The list below links to the work-in-progress data cards for models submitted to GEM. As part of our submission process, we ask participants a series of questions about their models. The current version of our model cards lists the provided answers verbatim. The submission form can be found here. The template used to produce the statements and can be found here: [download template].
- SimpleNERShared Task 2021Finetuned Transformer architecture for simplification.
- Self-Training, Acceptability Classifiers and Context-ConditioningShared Task 2021BART together with RoBERTa classifiers and context.
- POINTERShared Task 2021POINTER is a hybrid architecture, combining transformers with insertion-based networks.
- NUIG-DSIShared Task 2021Finetuned T5 with additional pretraining data.