Ondřej Dušek

e2e-metrics. E2E NLG Challenge Evaluation metrics

e2e-dataset. The E2E NLG Challenge Dataset

e2e-cleaning. Cleaned E2E NLG Challenge data + supporting scripts

ratpred. Referenceless quality estimation for natural language generation – predicting human ratings

context-menu. A simple Windows application that displays a ContextMenu for a given file

qsubmit. Wrapper for batch engine submission commands (remake of Obo's Perl qsubmit)

e2e-eval. E2E NLG Challenge – outputs of participating systems & raw human ratings

GEM-metrics. Automatic metrics for GEM tasks

qload. Son of Grid Engine/Sun Grid Engine queue load checker (parsing qstat)

multi_loss_example. A simple example for multiple losses with PyTorch

en-deep. MLProcess – a framework for batch parallel processing of various NLP tasks

customize. My Linux settings

e2e_challenge_web. E2E NLG Challenge Webpage

e2e-stats. Helper scripts used to compute statistics for the E2E NLG dataset + other datasets from the domain (BAGEL, SFREST)

wmt-trueskill. Data and code used in the 2014 WMT, "Efficient Elicitation of Annotations for Human Evaluation of Machine Translation"