Using droplast=True in DataLoaders

[xamm]

Currently, an error occurs when the epoch_size argument is not an integer multiple of batchsize.

attention-learn-to-route/options.py

Line 86 in c66da2c

assert opts.epoch_size % opts.batch_size == 0, "Epoch size must be integer multiple of batch size!"

I assume this is to ensure that there is no batch with less than the batchsize amount of data samples.

Is there any reason not to use the droplast parameter in DataLoaders to throw away such batches instead of generating an error?

[wouterkool]

Hi!

I don't think there would be a big problem with your suggestion, except that a very small batch could theoretically yield a noisy gradient. Please try and let me know if you run into problems.

[xamm]

It seems, that simply adding drop_last=True does not work as this assert

attention-learn-to-route/reinforce_baselines.py

Line 237 in c66da2c

assert (len(self.dataset) == len(self.baseline))

fails after the change.