Skip to content
Trending storyDeveloping

Jerry Liu says Unlimited OCR scores 46.2% on ParseBench, ranking #115

1 article1 sourcesince Oct 11Last article 1h ago ·

Overview

AISummary of 1 article

Jerry Liu says Unlimited OCR, a low-parameter OCR model, scores about 46.2% on ParseBench and ranks #115 there.

He calls this a good milestone with room to improve. In his account, the model processes documents quickly and reads order and layout well, slightly beats PaddleOCR on tables, and slightly trails it on layout.

Liu says the model ignores visual elements, formatting, and the most complex tables. The source is a single post by Liu on X, and no independent benchmark result is reported.

Written by AI from the articles below · updated Oct 11, 5:23 PM ET

Check the sources:

Latest developmentOct 11, 5:12 PM ET
Jerry Liu says Unlimited OCR ranks #115 on ParseBench at 46.2%

Article timeline

The articles in this story. Times are ET.

Oct 11
  1. Jerry LiuX
    Jerry Liu says Unlimited OCR ranks #115 on ParseBench at 46.2%

    AIJerry Liu says Unlimited OCR, a low-parameter OCR model, processes documents very quickly and reads order and layout well. He says it slightly beats PaddleOCR on tables and slightly trails it on layout, but ignores visual elements, formatting, and the most complex tables. He reports it scores about 46.2% and ranks #115 on ParseBench, which he calls a good milestone with room to improve.

    Image from @jerryjliu0's post

Heat trend

Not enough continuous observations to show a trend yet.