Said Taghadouini

@staghado

vit.cpp, modernbert, lightonocr

github.com/staghado Joined January 2024

448 Following 551 Followers 539 Posts

Latest posts 5

Pinned post

Said Taghadouini @staghado Oct 8

🚀 LightOnOCR-3 is out! 🦉

This time we go beyond OCR: text transcription, document layout, image descriptions and chart extraction, all in one model.

We release 3 sizes: 0.8B, 1B and 4B, all under Apache 2.0!

Said Taghadouini @staghado Oct 6

you would never guess the expected ground truth for this ParseBench sample!
ocr models are expected to read invisible text now!

Said Taghadouini @staghado Sep 9

our next model does pretty well on this!
h/t @baptaubertin

Yifei Hu Verified account @hu_yifei Sep 3

We trained a document parsing model that's good at differentiating muffin vs chihuahua. We also provide precise bounding boxes for human review.
If your current document parser struggles with muffin vs chihuahua, please consider @reductoai r-1.

View quoted post on X

Said Taghadouini @staghado Aug 14

don't look at the data because if you do you will keep delaying the run.

Said Taghadouini @staghado Jul 13

A good model is not defined by a single benchmark number.

Quality, speed, ease of deployment, and licensing all matter.

Alex Strick van Linschoten Verified account @strickvl Jul 13

New legal dataset up on the @huggingface Hub! Over the weekend I worked to finalise a snapshot of the official legal documents hosted by the Guantanamo Military Commissions. This is mostly court filings and transcripts from all the cases, around ~52GB of data.

Link to the dataset at the bottom of the 🧵

View quoted post on X