# Print iter number, loss, name and guess
if iter % print_every == 0:
guess, guess_i = char_from_output(output)
correct = '✓' if guess == target else '✗ (%s)' % target
print('%d %d%% (%s) %.4f %s / %s %s' % (iter, iter / n_iters * 100, timeSince(start), loss, sequence, guess, correct))
# Add current loss avg to list of losses
if iter % plot_every == 0:
all_losses.append(current_loss / plot_every)
current_loss = 0
if args.sanity:
print("-"*20)
print("Testing generation")
hidden = rnn.initHidden()
start = Variable(sequence_to_tensor("he"))
_, hidden = rnn(start[0], hidden)
output, hidden = rnn(start[1], hidden)
input = "he"
print("{}".format(input),end='')
for i in range(50):
hidden = rnn.initHidden()
start = Variable(sequence_to_tensor(input))
_, hidden = rnn(start[0], hidden)
output, hidden = rnn(start[1], hidden)
guess, _ = char_from_output(output)
input = input[1]+guess
print("{}".format(guess),end='')
torch.save(rnn.state_dict(),".saved_model")
print(
f'There are {n_categories} languages.\nNumber of family name per language:'
)
for categ in train_data.keys():
print(' {}\t {}'.format(categ, len(train_data[categ])))
### create model
n_hidden = 128
model_net = RNN(n_letters, n_hidden, n_categories)
### Tensorboard visualization of the network
# - create (any valid) input data
# - visualize the built model in tensorboeard
category, line, category_tensor, line_tensor = randomTrainingExample(
all_categories, train_data)
hidden = model_net.initHidden()
tb_writer.add_graph(model_net, (line_tensor[0], hidden))
# tb_writer.close()
#### training
criterion = torch.nn.NLLLoss() # the RNN already has a softmax as output
learning_rate = 0.005
def train(category_tensor, line_tensor):
hidden = model_net.initHidden()
model_net.zero_grad()
for i in range(line_tensor.size()[0]):
output, hidden = model_net(line_tensor[i], hidden)