keras - 简单的 Keras LSTM 模型的 PyTorch 版本

Question

尝试将 Keras 中的简单 LSTM 模型转换为 PyTorch 代码。 Keras 模型仅在 200 个 epoch 后收敛，而 PyTorch 模型:

需要更多的 epoch 才能达到相同的损失水平(200 对 ~8000)

似乎过度拟合输入，因为预测值不接近 100

这是 Keras 代码:

from numpy import array
from keras.models import Sequential
from keras.layers import LSTM
from keras.layers import Dense

X = array([10,20,30,20,30,40,30,40,50,40,50,60,50,60,70,60,70,80]).reshape((6,3,1))
y = array([40,50,60,70,80,90])
model = Sequential()
model.add(LSTM(50, activation='relu', recurrent_activation='sigmoid',  input_shape=(3, 1)))
model.add(Dense(1))
model.compile(optimizer='adam', loss='mse')
model.fit(X, y, epochs=200, verbose=1)
x_input = array([70, 80, 90]).reshape((1, 3, 1))
yhat = model.predict(x_input, verbose=0)
print(yhat)

这是等效的 PyTorch 代码:

from numpy import array
import torch
import torch.nn as nn
import torch.nn.functional as F

X = torch.tensor([10,20,30,20,30,40,30,40,50,40,50,60,50,60,70,60,70,80]).float().reshape(6,3,1)
y = torch.tensor([40,50,60,70,80,90]).float().reshape(6,1)

class Model(nn.Module):
  def __init__(self):
    super(Model, self).__init__()
    self.lstm = nn.LSTM(input_size=1, hidden_size=50, num_layers=1, batch_first=True)
    self.fc = nn.Linear(50, 1)

  def forward(self, x):
    batches = x.size(0)
    h0 = torch.zeros([1, batches, 50])
    c0 = torch.zeros([1, batches, 50])
    (x, _) = self.lstm(x, (h0, c0))
    x = x[:,-1,:]  # Keep only the output of the last iteration. Before shape (6,3,50), after shape (6,50)
    x = F.relu(x)
    x = self.fc(x)
    return x

model = Model()
criterion = nn.MSELoss()
optimizer = torch.optim.Adam(model.parameters())

n_epochs = 8000
for epoch in range(n_epochs):
  model.train()
  optimizer.zero_grad()
  y_ = model(X)
  loss = criterion(y_, y)
  loss.backward()
  optimizer.step()
  print(f"Epoch {epoch+1}/{n_epochs}, loss = {loss.item()}")

model.eval()
x_input = torch.tensor([70, 80, 90]).float().reshape((1, 3, 1))
yhat = model(x_input)
print(yhat)

唯一可能的区别是初始权重和偏差值，但我不认为稍微不同的权重和偏差可能会导致如此大的行为差异。
我在 PyTorch 代码中遗漏了什么？

Answer 1

行为差异是因为 LSTM API 中的激活函数。通过将激活更改为 tanh，我也可以在 Keras 中重现该问题。

model.add(LSTM(50, activation= 'tanh' , recurrent_activation='sigmoid', input_shape=(3, 1)))

pytorch LSTM API 中没有将激活函数更改为“relu”的选项。
https://pytorch.org/docs/stable/nn.html#lstm

从这里获取 LSTM 实现，https://github.com/huggingface/torchMoji/blob/master/torchmoji/lstm.py
并将 hardsigmoid/tanh 更改为 sigmoid/relu，模型也会在 pytorch 中收敛。