Pytorch-lightning: How to find the current epoch number inside the training loop?

[Deepayan137]

❓ Questions and Help

How to find the current epoch number inside the training loop?

I am training a Conv autoencoder on MNIST images and want to save the reconstructed images after every 10 epochs.

Code

```
class AutoEncoder(pl.LightningModule):
def __init__(self):
pass
def forward(self, x):
pass
def training_step(self, batch, batch_idx):
images = batch
out = self.forward(images)
loss = self.loss(out, images)
if epoch % 10:
save(images)
````
I am looking for something like this where the output images are saved every 10 epoch.

What have you tried?

I haven't tried anything since I am unable to find any documentation on current epoch no.

What's your environment?

I am using pytorch 1.4 and Lightning version 0.7

• OS: [e.g. iOS, Linux, Win]
• Packaging [e.g. pip, conda]
• Version [e.g. 0.5.2.1]

[github-actions[bot]]

Hi! thanks for your contribution!, great first issue!

[philip30]

self.current_epoch would do. (self is a lightning module). Though it starts with 0. You can find the definition of the module here.

https://github.com/PyTorchLightning/pytorch-lightning/blob/master/pytorch_lightning/core/lightning.py

Though I'd recommend you to use IDE e.g. pycharm so you can go to the definition of the class quickly.

[Deepayan137]

Thanks @philip30 :)

[jovenwayfarer]

@philip30 tqdm starts from epoch 1 but actually it is epoch 0?