Exploiting Pre-trained Encoder-Decoder Transformers for Sequence-to-Sequence Constituent Parsing

Open paper
Benchmark Model Rank Results
constituency-parsing-on-penn-treebankBART base + lexicalized bottom-up–F1 score: 94.52
constituency-parsing-on-penn-treebankBART base + lexicalized in-order–F1 score: 95.61
constituency-parsing-on-penn-treebankBART large + lexicalized in-order–F1 score: 96.24
constituency-parsing-on-penn-treebankBART large + lexicalized top-down–F1 score: 96.23
constituency-parsing-on-penn-treebankT5 large + lexicalized in-order–F1 score: 95.45

Rank counts only results with a code link.