List your company

Multi-Hop Question Answering on Bamboogle (token-level F1)

65.2Token-Level F1Oracle
3.42419.46235.551.538Apr 19, 2026

Evaluation results

MethodLinks
OracleMain agent backbone=Qwen2.5-7B-Instruct2026.0465.2
COSEARCHMain agent backbone=Qwen2.5-7B-Instruct2026.0460.1
OracleMain agent backbone=Qwen2.5-3B-Instruct2026.0458.9
Fixed RankerMain agent backbone=Qwen2.5-7B-Instruct2026.0458.8
Retrieval OnlyMain agent backbone=Qwen2.5-7B-Instruct2026.0457.4
ZeroSearchMain agent backbone=Qwen2.5-7B-Instruct2026.0457.2
Search-R1Main agent backbone=Qwen2.5-7B-Instruct2026.0455.8
COSEARCHMain agent backbone=Qwen2.5-3B-Instruct2026.0445.7
Fixed RankerMain agent backbone=Qwen2.5-3B-Instruct2026.0444.8
Retrieval OnlyMain agent backbone=Qwen2.5-3B-Instruct2026.0443.6
ZeroSearchMain agent backbone=Qwen2.5-3B-Instruct2026.0443.1
Search-o1Main agent backbone=Qwen2.5-7B-Instruct2026.0442.2
Search-R1Main agent backbone=Qwen2.5-3B-Instruct2026.0441.4
Direct InferenceMain agent backbone=Qwen2.5-7B-Instruct2026.0437.2
CoTMain agent backbone=Qwen2.5-7B-Instruct2026.0431
RAGMain agent backbone=Qwen2.5-7B-Instruct2026.0429.6
Search-o1Main agent backbone=Qwen2.5-3B-Instruct2026.0421.6
CoTMain agent backbone=Qwen2.5-3B-Instruct2026.0410
Direct InferenceMain agent backbone=Qwen2.5-3B-Instruct2026.046.8
RAGMain agent backbone=Qwen2.5-3B-Instruct2026.045.8