olmOCR 2 brings local, searchable transcription to archives
Ai2 reports that researchers are adapting the open olmOCR 2 model to extract searchable text from historical and archaeological document images, including handwriting and complex layouts. In the CENTURIA evaluation, LoRA fine-tuning on 57 records reduced transcription error from 15.58% to 1.11% across 450 test records, while ArtificeOCR uses the model locally with Tropy.
The reported adaptations demonstrate that olmOCR 2 can support locally deployed, expert-reviewed transcription workflows for institutions that cannot send archival materials to commercial APIs.
How researchers are adapting olmOCR 2 for historical and archaeological archives
October 28, 2026
Ai2
olmOCR 2 is Ai2’s open model for extracting searchable, structured text from document images, including handwriting and complex layouts. Its weights, fine-tuning data, and code let researchers run it locally and adapt it to their research needs.
#olmOCR 2#Ai2#ArtificeOCR#Tropy#CENTURIA#LEGION#TU Wien Computer Vision Lab#Austrian Archaeological Institute#Austrian Academy of Sciences#LoRA