Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
End-to-end RAG Generation on MS MARCO and Natural Questions
Loading...
50.1
F1 (QA)
SPI
44.588
46.019
47.45
48.881
Nov 12, 2025
F1 (QA)
BLEU-4
ROUGE-L
BERTScore
Updated 19d ago
Evaluation Results
Method
Method
Links
F1 (QA)
BLEU-4
ROUGE-L
BERTScore
SPI
Latency (ms)=92, Memor...
2025.11
50.1
47.8
44.2
87.9
Atlas
Latency (ms)=150, Memo...
2025.11
48.3
45.2
42.7
86
Retro
Latency (ms)=140, Memo...
2025.11
47.9
44.7
42.3
85.4
HyDE-RAG
Latency (ms)=155, Memo...
2025.11
47.8
44.9
42.1
85.6
ColBERTv2-RAG
Latency (ms)=128, Memo...
2025.11
47.2
44.3
41.8
85.2
SPLADE-RAG
Latency (ms)=115, Memo...
2025.11
46.5
43.1
40.6
84.1
DPR-RAG
Latency (ms)=145, Memo...
2025.11
44.8
41.2
38.9
82.3
Feedback
Search any
task
Search any
task