System Description for the CommonGen task with the POINTER model
Anna Shvets · 2021
In a current experiment we were testing Com-monGen dataset for structure-to-text task from GEM living benchmark with the constraint based POINTER model.POINTER represents a hybrid architecture, combining insertionbased and transformer paradigms, predicting the token and the insertion position at the same time.The text is therefore generated gradually in a parallel non-autoregressive manner, given the set of keywords.The pretrained model was fine-tuned on a training split of the Common-Gen dataset and the generation result was compared to the validation and challenge splits. 1 The received metrics outputs, which measure lexical equivalence, semantic similarity and diversity, are discussed in details in a present system description.