ICLR 2024 · 186 citations
  3. Unveiling Causal Reasoning in Large Language Models: Reality or Mirage?Haoang Chi, He Li, Wenjing Yang, Feng Liu et al.NeurIPS 2024 · 124 citations
  4. Is ChatGPT a Good Causal Reasoner? A Comprehensive EvaluationJinglong Gao, Xiao Ding, Bing Qin, Ting LiuEMNLP Findings 2023 · 66 citations
  5. WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-InstructHaipeng Luo, Qingfeng Sun, Can Xu, Pu Zhao et al.ICLR 2025

Related papers

  1. Group Causal Policy Optimization for Post-Training Large Language ModelsZiyin Gu, Jingyao Wang, Ran Zuo, Chuxiong Sun et al.AAAI 2026
  2. Quagmires in SFT-RL Post-Training: When High SFT Scores Mislead and What to Use InsteadFeiyang Kang, Michael Kuchnik, Karthik Padthe, Marin Vlastelica et al.ICLR 2026 · 27 citations
  3. ExpliCa: Evaluating Explicit Causal Reasoning in Large Language ModelsMartina Miliani, Serena Auriemma, Alessandro Bondielli, Emmanuele Chersoni et al.ACL Findings 2025 · 4 citations
  4. CounterBench: Evaluating and Improving Counterfactual Reasoning in Large Language ModelsYuefei Chen, Vivek K. Singh, Jing Ma, Ruixiang TangAAAI 2026 · 1 citation
  5. Playpen: An Environment for Exploring Learning From Dialogue Game FeedbackNicola Horst, Davide Mazzaccara, Antonia Schmidt, Michael Sullivan et al.EMNLP 2025 · 1 citation
Dusk over the sea between two cliffs drawn in fine vertical lines