A standard neural network implementation
Project description
Doki-net
Examples
Plenty of examples can be found in the aptly named examples folder.
Structure of nodes and edges
Each node and edge is a JSON object. The structure of each type of node and edge is shown below:
ANN:
node:
{
"bias" : 0.0, // Bias of the node
"type" : Type.RELU // Type refers to the activation function of a node and is set using an enum
}
edge:
"from" : 0 // The index of the node that this edge takes a value from
"to" : 1 // The index of the node that this edge sends the value to
"weight" : 0.1 // The weight of the edge
RNN (Fully-Recurrent):
node:
{
"alpha" : 0.0, // Alpha is the weight of recurrent signal
"bias" : 0.0, // Bias of the node
"type" : Type.RELU
}
CNN (In-Development):
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
doki-net-0.2.0.0.tar.gz
(5.5 kB
view hashes)
Built Distribution
doki_net-0.2.0.0-py3-none-any.whl
(38.0 kB
view hashes)
Close
Hashes for doki_net-0.2.0.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 674fa7334c141357158cc0913453c3f9c45f1538b97b2caeef0920d3c3906728 |
|
MD5 | 6d56e6770f7beec65f103bf74bdb0aa7 |
|
BLAKE2b-256 | 34d79a003e0b567a09ae7f59c029656c5b4576f6136c7e30e54804e9223a270f |