Enc Brit Glm Ocr: 2,724 OCR Transcriptions of the 1771 Encyclopaedia

Comprising 2,724 OCR transcriptions of the 1771 Encyclopaedia Britannica, generated by davanstrien in February 2026. It utilizes the GLM-OCR 0.9B model to convert historical document images into structured text outputs.

Use Cases

• Fine-tuning OCR models for historical text using the 'markdown' column
• Analyzing 18th-century linguistic patterns within the 'markdown' text
• Benchmarking GLM-OCR performance on archaic fonts found in the 'image' column

Strengths

• Uses SOTA GLM-OCR 0.9B model for transcription
• 2,724 paired image-to-text samples
• Outputs provided in structured markdown format

Limitations

• Small sample size of 2,724 records
• OCR results are model-generated and lack human verification
• Historical 18th-century typography may introduce specific character recognition errors

Provenance

Source

Encyclopaedia Britannica (1771 edition)

Collection Method

Model-generated OCR using the zai-org/GLM-OCR model

Time Range

1771

Freshness

Processed on 2026-02-18 using the GLM-OCR model.