List your company

Knowledge Question Answering on HLE

4.36AccuracyKTAE
2.57123.03563.53.9644Aug 5, 2025

Evaluation results

MethodLinks
KTAEModel Scale=Qwen3-1.7B2025.084.36
VRPOModel Scale=Qwen3-1.7B2025.084.22
GRPOModel Scale=Qwen3-1.7B2025.083.94
PPOModel Scale=Qwen3-1.7B2025.083.94
λ-GRPOModel Scale=Qwen3-1.7B2025.083.89
VRPOModel Scale=Qwen3-8B2025.083.89
λ-GRPOModel Scale=Qwen3-8B2025.083.57
Reinforce++Model Scale=Qwen3-1.7B2025.083.52
PPOModel Scale=Qwen3-8B2025.083.29
Reinforce++Model Scale=Qwen3-8B2025.083.29
GRPOModel Scale=Qwen3-8B2025.083.24
Dr.GRPOModel Scale=Qwen3-1.7B2025.083.1
Dr.GRPOModel Scale=Qwen3-8B2025.083.1
BaseModel Scale=Qwen3-8B2025.082.89
BaseModel Scale=Qwen3-1.7B2025.082.87
KTAEModel Scale=Qwen3-8B2025.082.64