net

package
v0.0.0-...-70b50b3 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Feb 29, 2024 License: MIT Imports: 8 Imported by: 0

Documentation

Index

Constants

This section is empty.

Variables

View Source 
var Layer_Activation_Functions map[string]func(*mat.VecDense, Layer) = map[string]func(x *mat.VecDense, L Layer){
	"NA":      NA,
	"SoftMax": SoftMax,
}
View Source 
var Node_Activation_Functions map[string]func(float64, *Node) float64 = map[string]func(float64, *Node) float64{
	"NodeLU":    NodeLU,
	"ReLU":      ReLU,
	"Log":       Log,
	"ExpE":      ExpE,
	"ExpSqrtE":  ExpSqrtE,
	"ExpPi":     ExpPi,
	"ExpSqrtPi": ExpSqrtPi,
	"Collatz":   Collatz,
}

▄▀█ █▀▀ ▀█▀ █ █░█ ▄▀█ ▀█▀ █ █▀█ █▄░█   █▀▀ █░█ █▄░█ █▀▀ ▀█▀ █ █▀█ █▄░█ █▀ █▀█ █▄▄ ░█░ █ ▀▄▀ █▀█ ░█░ █ █▄█ █░▀█   █▀░ █▄█ █░▀█ █▄▄ ░█░ █ █▄█ █░▀█ ▄█

we are using a dict to store the funtions. each function will be stored under a key that is a string so that each node can have a key to a actavation function all

dict to map funtions for the Node obj, this is done for dinamic calls to diffrent func. all func must be plan func and not methods with the parma(x float64, n Node) or cannot be placed inside the map

Functions

func Collatz 

func Collatz(x float64, n *Node) float64

the hailstone algorithom or Collatz conjecture https://en.wikipedia.org/wiki/Collatz_conjecture

func Compute_loss 

func Compute_loss(value float64) float64

func ExpE 

func ExpE(x float64, n *Node) float64

raw exponentiation via math.E

func ExpPi 

func ExpPi(x float64, n *Node) float64

raw exponentiation via PI

func ExpSqrtE 

func ExpSqrtE(x float64, n *Node) float64

raw exponentiation via math.SqrtE

func ExpSqrtPi 

func ExpSqrtPi(x float64, n *Node) float64

raw exponentiation via PI

func Log 

func Log(x float64, n *Node) float64

natral log: takes a float64 as var x and returns 0 if input < 0 else returns var log10(x)

func NA 

func NA(x *mat.VecDense, L Layer)

func NodeLU 

func NodeLU(x float64, n *Node) float64

func ReLU 

func ReLU(x float64, n *Node) float64

rectified linear: takes a float64 as var x and returns 0 if input < 0 else returns var x

func SoftMax 

func SoftMax(x *mat.VecDense, L Layer)

this function is intededed to be usesd for the Output of the forward pass so that the NN can be traned with the prodiction of how corect the model is

Types

type Gorebrum 

type Gorebrum struct {
	Layers         []*Layer
	Output         *mat.VecDense
	Width          int
	Depth          int
	Out_dimensions int
	In_dimensions  int
	Log            []string // for training
	// contains filtered or unexported fields
}

func Dcode_model 

func Dcode_model(name string) *Gorebrum

func New_Network 

func New_Network(
	In_dimensions int,
	Width int,
	Depth int,
	Out_dimensions int) *Gorebrum

this is the functon that makes a new network, New_Network takes 4 variabels as int as fallows

In_dimensions int, (this is the length of the vector that is pasesed into the network)

Width int, (the number of Layers includeing the frist and last Layers)

Depth int, (the lenght of the vectors in the hidden Layers)

Out_dimensions int, (the length of the Output layer)

the function will then return a *Gorebrum: 

	type Gorebrum struct {
	input          *mat.VecDense
	Layers         []*Layer
	Output         *mat.VecDense
	Width          int
	Depth          int
	Out_dimensions int
	In_dimensions  int
}

the Output layer will allwase be a softmax actavation

func (*Gorebrum) Computer_fraward_pass 

func (N *Gorebrum) Computer_fraward_pass()

the farward pass for the network

func (*Gorebrum) Display_Network 

func (N *Gorebrum) Display_Network()

func (*Gorebrum) Encode_model 

func (N *Gorebrum) Encode_model(name string)

encodes the network and wrigths the model to a gob file wich

func (Gorebrum) Get_Depth 

func (N Gorebrum) Get_Depth() int

func (Gorebrum) Get_Layers 

func (N Gorebrum) Get_Layers() []*Layer

func (Gorebrum) Get_Output 

func (N Gorebrum) Get_Output() *mat.VecDense

func (Gorebrum) Get_Width 

func (N Gorebrum) Get_Width() int

func (Gorebrum) Get_input 

func (N Gorebrum) Get_input() *mat.VecDense

func (Gorebrum) Get_layer 

func (N Gorebrum) Get_layer(index int) *Layer

func (*Gorebrum) New_Layer 

func (net *Gorebrum) New_Layer(length int, inputsLen int, Activation string, Index int)

