Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Long document retrieval on RepLiQA (full)
Loading...
84.22
F1 Score
AttentionRetriever
22.1008
38.2279
54.355
70.4821
Feb 12, 2026
F1 Score
Updated 1mo ago
Evaluation Results
Method
Method
Links
F1 Score
AttentionRetriever
Backbone=Qwen-2.5 3B
2026.02
84.22
AttentionRetriever
Backbone=LLaMA-3.2 3B
2026.02
83.39
SPScanner
Model Type=Autoregressive
2026.02
64.34
GritLM
Model Type=Dense
2026.02
51.41
BM25
Model Type=Sparse
2026.02
48.96
GTR
Model Type=Dense
2026.02
48.51
Qwen3
Model Type=Dense
2026.02
48.46
GTE-Qwen2
Model Type=Dense
2026.02
47.85
ANCE
Model Type=Dense
2026.02
46.86
DPR
Model Type=Dense
2026.02
36.02
CDE
Model Type=Dense
2026.02
24.49
Feedback
Search any
task
Search any
task
