List your company

Reasoning failure prediction and recovery on SAFIM L3

81Accuracythought-tree-based classifier
70.673.37678.7Apr 18, 2026

Evaluation results

MethodLinks
thought-tree-based classifierTarget Reasoning Model=R1, Evaluation Protocol=Generalization from L1 training2026.048117
thought-tree-based classifierTarget Reasoning Model=QwQ, Evaluation Protocol=Generalization from L1 training2026.04714