this is a funtion to add new Nodes to the layer

func (*Gorebrum) Set_Depth 

func (N *Gorebrum) Set_Depth(depth int)

func (*Gorebrum) Set_In_dimensions 

func (N *Gorebrum) Set_In_dimensions(inDimensions int)

func (*Gorebrum) Set_Layer 

func (N *Gorebrum) Set_Layer(index int, layer *Layer)

setters for the network

func (*Gorebrum) Set_Out_dimensions 

func (N *Gorebrum) Set_Out_dimensions(outDimensions int)

func (*Gorebrum) Set_Output 

func (N *Gorebrum) Set_Output(output *mat.VecDense)

func (*Gorebrum) Set_Width 

func (N *Gorebrum) Set_Width(width int)

func (*Gorebrum) Set_inputs 

func (N *Gorebrum) Set_inputs(inputs []float64)

sets an input to the seme vec that is passed to it

func (*Gorebrum) ToXML 

func (N *Gorebrum) ToXML() string

this displays the network in all its glory and wonder

func (*Gorebrum) Update_Output 

func (N *Gorebrum) Update_Output()

this updates the Output for the network

type Layer 

type Layer struct {
	Output     *mat.VecDense
	Nodes      []*Node
	Activation string

	Index int
	Log   []string // for training
	// contains filtered or unexported fields
}

this is the structur of the Layers in the network the goal is to apstrat all the

func (Layer) Compute_Layer 

func (L Layer) Compute_Layer()

Compute_Layer

func (Layer) Display_info 

func (L Layer) Display_info()

func (*Layer) Get_Activation 

func (L *Layer) Get_Activation() string

func (*Layer) Get_Index 

func (L *Layer) Get_Index() int

func (*Layer) Get_Nodes 

func (L *Layer) Get_Nodes() []*Node

func (*Layer) Get_parent_network 

func (L *Layer) Get_parent_network() *Gorebrum

func (*Layer) New_Node 

func (L *Layer) New_Node(Weights []float64, Bias float64, Activation string, i int)

makes a new node from the layer

func (*Layer) Set_Activation 

func (L *Layer) Set_Activation(activation string)

func (*Layer) Set_Index 

func (L *Layer) Set_Index(index int)

func (*Layer) Set_Nodes 

func (L *Layer) Set_Nodes(nodes []*Node)

func (*Layer) Set_Output 

func (L *Layer) Set_Output(output *mat.VecDense)

func (*Layer) Set_parent_network 

func (L *Layer) Set_parent_network(parent_network *Gorebrum)

setters

func (Layer) ToXML 

func (L Layer) ToXML() string

type Node 

type Node struct {
	Weights *mat.VecDense

	Bias       float64
	Activation string
	Output     float64

	Index int
	Log   []string // for training
	// contains filtered or unexported fields
}

this is the struct for creating Nodes this will eventualy be a nural network

func New_Node 

func New_Node(Weights []float64, Bias float64, Activation string) *Node

makes a node form a static state

func (*Node) Compute_node 

func (n *Node) Compute_node()

method for computing the Output of a node

func (Node) Display_info 

func (n Node) Display_info()

func (Node) Get_Activation 

func (n Node) Get_Activation() string

reternds the activation of the network.

func (Node) Get_Bias 

func (n Node) Get_Bias() float64

returns the bias of the node

func (Node) Get_Weights 

func (n Node) Get_Weights() *mat.VecDense

returns the weights of the node as a *mat.VecDence

func (Node) Get_input 

func (n Node) Get_input() *mat.VecDense

this function grabs the Outputs from the layer before it. unless it is the frist layer, in that case the input for the network is used

func (Node) Get_output 

func (n Node) Get_output() float64

returns the output of the node

func (Node) Get_parent_layer 

func (n Node) Get_parent_layer() *Layer

returs the address of the layer that the node is a part of

func (*Node) Set_Activation 

func (n *Node) Set_Activation(Activation string)

func (*Node) Set_Bias 

func (n *Node) Set_Bias(Bias float64)

sets the Bias of a node to a float64

func (*Node) Set_Output 

func (n *Node) Set_Output(Output float64)

func (*Node) Set_Weights 

func (n *Node) Set_Weights(Weights *mat.VecDense)

sets the weights of the node to a *mat.VecDence

func (*Node) Set_input 

func (n *Node) Set_input(input *mat.VecDense)

func (*Node) Set_parent_layer 

func (n *Node) Set_parent_layer(Layer *Layer)

func (Node) ToXML 

func (n Node) ToXML() string

type Trainer 

type Trainer struct {
	Traing_algoithms map[string]func(*Trainer)
	// contains filtered or unexported fields
}

func New_trainer 

func New_trainer(network *Gorebrum) *Trainer

this creates a new training obj for training a model on a data set doing it this way will give is freedom to implement different training strategies

func (*Trainer) Compute_Training_Pass 

func (T *Trainer) Compute_Training_Pass()

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.