This directory contains the evaluation scripts for the project.
I let the GroupedSamplingPipeLine translate all the talks from the ted_talks_iwslt dataset.
Then I use the evaluation/evaluatate_translation.py
script to calculate the BERT scores of the translated talks.
The results are saved in the comet ML project.