List your company

Multi-Hop Question Answering on Bamboogle (EM, F1, SD, SE)

54.4Exact Match (EM)SD-Search-Instruct
1.98415.59229.242.808May 7, 2025Jul 8, 2025Sep 9, 2025Nov 11, 2025Jan 12, 2026Mar 16, 2026May 18, 2026

Evaluation results

MethodLinks
SD-Search-InstructTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0554.4———
SD-Search-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0553.2———
AutoRefine-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0551.2———
Thinker-InstructTraining Pipeline=Synthesize → SFT, External Teacher=Qwen2.5-72B, Backbone=Qwen2.5-7B2026.0548———
StepSearch-BaseTraining Pipeline=Pure RL, External Teacher=GPT-4o, Backbone=Qwen2.5-7B2026.0546.7———
Search-R1-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0545.6———
MR-Search-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0545.2———
StepSearch-baseBackbone=Qwen2.5-7B2026.0442.751.92.0720.7
AutoSearchBackbone=Qwen2.5-7B2026.0441.954.91.5327.4
Search-R1-baseBackbone=Qwen2.5-7B2026.0441.153.42.5616.1
StepSearch-instructBackbone=Qwen2.5-7B2026.0440.352.42.0219.9
StepSearch-InstructTraining Pipeline=Pure RL, External Teacher=GPT-4o, Backbone=Qwen2.5-7B2026.0540———
Search-R1-instructBackbone=Qwen2.5-7B2026.0437.149.62.4215.3
Search-R1-InstructTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0536.8———
HIPRAG-instructBackbone=Qwen2.5-7B2026.0434.747.81.4523.9
StepSearch-instructBackbone=Qwen2.5-3B2026.0433.944.72.0716.4
ZEROSEARCH-baseBackbone=Qwen-2.5-7B, Variant=Base2025.0533.33———
AutoSearchBackbone=Qwen2.5-3B2026.0433.144.71.8318.1
StepSearch-baseBackbone=Qwen2.5-3B2026.043341.93.0310.9
Search-o1Backbone=LLaMA-3.2-3B2025.0532———
Search-R1-baseBackbone=Qwen-2.5-7B, Variant=Base2025.0530.56———
Search-o1Backbone=Qwen-2.5-7B2025.0530.4———
RAgentBackbone=Qwen-2.5-7B2025.0528———
R1-baseBackbone=Qwen-2.5-7B, Variant=Base2025.0527.78———
ZEROSEARCH-instBackbone=Qwen-2.5-7B, Variant=Instruct2025.0527.78———
Search-R1-instBackbone=LLaMA-3.2-3B, Variant=Instruct2025.0527.78———
RAgentBackbone=LLaMA-3.2-3B2025.0526.4———
Search-R1-instBackbone=Qwen-2.5-7B, Variant=Instruct2025.0526.39———
Search-R1-instructBackbone=Qwen2.5-3B2026.042536.91.9812.6
R1-instructBackbone=Qwen-2.5-7B, Variant=Instruct2025.0525———
CoTBackbone=Qwen-2.5-7B2025.0524———
Search-o1Backbone=Qwen-2.5-3B2025.0522.4———
CoTBackbone=LLaMA-3.2-3B2025.0521.6———
R1-instructBackbone=LLaMA-3.2-3B, Variant=Instruct2025.0520.83———
Search-R1-baseBackbone=LLaMA-3.2-3B, Variant=Base2025.0519.44———
ZEROSEARCH-instBackbone=LLaMA-3.2-3B, Variant=Instruct2025.0518.06———
RAGBackbone=Qwen-2.5-7B2025.0516.8———
Direct AnswerBackbone=Qwen-2.5-3B2025.0516.8———
Search-R1-baseBackbone=Qwen2.5-3B2026.0416.125.91.699.5
ZEROSEARCH-baseBackbone=LLaMA-3.2-3B, Variant=Base2025.0515.28———
RAGBackbone=Qwen-2.5-3B2025.0515.2———
Direct AnswerBackbone=Qwen-2.5-7B2025.0514.4———
ZEROSEARCH-baseBackbone=Qwen-2.5-3B, Variant=Base2025.0513.89———
RAgentBackbone=Qwen-2.5-3B2025.0513.6———
CoTBackbone=Qwen-2.5-3B2025.0512.8———
Search-R1-baseBackbone=Qwen-2.5-3B, Variant=Base2025.0512.5———
Search-R1-instBackbone=Qwen-2.5-3B, Variant=Instruct2025.0512.5———
RAGBackbone=LLaMA-3.2-3B2025.0511.2———
R1-instructBackbone=Qwen-2.5-3B, Variant=Instruct2025.0511.11———
ZEROSEARCH-instBackbone=Qwen-2.5-3B, Variant=Instruct2025.0511.11———
R1-baseBackbone=LLaMA-3.2-3B, Variant=Base2025.0511.11———
HIPRAG-baseBackbone=Qwen2.5-3B2026.0410.516.81.726.11
Direct AnswerBackbone=LLaMA-3.2-3B2025.058———
R1-baseBackbone=Qwen-2.5-3B, Variant=Base2025.055.56———
HIPRAG-instructBackbone=Qwen2.5-3B2026.04411.91.562.57