a small bug in "experiments/movingmnist/mnist_rnn_main.py"
igloooo opened this issue · comments
IG H commented
In the function "train" line 120, when creating a moving mnist factory, the parameter ctx_num is omitted, while the parameter batch_size is still divided by the number of gpus. This will cause dimension mismatch when running the file with multiple gpus.
Xingjian Shi commented
Yeah, thanks for pointing out this bug. I haven't tried using multiple GPUs for the MovingMNIST experiments.