stepfun-ai

/

GOT-OCR2_0

like 1.09k

This python file named 'modeling_got.py' is modified in order to run the model(GOT OCR2.0) on CPU instead of CUDA.This file is taken from the the original repo i.e., the file named 'modeling_GOT.py' and modified to make it run on CPU.
NOTE: The inference time is long (5-7 minutes) per image for plain ocr, as per my observation while running the inference.
This inference was done on the system MacBook Air M1(2020) with 8 GB ram, python 3.12.3(venv) on Visual Studio Code.