utf8_decode

convert back the u8 bytes to utf8

train_test_split

a random split into training and test sets

tokenizers

A proc-block takes a passage and a question as input and gives us BERT Encoding in form of input_ids, input_masks, segment_ids as output

text_extractor

It parses the text from a start logit to an end logit

tensor_input

Read an input from the environment.

support_vector_regression

a supervised learning models with associated learning algorithms that analyze data for regression analysis

support_vector_classifier

a binary classifier that uses an optimal hyperplane to separate the points in the input variable space by their class.

softmax

Find the normalised exponential function (softmax).

segment_output

This proc-block is useful in image segmentation.

prediction_errors

a proc-block to find Mean Absolute Error and Mean Squared Error. used for assessing evaluation and performace of the model.

password_strength

It takes the utf8 strings as input and return 0, 1, 2 based on the utf8 string length

parse

Parse a string tensor into a tensor of numeric values.

object_filter

A proc-block which filter the object detected by an Object Detection model to: 1. remove duplicate detection for a single object 2. remove the objects with low confidence

normalize

normalizes the input to the range [0, 1]

noise-filtering

reduces noise and applies a gain control algorithm within each frequency bin.

most_confident_indices

Given some confidence values, create a tensor containing the indices of the top N highest confidences.

modulo

Apply the modulus operator to each element in a tensor.

metric

for assessing model evaluation and performace of model

logistic_regression

a statistical model that models the probability of one event taking place by having the log-odds for the event be a linear combination of one or more independent variables.

linear_regression

a linear approach for modelling the relationship between a scalar response and one or more explanatory variables.

label

Using a wordlist, retrieve the label that corresponds to each element in a tensor.

image_input

Read an image from the environment.

image-normalization

A normalization routine takes the image matrix as input and fits their values to the range [0, 1] as f32's.

fft

converts a signal from its original domain (often time or space) to a representation in the frequency domain.

f1-score

a proc-block used to calculate f1-score

elastic_net

an extension of linear regression that adds L1 and L2 regularization penalties to the loss function during training

binary_classification

A proc-block takes a probability (0.0 to 1.0) score as input and divides the output into two classes ( 0 or 1) based on a threshold

audio_float_conversion

Converted values from i16 data type to a floating-point value.

argmax

Find the index of the largest element.

accuracy

calculates accuracy of predicted labels when compared to true labels