List your company

Preference Classification on Instruction Following (IF) (test)

61.4AccuracyBTPO
51.83254.31656.859.284Oct 17, 2025

Evaluation results

MethodLinks
BTPOBackbone=Qwen2.5-7B-Instruct2025.1061.4
RM-R1Backbone=Qwen2.5-7B-Instruct2025.1052.2
RM-R1-finetunedBackbone=Qwen2.5-7B-Instruct, Fine-tuned=true2025.1052.2
GRPO (pair)Backbone=Qwen2.5-7B-Instruct2025.1052.2