def __init__(self,
options_file: str,
weight_file: str,
requires_grad: bool = False) -> None:
super(_ElmoCharacterEncoder, self).__init__()
with open(cached_path(options_file), 'r') as fin:
self._options = json.load(fin)
self._weight_file = weight_file
self.output_dim = self._options['lstm']['projection_dim']
self.requires_grad = requires_grad
self._load_weights()
ELMoCharacterMapper.set_max_word_length(self._options['char_cnn']['max_characters_per_token'])
ELMoCharacterMapper.number_of_characters(self._options['char_cnn']['n_characters'])
# Cache the arrays for use in forward -- +1 due to masking. -> No need for that anymore (velmo takes care of that)
self._beginning_of_sentence_characters = torch.from_numpy(
numpy.array(ELMoCharacterMapper.beginning_of_sentence_characters) #+ 1
)
self._end_of_sentence_characters = torch.from_numpy(
numpy.array(ELMoCharacterMapper.end_of_sentence_characters) #+ 1
)
def get_batch(self, batch):
# sent in batch in decreasing order of lengths
# batch: (bsize, max_len, word_dim)
embed = np.zeros((len(batch), len(batch[0]), 50))
for i in range(len(batch)):
for j in range(len(batch[i])):
embed[i, j, :] = ELMoCharacterMapper.convert_word_to_char_ids(batch[i][j])
return torch.LongTensor(embed)
def elmo_sent_mapper(sentence, max_length, pad_token="~"):
word_list = []
for i in range(max_length):
word = sentence[i] if i < len(sentence) else pad_token
word_list.append(ELMoCharacterMapper.convert_word_to_char_ids(word))
return word_list
from allennlp.modules.elmo import Elmo
from allennlp.data.token_indexers.elmo_indexer import ELMoCharacterMapper
import torch
import numpy as np
a = torch.autograd.Variable(
torch.LongTensor(
np.reshape(ELMoCharacterMapper.convert_word_to_char_ids("hello"),
[1, 1, 50])))
options_file = "https://s3-us-west-2.amazonaws.com/allennlp/models/elmo/2x4096_512_2048cnn_2xhighway/elmo_2x4096_512_2048cnn_2xhighway_options.json"
weight_file = "https://s3-us-west-2.amazonaws.com/allennlp/models/elmo/2x4096_512_2048cnn_2xhighway/elmo_2x4096_512_2048cnn_2xhighway_weights.hdf5"
elmo_instance = Elmo(options_file, weight_file, 1)
print(elmo_instance(a))