List your company

General Question Answering on TriviaQA (token-level F1)

76.4Token-Level F1Oracle
32.61643.98355.3566.717Apr 19, 2026

Evaluation results

MethodLinks
OracleMain agent backbone=Qwen2.5-7B-Instruct2026.0476.4
OracleMain agent backbone=Qwen2.5-3B-Instruct2026.0474.9
COSEARCHMain agent backbone=Qwen2.5-7B-Instruct2026.0474.7
Fixed RankerMain agent backbone=Qwen2.5-7B-Instruct2026.0473.5
Retrieval OnlyMain agent backbone=Qwen2.5-7B-Instruct2026.0472.8
COSEARCHMain agent backbone=Qwen2.5-3B-Instruct2026.0471.4
Fixed RankerMain agent backbone=Qwen2.5-3B-Instruct2026.0469.7
ZeroSearchMain agent backbone=Qwen2.5-7B-Instruct2026.0469.3
RAGMain agent backbone=Qwen2.5-7B-Instruct2026.0468.9
Search-R1Main agent backbone=Qwen2.5-7B-Instruct2026.0468.1
Retrieval OnlyMain agent backbone=Qwen2.5-3B-Instruct2026.0465.1
ZeroSearchMain agent backbone=Qwen2.5-3B-Instruct2026.0464.2
Search-R1Main agent backbone=Qwen2.5-3B-Instruct2026.0463.7
Search-o1Main agent backbone=Qwen2.5-7B-Instruct2026.0452.6
Direct InferenceMain agent backbone=Qwen2.5-7B-Instruct2026.0450.4
Search-o1Main agent backbone=Qwen2.5-3B-Instruct2026.0450.1
RAGMain agent backbone=Qwen2.5-3B-Instruct2026.0447.8
CoTMain agent backbone=Qwen2.5-7B-Instruct2026.0445.6
CoTMain agent backbone=Qwen2.5-3B-Instruct2026.0436.3
Direct InferenceMain agent backbone=Qwen2.5-3B-Instruct2026.0434.3