List your company

Translator-call mode selection on MMLU ProX Lite

76.45Macro F1LUAR
39.956449.430758.90568.3793Jun 1, 2026

Evaluation results

MethodLinks
LUARBackbone=Qwen3-4B2026.0676.45
LUARBackbone=Qwen3-8B2026.0668.18
ST(qr)Backbone=Qwen3-4B2026.0665.44
BOUNDARY-SFTBackbone=Qwen3-4B2026.0665.21
ST(qr)Backbone=Qwen3-8B2026.0661.13
NATIVE-TOOL-USEBackbone=Qwen3-4B2026.0660.03
BOUNDARY-SFTBackbone=Qwen3-8B2026.0659.47
ST(q)Backbone=Qwen3-4B2026.0648.38
SELF-ASSESSMENTBackbone=Qwen3-8B2026.0648.04
SELF-ASSESSMENTBackbone=Qwen3-4B2026.0647.83
ST(q)Backbone=Qwen3-8B2026.0646.8
NATIVE-TOOL-USEBackbone=Qwen3-8B2026.0641.36