51. Tokenization Beyond Text

Learning objectives

Talk about tokenization beyond text

You can add notes on each slide with blocks like this!
Load a deck in the browser and type “s” to see these notes.

Tokenization Beyond Text

Text

tokenization of text

image source: Murilo Gustineli

Images

Vectorization

tokenization of images

images are partitioned into patches
each patch is flattened into a vector
image source: Shusen Wang

Positional Encoding

positional encoding of patches

image source: Shusen Wang

Audio

Abstraction

audio abstraction

image source: Valerio Velardo

Music

beat
timbre
pitch
harmony
…
source: Valerio Velardo

Fourier

signal domain

image source: Valerio Velardo

Trends

digital signal processing $\rightarrow$ rule-based systems
traditional ML $\rightarrow$ feature engineering
deep learning $\rightarrow$ automatic feature engineering
source: Valerio Velardo

DSLC.io/genai | DSLC.io

51. Tokenization Beyond Text

Slides
Tools
Close

51. Tokenization Beyond Text
Learning objectives
Tokenization Beyond Text
Text
Images
Vectorization
Positional Encoding
Audio
Abstraction
Music
Fourier
Trends

f Fullscreen
s Speaker View
o Slide Overview
e PDF Export Mode
r Scroll View Mode
? Keyboard Help