nait 3.0.8

Machine learning library

Nait

Provides functions and classes for creating and

training neural networks.

---

• Model

• Model.train

• Model.predict

• Model.mutate

• Layer

• Layer.forward

• Layer.mutate

• Neuron

• Neuron.parse

• Neuron.mutate

• Activation

• Activation.calculate

• evaluate

• save

• load

---

Model

class

Syntax

Model(layers)

Arguments

• layers list

---

Model.train

method

Syntax

train(x=[], y=[], epochs=5000, rate=1, batch_size=10, sample_size=None, square_loss=True, visual=True, loss_function=None)

Arguments

• x list

• y list

• epochs int

• rate int / float

• batch_size int

• sample_size int

• square_loss bool

• visual bool

• loss_function function

---

Model.predict

method

Syntax

predict(inputs)

Arguments

• inputs list

---

Model.mutate

method

Syntax

mutate(rate)

Arguments

• rate int / float

---

Layer

class

Syntax

Layer(inputs=4, neurons=4, activation="linear")

Arguments

• inputs int

• neurons int

• activation "linear" / "step" / "relu" / "tanh" / "hardmax" / "sigmoid" / "softmax" / "leaky relu"

---

Layer.forward

method

Syntax

forward(inputs)

Arguments

• inputs list

---

Layer.mutate

method

Syntax

mutate(rate)

Arguments

• rate int / float

---

Neuron

class

Syntax

Neuron(weights, bias)

Arguments

• weights list

• bias int / float

---

Neuron.parse

method

Syntax

parse(inputs)

Arguments

• inputs list

---

Neuron.mutate

method

Syntax

mutate(rate)

Arguments

• rate int / float

---

Activation

class

Syntax

Activation(function)

Arguments

• function "linear" / "step" / "relu" / "tanh" / "hardmax" / "sigmoid" / "softmax" / "leaky relu"

---

Activation.calculate

method

Syntax

calculate(x)

Arguments

• x int / float

---

save

function

Syntax

save(model, path)

Arguments

• model Model

• path str

---

load

function

Syntax

load(path)

Arguments

• path str

---

Changelog - Nait 3.0.8

---

• Fixed 'save()'

---