NLP Architect Model Zoo

Model

Description

Links

Sparse GNMT

90% sparse GNMT model and a 2x2 block sparse translating German to English trained on Europarl-v7 1 , Common Crawl and News Commentary 11 datasets

Intent Extraction

A MultiTaskIntentModel intent extraction and slot tagging model, trained on SNIPS NLU dataset

Named Entity Recognition

A NERCRF model trained on CoNLL 2003 dataset

Dependency parser

Graph-based dependency parser using BiLSTM feature extractors

model

Machine comprehension

Match LSTM model trained on SQuAD dataset

Word chunker

A word chunker model trained on CoNLL 2000 dataset

References

1

Europarl-v7: A Parallel Corpus for Statistical Machine Translation, Philipp Koehn, MT Summit 2005