3 ms·
Wouldn't that just bias itself to the shape of the text extracted from the OCR against the shape of the raw text alone? It doesn't seem like it would be a great
by someothherguyy 2y ago
Wouldn't that just bias itself to the shape of the text extracted from the OCR against the shape of the raw text alone? It doesn't seem like it would be a great benchmark for estimating semantic accuracy?