Dataset Viewer
Auto-converted to Parquet Duplicate
model
stringclasses
5 values
elo
int64
1.38k
1.56k
params
stringclasses
1 value
wins
int64
57
116
losses
int64
76
134
ties
int64
0
7
win_pct
int64
29
60
elo_low
int64
1.33k
1.51k
elo_high
int64
1.43k
1.6k
glm-ocr
1,556
116
78
0
60
1,513
1,599
lighton-ocr-2
1,538
99
76
0
57
1,496
1,580
nuextract3
1,532
104
82
7
54
1,495
1,573
nuextract3-t0rep
1,491
91
97
6
47
1,451
1,531
nuextract3-think
1,383
57
134
3
29
1,334
1,427

OCR Bench Results: ocr-bench-britannica

VLM-as-judge pairwise evaluation of OCR models. Rankings depend on document type — there is no single best OCR model.

Leaderboard

Rank Model Params ELO 95% CI Wins Losses Ties Win%
1 glm-ocr 1556 1513–1599 116 78 0 60%
2 lighton-ocr-2 1538 1496–1580 99 76 0 57%
3 nuextract3 1532 1495–1573 104 82 7 54%
4 nuextract3-t0rep 1491 1451–1531 91 97 6 47%
5 nuextract3-think 1383 1334–1427 57 134 3 29%

Details

Configs

  • load_dataset("davanstrien/ocr-bench-britannica-nuextract3-sweep") — leaderboard table
  • load_dataset("davanstrien/ocr-bench-britannica-nuextract3-sweep", name="comparisons") — full pairwise comparison log
  • load_dataset("davanstrien/ocr-bench-britannica-nuextract3-sweep", name="metadata") — evaluation run history

Generated by ocr-bench

Downloads last month
36