Given a simple 2 layer neural network, the traditional idea is to compute the gradient w.r.t. the weights/model parameters. For an experiment, I want to compute the gradient of the error w.r.t the input. Are there existing Pytorch methods that can allow me to do this?
More concretely, consider the following neural network:
import torch.nn as nn
import torch.nn.functional as F
class NeuralNet(nn.Module):
def __init__(self, n_features, n_hidden, n_classes, dropout):
super(NeuralNet, self).__init__()
self.fc1 = nn.Linear(n_features, n_hidden)
self.sigmoid = nn.Sigmoid()
self.fc2 = nn.Linear(n_hidden, n_classes)
self.dropout = dropout
def forward(self, x):
x = self.sigmoid(self.fc1(x))
x = F.dropout(x, self.dropout, training=self.training)
x = self.fc2(x)
return F.log_softmax(x, dim=1)
I instantiate the model and an optimizer for the weights as follows:
import torch.optim as optim
model = NeuralNet(n_features=args.n_features,
n_hidden=args.n_hidden,
n_classes=args.n_classes,
dropout=args.dropout)
optimizer_w = optim.SGD(model.parameters(), lr=0.001)
While training, I update the weights as usual. Now, given that I have values for the weights, I should be able to use them to compute the gradient w.r.t. the input. I am unable to figure out how.
def train(epoch):
t = time.time()
model.train()
optimizer.zero_grad()
output = model(features)
loss_train = F.nll_loss(output[idx_train], labels[idx_train])
acc_train = accuracy(output[idx_train], labels[idx_train])
loss_train.backward()
optimizer_w.step()
# grad_features = loss_train.backward() w.r.t to features
# features -= 0.001 * grad_features
for epoch in range(args.epochs):
train(epoch)
It is possible, just set input.requires_grad = True for each input batch you're feeding in, and then after loss.backward() you should see that input.grad holds the expected gradient. In other words, if your input to the model (which you call features in your code) is some M x N x ... tensor, features.grad will be a tensor of the same shape, where each element of grad holds the gradient with respect to the corresponding element of features. In my comments below, I use i as a generalized index - if your parameters has for instance 3 dimensions, replace it with features.grad[i, j, k], etc.
Regarding the error you're getting: PyTorch operations build a tree representing the mathematical operation they are describing, which is then used for differentiation. For instance c = a + b will create a tree where a and b are leaf nodes and c is not a leaf (since it results from other expressions). Your model is the expression, and its inputs as well as parameters are the leaves, whereas all intermediate and final outputs are not leaves. You can think of leaves as "constants" or "parameters" and of all other variables as of functions of those. This message tells you that you can only set requires_grad of leaf variables.
Your problem is that at the first iteration, features is random (or however else you initialize) and is therefore a valid leaf. After your first iteration, features is no longer a leaf, since it becomes an expression calculated based on the previous ones. In pseudocode, you have
f_1 = initial_value # valid leaf
f_2 = f_1 + your_grad_stuff # not a leaf: f_2 is a function of f_1
to deal with that you need to use detach, which breaks the links in the tree, and makes the autograd treat a tensor as if it was constant, no matter how it was created. In particular, no gradient calculations will be backpropagated through detach. So you need something like
features = features.detach() - 0.01 * features.grad
Note: perhaps you need to sprinkle a couple more detaches here and there, which is hard to say without seeing your whole code and knowing the exact purpose.
Related
I was reading this blog from PyTorch. Just before the AutoGrad in training Section , it is mentioned
Be aware that only leaf nodes of the computation have their gradients computed. If you tried, for example, print(c.grad) you’d get back None. In this simple example, only the input is a leaf node, so only it has gradients computed.
Then weights are also considered to be leaf nodes. In the subsequent AutoGrad in training Section this below code block is executed.
BATCH_SIZE = 16
DIM_IN = 1000
HIDDEN_SIZE = 100
DIM_OUT = 10
class TinyModel(torch.nn.Module):
def __init__(self):
super(TinyModel, self).__init__()
self.layer1 = torch.nn.Linear(1000, 100)
self.relu = torch.nn.ReLU()
self.layer2 = torch.nn.Linear(100, 10)
def forward(self, x):
x = self.layer1(x)
x = self.relu(x)
x = self.layer2(x)
return x
some_input = torch.randn(BATCH_SIZE, DIM_IN, requires_grad=False)
ideal_output = torch.randn(BATCH_SIZE, DIM_OUT, requires_grad=False)
model = TinyModel()
When
print(model.layer2.weight.grad)
is executed it is shown as None.
