List your company
ResearchFormerly WizWand

Knowledge-Intensive Reasoning on HQA

87Average ScoreAutoTraj
3.2825.01546.7568.485Jan 30, 2026
Updated 8mo ago

Evaluation results

MethodLinks
AutoTrajCategory=SFT-RL TIR Methods2026.0187
Qwen2.5-7B-InstructFramework=Multi-Dimensional Trajectory Evaluation2026.0185
AutoTIRCategory=RL-only TIR Methods2026.0185
R1-SearcherCategory=RL-only TIR Methods2026.0183
ReSearchCategory=RL-only TIR Methods2026.0182
Tool-Star-SFTCategory=SFT-only TIR Methods2026.0177
Vanilla SFT-RL TIRCategory=SFT-RL TIR Methods2026.0175
Tool-StarCategory=SFT-RL TIR Methods2026.0174
ToRLCategory=RL-only TIR Methods2026.0172
AutoTIRBackbone=Qwen2.5-7B, Training Paradigm=RL-only2026.0130.5
ReSearchBackbone=Qwen2.5-7B, Training Paradigm=RL-only2026.0128.5
AutoTrajBackbone=Qwen2.5-7B, Training Paradigm=SFT–RL TIR, SFT Trajectory Count=13K2026.0128.5
Vanilla SFT-RL TIRBackbone=Qwen2.5-7B, Training Paradigm=SFT–RL TIR2026.0128
Tool-StarBackbone=Qwen2.5-7B, Training Paradigm=SFT–RL TIR, SFT Trajectory Count=54K2026.0127
Tool-Star-SFTBackbone=Qwen2.5-7B, Training Paradigm=SFT-only, SFT Trajectory Count=54K2026.0126.5
Qwen2.5-7B-InstructBackbone=Qwen2.5-7B, Training Paradigm=Instruct2026.0124.5
R1-SearcherBackbone=Qwen2.5-7B, Training Paradigm=RL-only2026.0119.5
ToRLBackbone=Qwen2.5-7B, Training Paradigm=RL-only2026.016.5