List your company
ResearchFormerly WizWand

Knowledge-Intensive Reasoning on Bamboogle (F1 score)

73.8F1Llama3.1-8B + ARPO
9.21625.98342.7559.517Dec 11, 2025Jan 8, 2026Feb 6, 2026Mar 7, 2026Apr 4, 2026May 3, 2026Jun 1, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Llama3.1-8B + ARPOBase Model=Llama3.1-8B, Algorithm=ARPO2025.1273.8———
Qwen2.5-7B + ARPOBase Model=Qwen2.5-7B, Algorithm=ARPO2025.1271.5———
Llama3.1-8B + Reinforce ++Base Model=Llama3.1-8B, Algorithm=Reinforce ++2025.1269.1———
Qwen2.5-7B + GRPOBase Model=Qwen2.5-7B, Algorithm=GRPO2025.1268.4———
Llama3.1-8B + GRPOBase Model=Llama3.1-8B, Algorithm=GRPO2025.1268.2———
Llama3.1-8B + DAPOBase Model=Llama3.1-8B, Algorithm=DAPO2025.1267.3———
Qwen2.5-3B + ARPOBase Model=Qwen2.5-3B, Algorithm=ARPO2025.1266.8———
Qwen2.5-3B + Reinforce ++Base Model=Qwen2.5-3B, Algorithm=Reinforce ++2025.1265.7———
Qwen2.5-7B + DAPOBase Model=Qwen2.5-7B, Algorithm=DAPO2025.1265.5———
Qwen2.5-3B + GRPOBase Model=Qwen2.5-3B, Algorithm=GRPO2025.1265.2———
Qwen2.5-7B + Reinforce ++Base Model=Qwen2.5-7B, Algorithm=Reinforce ++2025.1264.9———
Qwen2.5-3B + DAPOBase Model=Qwen2.5-3B, Algorithm=DAPO2025.1264.8———
EAPOBackbone=Llama3.1-8B-Instruct2026.0661.747.92.19—
Llama3.1-8B + TIR PromptingBase Model=Llama3.1-8B, Algorithm=TIR Prompting2025.1258.4———
Reinforce++Backbone=Llama3.1-8B-Instruct2026.0658.423.72.2—
GRPOBackbone=Llama3.1-8B-Instruct2026.0657.442.82.94—
Llama3.1-8BBase Model=Llama3.1-8B, Algorithm=Direct Reasoning2025.1240———
TIRBackbone=Llama3.1-8B-Instruct2026.0630.615.82.42—
BaseBackbone=Llama3.1-8B-Instruct2026.0626.712.5——
Qwen2.5-7BBase Model=Qwen2.5-7B, Algorithm=Direct Reasoning2025.1224———
Qwen2.5-7B + TIR PromptingBase Model=Qwen2.5-7B, Algorithm=TIR Prompting2025.1223.6———
Qwen2.5-3B + TIR PromptingBase Model=Qwen2.5-3B, Algorithm=TIR Prompting2025.1216.4———
Qwen2.5-3BBase Model=Qwen2.5-3B, Algorithm=Direct Reasoning2025.1211.7———
APPOBackbone=Llama3.1-8B-Instruct2026.06———76.8
APPOBackbone=Qwen2.5-7B-Instruct2026.06———77.6
ARPOBackbone=Llama3.1-8B-Instruct2026.06———72.8
ARPOBackbone=Qwen2.5-7B-Instruct2026.06———71.5
CISPOBackbone=Llama3.1-8B-Instruct2026.06———71.8
CISPOBackbone=Qwen2.5-7B-Instruct2026.06———70.2
DAPOBackbone=Llama3.1-8B-Instruct2026.06———67.3
DAPOBackbone=Qwen2.5-7B-Instruct2026.06———65.5
GIGPOBackbone=Llama3.1-8B-Instruct2026.06———72.1
GIGPOBackbone=Qwen2.5-7B-Instruct2026.06———70.1
GPPOBackbone=Llama3.1-8B-Instruct2026.06———71.9
GPPOBackbone=Qwen2.5-7B-Instruct2026.06———72.4
GRPOBackbone=Llama3.1-8B-Instruct2026.06———68.2
GRPOBackbone=Qwen2.5-7B-Instruct2026.06———68.4
Reinforce++Backbone=Llama3.1-8B-Instruct2026.06———69.1
Reinforce++Backbone=Qwen2.5-7B-Instruct2026.06———64.9
TIR PromptingBackbone=Llama3.1-8B-Instruct2026.06———58.4
TIR PromptingBackbone=Qwen2.5-7B-Instruct2026.06———23.6
Zero-shotBackbone=Llama3.1-8B-Instruct2026.06———40
Zero-shotBackbone=Qwen2.5-7B-Instruct2026.06———24