Why do we need to call zero_grad() in PyTorch? [duplicate]

Question

This question already has an answer here:

• Why do we need to explicitly call zero_grad()? 4 answers

The method zero_grad() needs to be called during training. But the documentation is not very helpful

|  zero_grad(self)
|      Sets gradients of all model parameters to zero.

Why do we need to call this method?

Answer 1

In PyTorch, we need to set the gradients to zero before starting to do backpropragation because PyTorch accumulates the gradients on subsequent backward passes. This is convenient while training RNNs. So, the default action is to accumulate (i.e. sum) the gradients on every loss.backward() call.

Because of this, when you start your training loop, ideally you should zero out the gradients so that you do the parameter update correctly. Else the gradient would point in some other direction than the intended direction towards the minimum (or maximum, in case of maximization objectives).

Here is a simple example:

import torch
from torch.autograd import Variable
import torch.optim as optim

def linear_model(x, W, b):
    return torch.matmul(x, W) + b

data, targets = ...

W = Variable(torch.randn(4, 3), requires_grad=True)
b = Variable(torch.randn(3), requires_grad=True)

optimizer = optim.Adam([W, b])

for sample, target in zip(data, targets):
    # clear out the gradients of all Variables 
    # in this optimizer (i.e. W, b)
    optimizer.zero_grad()
    output = linear_model(sample, W, b)
    loss = (output - target) ** 2
    loss.backward()
    optimizer.step()

---

Alternatively, if you're doing a vanilla gradient descent, then:

W = Variable(torch.randn(4, 3), requires_grad=True)
b = Variable(torch.randn(3), requires_grad=True)

for sample, target in zip(data, targets):
    # clear out the gradients of Variables 
    # (i.e. W, b)
    W.grad.data.zero_()
    b.grad.data.zero_()

    output = linear_model(sample, W, b)
    loss = (output - target) ** 2
    loss.backward()

    W -= learning_rate * W.grad.data
    b -= learning_rate * b.grad.data

---

Note: The accumulation (i.e. sum) of gradients happen when .backward() is called on the loss tensor.