def initialize_models(voc):
print('Building encoder and decoder ...')
# initialize word embeddings
embedding = nn.Embedding(voc.num_words, c.HIDDEN_SIZE)
if c.LOAD_FILENAME:
embedding.load_state_dict(embedding_sd)
# initialize encoder & decoder models
encoder = EncoderRNN(c.HIDDEN_SIZE, embedding, c.ENCODER_N_LAYERS,
c.DROPOUT)
decoder = LuongAttnDecoderRNN(c.ATTN_MODEL, embedding, c.HIDDEN_SIZE,
voc.num_words, c.DECODER_N_LAYERS, c.DROPOUT)
if c.LOAD_FILENAME:
encoder.load_state_dict(encoder_sd)
decoder.load_state_dict(decoder_sd)
# Use appropriate device
encoder = encoder.to(c.DEVICE)
decoder = decoder.to(c.DEVICE)
print('Models built and ready to go!')
return embedding, encoder, decoder
def main():
"""primary entry
"""
voc, pairs = loadPreparedData()
print('Building encoder and decoder ...')
# Initialize word embeddings
#embedding = nn.Embedding(voc.num_words, params.hidden_size)
embedding = nn.Embedding(voc.num_words, params.embedding_size)
# Initialize encoder & decoder models
encoder = EncoderRNN(embedding, params.hidden_size,
params.encoder_n_layers, params.dropout)
decoder = LuongAttnDecoderRNN(params.attn_model, embedding,
params.hidden_size, voc.num_words,
params.decoder_n_layers, params.dropout)
# Use appropriate device
encoder = encoder.to(params.device)
decoder = decoder.to(params.device)
print('Models built and ready to go!')
# Ensure dropout layers are in train mode
encoder.train()
decoder.train()
# Initialize optimizers
print('Building optimizers ...')
encoder_optimizer = optim.Adam(encoder.parameters(),
lr=params.learning_rate)
decoder_optimizer = optim.Adam(decoder.parameters(),
lr=params.learning_rate *
params.decoder_learning_ratio)
# Run training iterations
print("Starting Training!")
trainIters(voc,
pairs,
encoder,
decoder,
encoder_optimizer,
decoder_optimizer,
embedding,
params.encoder_n_layers,
params.decoder_n_layers,
params.save_dir,
params.n_iteration,
params.batch_size,
params.print_every,
params.save_every,
params.clip,
params.corpus_name,
load_filename=None)
checkpoint = torch.load(ckpt)
encoder_sd = checkpoint['en']
decoder_sd = checkpoint['de']
encoder_optimizer_sd = checkpoint['en_opt']
decoder_optimizer_sd = checkpoint['de_opt']
embedding_sd = checkpoint['embedding']
voc.__dict__ = checkpoint['voc_dict']
# Initialize word embeddings
embedding = nn.Embedding(voc.num_words, hidden_size)
embedding.load_state_dict(embedding_sd)
# Initialize encoder & decoder models
encoder = EncoderRNN(hidden_size, embedding, encoder_n_layers, dropout)
decoder = LuongAttnDecoderRNN(attn_model, embedding, hidden_size,
voc.num_words, decoder_n_layers, dropout)
encoder.load_state_dict(encoder_sd)
decoder.load_state_dict(decoder_sd)
encoder = encoder.to(device)
decoder = decoder.to(device)
# Set dropout layers to eval mode
encoder.eval()
decoder.eval()
# Initialize search module
searcher = GreedySearchDecoder(encoder, decoder)
# Begin chatting (uncomment and run the following line to begin)
evaluateInput(encoder, decoder, searcher, voc)
num_workers=4,
collate_fn=collate_fn)
### Initialize model
encoder = EncoderRNN(embedding=src_embedding,
rnn_type=opts.rnn_type,
hidden_size=opts.hidden_size,
num_layers=opts.num_layers,
dropout=opts.dropout,
bidirectional=opts.bidirectional)
decoder = LuongAttnDecoderRNN(encoder, embedding=tgt_embedding,
attention=opts.attention,
tie_embeddings=opts.tie_embeddings,
dropout=opts.dropout,
tie_ext_feature=opts.tie_ext_feature,
ext_rate_embedding=ext_rate_emebdding,
ext_appcate_embedding=ext_appcate_embedding, ext_seqlen_embedding=ext_seqlen_embedding, ext_senti_embedding=ext_senti_embedding
)
print("emb start")
if opts.pretrained_embeddings:
if opts.use_keyword:
train_embedding = Embedding(filename=args.glove_filename, embedding_size=embedding_size).load_embedding_plus_keyword(train_dataset.src_vocab)
target_embedding = Embedding(filename=args.glove_filename, embedding_size=embedding_size).load_embedding_plus_keyword(
train_dataset.tgt_vocab)
else:
train_embedding = Embedding(filename=args.glove_filename, embedding_size=embedding_size).load_embedding(train_dataset.src_vocab)
target_embedding = Embedding(filename=args.glove_filename, embedding_size=embedding_size).load_embedding(train_dataset.tgt_vocab)
encoder.embedding.weight.data.copy_(train_embedding)
decoder.embedding.weight.data.copy_(target_embedding)
#checkpoint = torch.load(loadFilename, map_location=torch.device('cpu'))
encoder_sd = checkpoint['en']
decoder_sd = checkpoint['de']
encoder_optimizer_sd = checkpoint['en_opt']
decoder_optimizer_sd = checkpoint['de_opt']
embedding_sd = checkpoint['embedding']
voc.__dict__ = checkpoint['voc_dict']
print('Building encoder and decoder ...')
# Initialize word embeddings
embedding = nn.Embedding(voc.num_words, hidden_size)
if loadFilename:
embedding.load_state_dict(embedding_sd)
# Initialize encoder & decoder models
encoder = EncoderRNN(hidden_size, embedding, encoder_n_layers, dropout)
decoder = LuongAttnDecoderRNN(attn_model, embedding, hidden_size,
voc.num_words, decoder_n_layers, dropout)
if loadFilename:
encoder.load_state_dict(encoder_sd)
decoder.load_state_dict(decoder_sd)
# Use appropriate device
encoder = encoder.to(device)
decoder = decoder.to(device)
print('Models built and ready to go!')
# Ensure dropout layers are in train mode
encoder.train()
decoder.train()
# Initialize optimizers
print('Building optimizers ...')
encoder_optimizer = optim.Adam(encoder.parameters(), lr=learning_rate)