But after training with the below code snippet, the weights have gradients,
optimizer = torch.optim.SGD(model.parameters(), lr=0.001)
prediction = model(some_input)
loss = (ideal_output - prediction).pow(2).sum()
loss.backward()
print(model.layer2.weight.grad[0][0:10])
So is my understanding correct? i.e when weights are initialised by calling TinyModel(), the requires_autograd is set to False. Only when the training starts happening with loss.backward() then the requires_autograd is set to True and the gradient is kept track?
But in other examples, when we create PyTorch models from scratch, where we initialise the weights randomly along with requires_grad=True, the gradient is tracked from beginning.
Or is the gradient tracking generally enabled only when it is started to be trained? If so why initially it was returning None in the above example.
Thank You in advance
I assume by requires_autograd you mean requires_grad.
when weights are initialised by calling TinyModel(), the requires_autograd is set to False.
No, this isn't true. The attribute model.layer2.weight is an instance of nn.Parameter which has requires_grad == True by default. You can verify this yourself:
model = TinyModel()
assert isinstance(model.layer2.weight, nn.Parameter)
assert model.layer2.weight.requires_grad
assert all(p.requires_grad for p in model.parameters())
why initially it was returning None in the above example
The value of model.layer2.weight.grad is None because at that point no gradient is computed yet. In fact, no forward computation is even computed yet. When loss.backward() is executed, the autograd engine computes the gradient of all tensor p with p.requires_grad == True and stores this gradient in p.grad. That's why model.layer2.weight.grad is no longer None after loss.backward().
I have this model in pytorch that I have been using for sequence classification.
class RoBERT_Model(nn.Module):
def __init__(self, hidden_size = 100):
self.hidden_size = hidden_size
super(RoBERT_Model, self).__init__()
self.lstm = nn.LSTM(768, hidden_size, num_layers=1, bidirectional=False)
self.out = nn.Linear(hidden_size, 2)
def forward(self, grouped_pooled_outs):
# chunks_emb = pooled_out.split_with_sizes(lengt) # splits the input tensor into a list of tensors where the length of each sublist is determined by length
seq_lengths = torch.LongTensor([x for x in map(len, grouped_pooled_outs)]) # gets the length of each sublist in chunks_emb and returns it as an array
batch_emb_pad = nn.utils.rnn.pad_sequence(grouped_pooled_outs, padding_value=-91, batch_first=True) # pads each sublist in chunks_emb to the largest sublist with value -91
batch_emb = batch_emb_pad.transpose(0, 1) # (B,L,D) -> (L,B,D)
lstm_input = nn.utils.rnn.pack_padded_sequence(batch_emb, seq_lengths, batch_first=False, enforce_sorted=False) # seq_lengths.cpu().numpy()
packed_output, (h_t, h_c) = self.lstm(lstm_input, ) # (h_t, h_c))
# output, _ = nn.utils.rnn.pad_packed_sequence(packed_output, padding_value=-91)
h_t = h_t.view(-1, self.hidden_size) # (-1, 100)
return self.out(h_t) # logits
The issue that I am having is that I am not entirely convinced of what data is being passed to the final classification layer. I believe what is being done is that only the final LSTM cell in the last layer is being used for classification. That is there are hidden_size features that are passed to the feedforward layer.
I have depicted what I believe is going on in this figure here:
Is this understanding correct? Am I missing anything?
Thanks.
Your code is a basic LSTM for classification, working with a single rnn layer.
In your picture you have multiple LSTM layers, while, in reality, there is only one, H_n^0 in the picture.
Your input to LSTM is of shape (B, L, D) as correctly pointed out in the comment.
packed_output and h_c is not used at all, hence you can change this line to: _, (h_t, _) = self.lstm(lstm_input) in order no to clutter the picture further
h_t is output of last step for each batch element, in general (B, D * L, hidden_size). As this neural network is not bidirectional D=1, as you have a single layer L=1 as well, hence the output is of shape (B, 1, hidden_size).
