ELBERT model based on huggingface's albert implementation with early exit mechanism. Based on transformers version 4.31.0
based on the paper: https://arxiv.org/pdf/2107.00175.pdf
and the corresponding github repo: https://github.com/shakeley/ELBERT/tree/master
as part of an academic project with Dr.Roy Schwartz lab at the Hebrew University of Jerusalem: https://github.com/schwartz-lab-NLP
pip install -r requirements.txt
pip install transformers[torch]
and then to run the experiments simply run the main.py file or:
from main import run
run()