Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Evidence Retrieval on NQ
Loading...
31.9
Recall@5
Qwen3-Emb-4B + Jina reranker
6.524
13.112
19.7
26.288
May 7, 2026
Recall@5
Recall@10
Recall@20
Updated 26d ago
Evaluation Results
Method
Method
Links
Recall@5
Recall@10
Recall@20
Qwen3-Emb-4B + Jina reranker
parameters=4B, reranke...
2026.05
31.9
42
50.9
Qwen3-Emb-4B
parameters=4B
2026.05
30.3
40
50.5
BGE
variant=BGE-large
2026.05
29.6
39
47.5
INTRA
2026.05
29.1
38.3
45.9
Qwen3-Emb-0.6B
parameters=0.6B
2026.05
25.6
34.4
42
Hybrid RAG
fusion=reciprocal rank...
2026.05
22.9
35.5
49.8
MaxSim
type=late-interaction,...
2026.05
20.9
29.5
39
BM25
type=sparse lexical
2026.05
14
21.9
32.7
TF-IDF
type=sparse lexical
2026.05
7.5
11.3
16.2
Feedback
Search any
task
Search any
task