GlotOCR Bench shows that OCR models perform well on fewer than 10 scripts and fail to generalize beyond about 30, with results tracking pretraining coverage and models hallucinating from known scripts on unfamiliar ones.
Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 3roles
background 1polarities
background 1representative citing papers
TypewriterLM is a 7.24B language model pretrained exclusively on pre-1913 English text using a 54B-token corpus, lexically grounded instruction tuning, and the History-Event benchmark for temporal consistency.
OCR for basic letters in Old Church Slavonic can reach 2-3% error rate with modern systems, and a new image-only pipeline of glyph extraction, clustering, and pairwise comparison can produce stemmata for historical manuscripts.
citing papers explorer
-
GlotOCR Bench: OCR Models Still Struggle Beyond a Handful of Unicode Scripts
GlotOCR Bench shows that OCR models perform well on fewer than 10 scripts and fail to generalize beyond about 30, with results tracking pretraining coverage and models hallucinating from known scripts on unfamiliar ones.
-
Pretraining Language Models on Historical Text
TypewriterLM is a 7.24B language model pretrained exclusively on pre-1913 English text using a 54B-token corpus, lexically grounded instruction tuning, and the History-Event benchmark for temporal consistency.
-
The Devil is in the Details -- From OCR for Old Church Slavonic to Purely Visual Stemma Reconstruction
OCR for basic letters in Old Church Slavonic can reach 2-3% error rate with modern systems, and a new image-only pipeline of glyph extraction, clustering, and pairwise comparison can produce stemmata for historical manuscripts.