def test():
encoder_test = sm.EncoderRNN(10, 10, 2)
decoder_test = sm.AttnDecoderRNN('general', 10, 10, 2)
if torch.cuda.is_available():
encoder_test.cuda()
decoder_test.cuda()
encoder_hidden = encoder_test.init_hidden()
word_input = cuda_variable(torch.LongTensor([1, 2, 3]))
encoder_outputs, encoder_hidden = encoder_test(word_input, encoder_hidden)
print(encoder_outputs.size())
word_target = cuda_variable(torch.LongTensor([1, 2, 3]))
decoder_attns = torch.zeros(1, 3, 3)
decoder_hidden = encoder_hidden
decoder_context = cuda_variable(torch.zeros(1, decoder_test.hidden_size))
for c in range(len(word_target)):
decoder_output, decoder_context, decoder_hidden, decoder_attn = \
decoder_test(word_target[c], decoder_context,
decoder_hidden, encoder_outputs)
print(decoder_output.size(), decoder_hidden.size(), decoder_attn.size())
decoder_attns[0, c] = decoder_attn.squeeze(0).cpu().data
output_dist = output.data.view(-1).div(temperature).exp()
top_i = torch.multinomial(output_dist, 1)[0]
# Stop at the EOS
if top_i is EOS_token:
break
predicted_char = chr(top_i)
predicted += predicted_char
dec_input = str2tensor(predicted_char)
return enc_input, predicted
encoder = sm.EncoderRNN(N_CHARS, HIDDEN_SIZE, N_LAYERS)
decoder = sm.DecoderRNN(HIDDEN_SIZE, N_CHARS, N_LAYERS)
if torch.cuda.is_available():
decoder.cuda()
encoder.cuda()
print(encoder, decoder)
test()
params = list(encoder.parameters()) + list(decoder.parameters())
optimizer = torch.optim.Adam(params, lr=0.001)
criterion = nn.CrossEntropyLoss()
train_loader = DataLoader(dataset=TextDataset(),
batch_size=BATCH_SIZE,
shuffle=True,