This output is reshaped into nn.Linear compatible (this line: h_t = h_t.view(-1, self.hidden_size)) and will give you output of shape (B, hidden_size)
This input is fed to a single nn.Linear layer.
In general, the output of the last time step from RNN is used for each element in the batch, in your picture H_n^0 and simply fed to the classifier.
By the way, having self.out = nn.Linear(hidden_size, 2) in classification is probably counter-productive; most likely your are performing binary classification and self.out = nn.Linear(hidden_size, 1) with torch.nn.BCEWithLogitsLoss might be used. Single logit contains information whether the label should be 0 or 1; everything smaller than 0 is more likely to be 0 according to nn, everything above 0 is considered as a 1 label.
So i am new to deep learning and started learning PyTorch. I created a classifier model with following structure.
class model(nn.Module):
def __init__(self):
super(model, self).__init__()
resnet = models.resnet34(pretrained=True)
layers = list(resnet.children())[:8]
self.features1 = nn.Sequential(*layers[:6])
self.features2 = nn.Sequential(*layers[6:])
self.classifier = nn.Sequential(nn.BatchNorm1d(512), nn.Linear(512, 3))
def forward(self, x):
x = self.features1(x)
x = self.features2(x)
x = F.relu(x)
x = nn.AdaptiveAvgPool2d((1,1))(x)
x = x.view(x.shape[0], -1)
return self.classifier(x)
So basically I wanted to classify among three things {0,1,2}. While evaluating, I passed the image it returned a Tensor with three values like below
(tensor([[-0.1526, 1.3511, -1.0384]], device='cuda:0', grad_fn=<AddmmBackward>)
So my question is what are these three numbers? Are they probability ?
P.S. Please pardon me If I asked something too silly.
The final layer nn.Linear (fully connected layer) of self.classifier of your model produces values, that we can call a scores, for example, it may be: [10.3, -3.5, -12.0], the same you can see in your example as well: [-0.1526, 1.3511, -1.0384] which are not normalized and cannot be interpreted as probabilities.
As you can see it's just a kind of "raw unscaled" network output, in other words these values are not normalized, and it's hard to use them or interpret the results, that's why the common practice is converting them to normalized probability distribution by using softmax after the final layer, as #skinny_func has already described. After that you will get the probabilities in the range of 0 and 1, which is more intuitive representation.
So after training what you would want to do is to apply softmax to the output tensor to extract the probability of each class, then you choose the maximal value (highest probability).
in your case:
prob = torch.nn.functional.softmax(model(x), dim=1)
_, pred_class = torch.max(prob, dim=1)
I have to train a network on unlabelled data of binary type (True/False), which sounds like unsupervised learning. This is what the normalised data look like:
array([[-0.05744527, -1.03575495, -0.1940105 , -1.15348956, -0.62664491,
-0.98484037],
[-0.05497629, -0.50935675, -0.19396862, -0.68990988, -0.10551919,
-0.72375012],
[-0.03275552, 0.31480204, -0.1834951 , 0.23724946, 0.15504367,
0.29810553],
...,
[-0.05744527, -0.68482282, -0.1940105 , -0.87534175, -0.23580062,
-0.98484037],
[-0.05744527, -1.50366446, -0.1940105 , -1.52435329, -1.14777063,
-0.98484037],
[-0.05744527, -1.26970971, -0.1940105 , -1.33892142, -0.88720777,
-0.98484037]])
However, I do have a constraint on the total number of True labels in my data. This doesn't mean I can build a classical custom loss function in Keras taking (y_true, y_pred) arguments as required: my external constraint is just on the predicted total of True and False, not on the individual labels.
My question is whether there is a somewhat "standard" approach to this kind of problems, and how that is implementable in Keras.
POSSIBLE SOLUTION
Should I assign y_true randomly as 0/1, have a network return y_pred as 1/0 with a sigmoid activation function, and then define my loss function as
sum_y_true = 500 # arbitrary constant known a priori
def loss_function(y_true, y_pred):
loss = np.abs(y_pred.sum() - sum_y_true)
return loss
In the end, I went with the following solution, which worked.
