I am training a model on 3 GPUs by splitting the layers into the three but after training them on the data after 2 3 epochs accuracy and loss both explode and don’t change
accuracy stays the same at 0.5
and loss oscillates arround 25
I am training a model on 3 GPUs by splitting the layers into the three but after training them on the data after 2 3 epochs accuracy and loss both explode and don’t change
accuracy stays the same at 0.5
and loss oscillates arround 25