List your company

Reasoning on AGIEval (Acc, AUROC, ECE)

37.3AccuracyRLVR
18.68423.51728.3533.183Jul 2, 2026

Evaluation results

MethodLinks
RLVRBackbone=Llama-3.2-3B-Instruct2026.0737.30.5220.468
C3RL (Ours)Backbone=Llama-3.2-3B-Instruct2026.0737.30.660.246
C3RL w/o RefBackbone=Llama-3.2-3B-Instruct2026.0737.20.6520.218
BaseBackbone=Llama-3.2-3B-Instruct2026.0734.80.5570.423
RLCRBackbone=Llama-3.2-3B-Instruct2026.0734.40.6580.111
SFT+RefBackbone=Llama-3.2-3B-Instruct2026.0732.10.5940.24
SCBackbone=Llama-3.2-3B-Instruct2026.0728.60.6440.352
SaySelfBackbone=Llama-3.2-3B-Instruct2026.0719.40.6110.072