alea-institute/alea-correction-pairs

ALEA correction pairs v0.2 Spelling, typing and OCR correction examples over clean passages of legal, regulatory, financial and educational text. The errors are injected by code and the answers are the original passages, so every answer is known exactly. No language model was used to build this dataset; about 30% of the passages come from OpenGloss and OpenGloss-Edu, which are model-generated text (see Provenance). config documents ALEA tokens what correction_pairs… See the full description on the dataset page: https://huggingface.co/datasets/alea-institute/alea-correction-pairs.

<iframe src="https://silytics.si/embed/item/a7a2f3ba-2143-4f62-9349-d7db35b19485" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.