A small open model that reads one printed table row out of a scanned nineteenth-century meteorological table and returns its cells as text. It is not a chatbot, not a general OCR model, and not a climate model. It reads a strip of paper and writes back numbers. Trained and evaluated entirely on Brazilian historical records, but the interesting result is that it reads layouts it has never seen — and usually with no labelled rows at all. Of the 18 layouts this project describes, two ever needed hand-labelling; the other fifteen were added by writing a JSON description of the columns and running the pipeline. docs/ADDING-A-PUBLICATION.md is the path and the evidence. Base: Qwen/Qwen3.5-2B…
Independent publisher
Gabriel Bueno
akagabi
Models in Library1
Datasets in Library1
Models on Hugging Face2
Followers3
Models
Datasets
Daily meteorological observations from Brazilian observatories, 1883–1890, transcribed from printed nineteenth-century tables by an open 2B model running offline, with every row carrying its provenance and a quality verdict. readings a day), air temperature (mean, max, min, in-shelter and unsheltered), vapour tension, relative humidity, wind direction and force, cloudiness, rainfall, evaporation, ozone, and for Cuyabá the height of the Rio Cuyabá. Each row is one printed line: valuesasprinted (what the page shows, no silent correction), values (the publication's conventions undone in code, e.g. the barometer's elided leading digit restored), markers (words the page prints instead of a…