neat-python

neat-python copied to clipboard

Implementing backpropagation

7

Hi. [ Disclaimer: I am pretty new - to github, ANNs and NEAT as well. ] I was wondering if executing backpropagation with a certain scheme might help training. Is...

morsemind

Deprecated activation functions?

1

I get the following error when using activation functions `elu`, `lelu`, and `selu`: > InvalidActivationFunction: No such activation function: 'elu' (similar for the latter two). Have these been deprecated? If...

fdetsch

Coevolution or Solution with Multiple Networks

1

Hi, I want to be able to evolve multiple networks as part of project. I would prefer to do this using coevolution if that is possible within this library. If...

1234shel

PyPi release 0.92 doesn't have the activations "elu" "selu" and probably other activations, the docs however, list them as supported activations. I think the PyPi package should be updated, since...

ShoofLLC

Hi, the StringAttributes instances have their "options=" parameters overwritten because all StringAttribute objects uses the same memmory location for options as if it was a static variable. This seem to...

aDecoy

How can we use neat-python for classification?

1

Is there any ready-made loss or activation functions (like softmax and cross-entropy)for categorical labels in neat-python? Or some best practices for classification task? It seems now I could only do...

oushu1zhangxiangxuan1

neat.distributed: use socket module instead of multiprocessing.managers

3

I finally found time to complete the rewrite of `distributed.py`. This PR replaces the usage of the `multiprocessing.managers` module with the `socket` module. Here are some advantages of this change:...

bennr01

Make configuration file optional

1

I find the current config setup to be pretty inflexible. I'd like to load configuration parameters from a database and provide the configuration parameters as kwargs. Should the constructor, for...

dyangelo-grullon

How can the activation function for the output be set? In other words, I need the outputs for num_outputs > 1 with an nn.RecurrentNetwork as `oneHot` encoded. Whats the best...

macd2

I want to allow creation only for connections between neighbourhood layers In XOR example I want to not allow for creation of connections like that.  Can I do this...

Bienqq