1) Define batches in your dataframe df with a batch_id column, so that in each batch Y_train is your identical "batch ground truth" (in my case, the total number of True labels in the batch). You can then pass these instances together to the network. This can be done with a generator:
def grouper(g,x,y):
while True:
for gr in g.unique():
# this assigns indices to the entire set of values in g,
# then subsects to all the rows in which g == gr
indices = g == gr
yield (x[indices],y[indices])
# train set
train_generator = grouper(df.loc[df['set'] == 'train','batch_id'], X_train, Y_train)
# validation set
val_generator = grouper(df.loc[df['set'] == 'val','batch_id'], X_val, Y_val)
2) define a custom loss function, to track how close the total number of instances predicted as true matches the ground truth:
def custom_delta(y_true, y_pred):
loss = K.abs(K.mean(y_true) - K.sum(y_pred))
return loss
def custom_wrapper():
def custom_loss_function(y_true, y_pred):
return custom_delta(y_true, y_pred)
return custom_loss_function
Note that here
a) Each y_true label is already the sum of the ground truth in our batch (cause we don't have individual values). That's why y_true is not summed over;
b) K.mean is actually a bit of an overkill to extract a single scalar from this uniform tensor, in which all y_true values in each batch are identical - K.min or K.max would also work, but I haven't tested whether their performance is faster.
3) Use fit_generator instead of fit:
fmodel = Sequential()
# ...your layers...
# Create the loss function object using the wrapper function above
loss_ = custom_wrapper()
fmodel.compile(loss=loss_, optimizer='adam')
history1 = fmodel.fit_generator(train_generator, steps_per_epoch=total_batches,
validation_data=val_generator,
validation_steps=df.loc[encs.df['set'] == 'val','batch_id'].nunique(),
epochs=20, verbose = 2)
This way the problem is basically addressed as one of supervised learning, although without individual labels, which means that notions like true/false positive are meaningless here.
This approach not only managed to give me a y_pred that closely matches the totals I know per batch. It actually finds two groups (True/False) that occupy the expected different portions of parameter space.
in keras, I want to customize my loss function which not only takes (y_true, y_pred) as input but also need to use the output from the internal layer of the network as the label for an output layer.This picture shows the Network Layout
Here, the internal output is xn, which is a 1D feature vector. in the upper right corner, the output is xn', which is the prediction of xn. In other words, xn is the label for xn'.
While [Ax, Ay] is traditionally known as y_true, and [Ax',Ay'] is y_pred.
I want to combine these two loss components into one and train the network jointly.
Any ideas or thoughts are much appreciated!
I have figured out a way out, in case anyone is searching for the same, I posted here (based on the network given in this post):
The idea is to define the customized loss function and use it as the output of the network. (Notation: A is the true label of variable A, and A' is the predicted value of variable A)
def customized_loss(args):
#A is from the training data
#S is the internal state
A, A', S, S' = args
#customize your own loss components
loss1 = K.mean(K.square(A - A'), axis=-1)
loss2 = K.mean(K.square(S - S'), axis=-1)
#adjust the weight between loss components
return 0.5 * loss1 + 0.5 * loss2
def model():
#define other inputs
A = Input(...) # define input A
#construct your model
cnn_model = Sequential()
...
# get true internal state
S = cnn_model(prev_layer_output0)
# get predicted internal state output
S' = Dense(...)(prev_layer_output1)
# get predicted A output
A' = Dense(...)(prev_layer_output2)
# customized loss function
loss_out = Lambda(customized_loss, output_shape=(1,), name='joint_loss')([A, A', S, S'])
model = Model(input=[...], output=[loss_out])
return model
def train():
m = model()
opt = 'adam'
model.compile(loss={'joint_loss': lambda y_true, y_pred:y_pred}, optimizer = opt)
# train the model
....
First of all you should be using the Functional API. Then you should define the network output as the output plus the result from the internal layer, merge them into a single output (by concatenating), and then make a custom loss function that then splits the merged output into two parts and does the loss computations on its own.
Something like:
def customLoss(y_true, y_pred):
#loss here
internalLayer = Convolution2D()(inputs) #or other layers
internalModel = Model(input=inputs, output=internalLayer)
tmpOut = Dense(...)(internalModel)
mergedOut = merge([tmpOut, mergedOut], mode = "concat", axis = -1)
fullModel = Model(input=inputs, output=mergedOut)
fullModel.compile(loss = customLoss, optimizer = "whatever")
I have my reservations regarding this implementation. The loss computed at the merged layer is propagated back to both the merged branches. Generally you would like to propagate it through just one layer.