I notice that when I run summary() on my models I get multiple connections. I also get GpuArrayExceptions: out of memory exceptions which I think might be related.
I’m running my version of dogscats-ensemble notebook, and using the following def of get_conv_model.
conv_layers, fc_layers = split_at(vgg_eqv_model, Convolution2D)
conv_model = Sequential(conv_layers)
last_conv_idx = len(conv_layers)-1
return conv_model, fc_layers, last_conv_idx
Any ideas? Thanks!