nn-utils
Table of contents
“nn-utils” holds the helper functions for the main “nn” app.
Global variables
global_weight_high
Minecraft block representing positive weight (i.e. +1), default: minecraft:purple_stained_glass
global_weight_low
Minecraft block representing negative weight (i.e. -1), default: minecraft:lime_stained_glass
global_activations_high
Minecraft block representing positive activation (i.e. +1), default: minecraft:white_concrete
global_activations_low
Minecraft block representing negative activation (i.e. -1), default: minecraft:black_concrete
Functions
set_block_weight_high
Change Minecraft block (global variable) corresponding to positive weight. Takes in block name b as argument.
set_block_weight_low
Change Minecraft block (global variable) corresponding to negative weight. Takes in block name b as argument.
set_block_activation_high
Change Minecraft block (global variable) corresponding to positive activation. Takes in block name b as argument.
set_block_activation_low
Change Minecraft block (global variable) corresponding to negative activation. Takes in block name b as argument.
is_high
Takes block_pos as argument and returns if block at block_pos is either high weight or high activation or not. Note that block_pos is a list of 3 coordinates \((x, y, z)\).
is_low
Takes block_pos as argument and returns if block at block_pos is either low weight or low activation or not. Note that block_pos is a list of 3 coordinates \((x, y, z)\).
list_abs
Takes a list_input argument. Returns \(abs(x)\) for $x$ in list_input as a list.
list_str
Takes an input list (list_input) and returns the string representation of the list. For example,
list_input = l(1, 2, 3, 4);
list_str(list_input) => '[ 1 2 3 4 ]'
sign
Take an integer input x and returns sign of x. 0 is considered positive by this function. Note that this is different from signum function \(sgn(x)\) which returns 0 if x = 0.
list_sign
Takes a list as input and returns the list corresponding to sign of individual element. For example,
list_input = l(-1, 0, 1, 2, -3);
list_sign(list_input) => l(-1, 1, 1, 1, -1)
concatenate_lists
Takes 2 list inputs (list1 and list2) and returns the list formed by concatenating list2 at the end of list1. For example,
list1 = l(1, 2, 3);
list2 = l(4, 5, 6);
concatenate_lists(list1, list2) => l(1, 2, 3, 4, 5, 6)
xnor
Takes to integer inputs bit1 and bit2 and returns exclusive NOR of bit1 and bit2. For example,
bit1 = 0; bit2 = 1;
xnor(bit1, bit2) => 0
multiply_vectors
Takes in 2 vectors (scarpet lists) vec_a and vec_b containing bit representation of signs of weights and returns the list output consisting of bit representation of multiplication of every element. In short this multiplication is element wise XNOR of 2 vectors.
reduce_vector
Takes in vector vec as input and returns the sign of sum of all the numbers in vec based on bitcounting. For example,
vec = l(1, 1, 0, 1, 0, 0, 0);
reduce_vector(vec) => 0
vec = l(1, 1, 1, 0, 0, 0);
reduce_vector(vec) => 1