List your company

Multi-task Language Understanding on MMLU (Acc, AUROC, ECE)

64AccuracyC3RL (Ours)
36.33643.51850.757.882Jul 2, 2026

Evaluation results

MethodLinks
C3RL (Ours)Backbone=Llama-3.2-3B-Instruct2026.0764688.9
C3RL w/o RefBackbone=Llama-3.2-3B-Instruct2026.0763.568.88.7
RLVRBackbone=Llama-3.2-3B-Instruct2026.0763.453.720.8
RLCRBackbone=Llama-3.2-3B-Instruct2026.0762.36715.3
SCBackbone=Llama-3.2-3B-Instruct2026.0759.973.315.8
BaseBackbone=Llama-3.2-3B-Instruct2026.0759.658.627.1
SFT+RefBackbone=Llama-3.2-3B-Instruct2026.0752.862.816.2
SaySelfBackbone=Llama-3.2-3B-Instruct2026.0737.471.511