batch_loader = BatchLoader('')
parameters = Parameters(batch_loader.max_word_len,
batch_loader.max_seq_len,
batch_loader.words_vocab_size,
batch_loader.chars_vocab_size)
rvae = RVAE(parameters)
if args.use_trained:
rvae.load_state_dict(t.load('trained_RVAE'))
if args.use_cuda:
rvae = rvae.cuda()
optimizer = Adam(rvae.learnable_parameters(), args.learning_rate)
train_step = rvae.trainer(optimizer, batch_loader)
validate = rvae.validater(batch_loader)
ce_result = []
kld_result = []
for iteration in range(args.num_iterations):
cross_entropy, kld, coef = train_step(iteration, args.batch_size, args.use_cuda, args.dropout)
if iteration % 5 == 0:
print('\n')
print('------------TRAIN-------------')
print('----------ITERATION-----------')
print(iteration)
print('--------CROSS-ENTROPY---------')
batch_loader = BatchLoader('')
parameters = Parameters(batch_loader.max_word_len,
batch_loader.max_seq_len,
batch_loader.words_vocab_size,
batch_loader.chars_vocab_size)
rvae = RVAE(parameters)
if args.use_trained:
rvae.load_state_dict(t.load('trained_RVAE'))
if args.use_cuda:
rvae = rvae.cuda()
optimizer = Adam(rvae.learnable_parameters(), args.learning_rate)
train_step = rvae.trainer(optimizer)
# validate = rvae.validater()
ce_result = []
kld_result = []
# training_data = batch_loader.training_data('train')
# validation_data = batch_loader.training_data('valid')
for iteration in range(args.num_iterations):
print(f"-----Iteration: {iteration}-------------")
x = 0
bar = progressbar.ProgressBar(maxval=130001,
widgets=[progressbar.Bar('=', '[', ']'), ' ', progressbar.Percentage()])
bar.start()
while True:
input = batch_loader.next_batch(x, args.batch_size, "train")
with open('train.txt', 'r') as f:
data = f.readlines()
preprocess = Preprocess(embedding_model)
input = preprocess.to_sequence(data)
# embedding=preprocess.embedding()
# np.save('embedding',embedding)
batch_loader = Batch(input, 0.7)
params=Parameter(word_embed_size=300,encode_rnn_size=600,latent_variable_size=1400,\
decode_rnn_size=600,vocab_size=preprocess.vocab_size,embedding_path='embedding.npy')
model = RVAE(params)
model = model.cuda()
optimizer = Adam(model.learnable_parameters(), 1e-3)
train_step = model.trainer(optimizer)
use_cuda = t.cuda.is_available()
ce_list = []
kld_list = []
coef_list = []
test_batch = batch_loader.test_next_batch(1)
for i, batch in enumerate(batch_loader.train_next_batch(1)):
# if i%20==0:
# sample=next(test_batch)
# sentence=model.sample(10,sample,use_cuda)
# sentence=[preprocess.index_to_word(i) for i in sentence]
# print(' '.join(sentence))
# break
ce, kld, coef = train_step(batch, 0.2, use_cuda)
batch_loader_2.max_seq_len,
batch_loader_2.words_vocab_size,
batch_loader_2.chars_vocab_size)
'''=================================================================================================
'''
rvae = RVAE(parameters, parameters_2)
if args.use_trained:
rvae.load_state_dict(t.load('trained_RVAE'))
if args.use_cuda:
rvae = rvae.cuda()
optimizer = Adam(rvae.learnable_parameters(), args.learning_rate)
train_step = rvae.trainer(
optimizer, batch_loader,
batch_loader_2) # batchloader里面是原始句子，batechloader2里面存储的是释义句
validate = rvae.validater(batch_loader, batch_loader_2)
ce_result = []
kld_result = []
start_index = 0
# start_index_2 = 0
for iteration in range(args.num_iterations):
#This needs to be changed ##这一步必须保证不大于训练数据数量-每一批数据的大小，否则越界报错######################
start_index = (start_index + 1) % (49999 - args.batch_size)
#start_index = (start_index+args.batch_size)%149163 #计算交叉熵损失，等
cross_entropy, kld, coef = train_step(iteration, args.batch_size,
args.use_cuda, args.dropout,
