I was following the the tutorial on PyTorch website for seq2seq modelling, following are the parts of the code I am using:
class EncoderRNN(nn.Module):
def __init__(self, input_size, hidden_size):
super(EncoderRNN, self).__init__()
self.hidden_size = hidden_size
self.embedding = nn.Embedding(input_size, hidden_size)
self.gru = nn.GRU(hidden_size, hidden_size)
def forward(self, input, hidden):
embedded = self.embedding(input).view(1, 1, -1)
output = embedded
output, hidden = self.gru(output, hidden)
return output, hidden
def initHidden(self):
return torch.zeros(1, 1, self.hidden_size, device=device)
class AttnDecoderRNN(nn.Module):
def __init__(self, hidden_size, output_size, dropout_p=0.1, max_length=MAX_LENGTH):
super(AttnDecoderRNN, self).__init__()
self.hidden_size = hidden_size
self.output_size = output_size
self.dropout_p = dropout_p
self.max_length = max_length
self.embedding = nn.Embedding(self.output_size, self.hidden_size)
self.attn = nn.Linear(self.hidden_size * 2, self.max_length)
self.attn_combine = nn.Linear(self.hidden_size * 2, self.hidden_size)
self.dropout = nn.Dropout(self.dropout_p)
self.gru = nn.GRU(self.hidden_size, self.hidden_size)
self.out = nn.Linear(self.hidden_size, self.output_size)
def forward(self, input, hidden, encoder_outputs):
embedded = self.embedding(input).view(1, 1, -1)
embedded = self.dropout(embedded)
attn_weights = F.softmax(
self.attn(torch.cat((embedded[0], hidden[0]), 1)), dim=1)
attn_applied = torch.bmm(attn_weights.unsqueeze(0),
encoder_outputs.unsqueeze(0))
output = torch.cat((embedded[0], attn_applied[0]), 1)
output = self.attn_combine(output).unsqueeze(0)
output = F.relu(output)
output, hidden = self.gru(output, hidden)
output = F.log_softmax(self.out(output[0]), dim=1)
return output, hidden, attn_weights
def initHidden(self):
return torch.zeros(1, 1, self.hidden_size, device=device)
Once the encoder and decoder have been defined, the training proceeds as follows:
def train(input_tensor, target_tensor, encoder, decoder, encoder_optimizer, decoder_optimizer, criterion, max_length=MAX_LENGTH):
encoder_hidden = encoder.initHidden()
encoder_optimizer.zero_grad()
decoder_optimizer.zero_grad()
input_length = input_tensor.size(0)
target_length = target_tensor.size(0)
encoder_outputs = torch.zeros(max_length, encoder.hidden_size, device=device)
loss = 0
for ei in range(input_length):
encoder_output, encoder_hidden = encoder(input_tensor[ei], encoder_hidden)
encoder_outputs[ei] = encoder_output[0, 0]
decoder_input = torch.tensor([[SOS_token]], device=device)
decoder_hidden = encoder_hidden
for di in range(target_length):
decoder_output, decoder_hidden, decoder_attention = decoder(decoder_input, decoder_hidden, encoder_outputs)
loss += criterion(decoder_output, target_tensor[di])
decoder_input = target_tensor[di] # Teacher forcing
loss.backward()
encoder_optimizer.step()
decoder_optimizer.step()
return loss.item() / target_length
def trainIters(encoder, decoder, n_iters, print_every=1000, plot_every=100, learning_rate=0.01):
encoder_optimizer = optim.SGD(encoder.parameters(), lr=learning_rate)
decoder_optimizer = optim.SGD(decoder.parameters(), lr=learning_rate)
criterion = nn.NLLLoss()
for iter in range(1, n_iters + 1):
input_tensor = torch.from_numpy(np.array([np.random.randint(2) for i in range(10)]).reshape(10,1))
target_tensor = torch.from_numpy(np.array([np.random.randint(2) for i in range(10)]).reshape(10,1))
loss = train(input_tensor, target_tensor, encoder, decoder, encoder_optimizer, decoder_optimizer, criterion)
print("loss===>",loss)
hidden_size = 256
encoder1 = EncoderRNN(50, hidden_size).to(device)
attn_decoder1 = AttnDecoderRNN(hidden_size, 50).to(device)
trainIters(encoder1, attn_decoder1, 10000, print_every=5000)
This code works for translation of sentences. It picks a word, generates its embedding and then gets a hidden state via a GRU in the encoder. This hidden state is then passed to the decoder word by word and for each word there is a vector of probabilities. Then the Negative log likelihood loss is computed given the target label and vector of probabilities.
My question is basically how to adapt this to a time series forecasting model? I have a time series data divided into two parts, sequence 1 and 2. I wish to predict sequence 2. It is clear to me that I need the MSE Loss instead of the classification loss. Also, I believe there is no need to generate embeddings for a particular value in the time series. Could someone experienced with PyTorch please explain how to modify this?
here is my implementation