[ allennlp.tools.create_elmo_embeddings_from_vocab ]
def main( vocab_path: str, elmo_config_path: str, elmo_weights_path: str, output_dir: str, batch_size: int, device: int, use_custom_oov_token: bool = False )
Creates ELMo word representations from a vocabulary file. These
word representations are independent - they are the result of running
the CNN and Highway layers of the ELMo model, but not the Bidirectional LSTM.
ELMo requires 2 additional tokens:
and . The first token
in this file is assumed to be an unknown token.
This script produces two artifacts: A new vocabulary file
and tokens inserted and a glove formatted embedding
file containing word : vector pairs, one per line, with all values
separated by a space.