facebookresearch / nougat

Implementation of Nougat Neural Optical Understanding for Academic Documents
https://facebookresearch.github.io/nougat/
MIT License
8.98k stars 567 forks source link

How many params does nougat have? #156

Open brando90 opened 1 year ago

brando90 commented 1 year ago

Based on the README, Nougat has two main models:

The README mentions that the base model uses a Swin Transformer encoder and an mBART decoder. The mBART decoder has 10 layers and the whole architecture has 350M parameters.

The smaller model has a slightly smaller sequence length and only 4 decoder layers, with a total of 250M parameters.

So in summary:

Am I right?

lukas-blecher commented 11 months ago

Yes, correct.

xsank commented 8 months ago

Will the nougat provide large model later?