AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

From Shortcuts to Reasoning: Robust Post-Training of Theory of Mind with Reinforcement Learning

Jike Zhong, Yuxiang Lai, Ming Li, Yuheng Li, Wuao Liu, Behzad Dariush, Konstantinos Psounis, Shao-Yuan Lo

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning for reasoning data · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace SFT with Thinking-RFTQwen2.5-7B-Instruct · OpenToM overallaccuracy (%): 83.14 -> 89.14Improved
add explicit thinking to RFT (ablation)No-Thinking-RFT · OpenToM overall (7B)accuracy (%): 78.71 -> 89.14Improved
replace SFT with Thinking-RFTQwen2.5-VL-7B-Instruct · MMToM / MuMA-ToM averageaccuracy (%): 74.75 -> 82.20Improved
replace SFT with Thinking-RFT on shortcut-prone ExploreToMQwen2.5-7B-Instruct · ExploreToM in-domainaccuracy (%): 95.8 -> 94.3Hurt
train on shortcut-prone ExploreToMQwen2.5-7B-Instruct · Hi-ToM OODaccuracy (%): 43.6 -> 35.3Hurt
scale model from 3B to 7B on shortcut-prone ExploreToMSFT · ExploreToM in-domainaccuracy (%): 96.4 -> 95.8No clear effect
scale model from 3B to 7B on shortcut-free OpenToMThinking-RFT · OpenToM overallaccuracy (%): 83.0 -> 89.1Improved
replace SFT with Thinking-RFT for second-order generalization (train on first order)Qwen2.5-7B-Instruct · OpenToM second-orderaccuracy (%): 65.33 -> 74.33Improved
replace SFT with Thinking-RFT for unseen environmentsQwen2.5-VL-7B-Instruct · MMToM unseen environmentsaccuracy drop: -7.11 -> -2.4Improved
replace SFT with Thinking-RFT for cross-dataset transferQwen2.5-7B-Instruct · OpenToM->ToMATOaccuracy (%): 56.8 -> 70.4Improved
replace SFT with Thinking-RFT under counterfactual rewritesQwen2.5-7B-Instruct · OpenToM counterfactualaccuracy (%): 66.00 -> 81.24Improved
replace SFT with Thinking-RFT on first-order OpenToMQwen2.5-7B-Instruct · OpenToM first-orderaccuracy (%): 93.00 -> 95.00No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy