KiKaBeN

VAE: Variational Auto-Encoder (2013)

Understanding the Auto-Encoding Variational Bayes Paper

CLIP: Learning Transferable Visual Models From Natural Language Supervision (2021)

Bridging the Gap Between Vision and Language - A Look at OpenAI’s CLIP Model

LLaMA (2023)

Open and Efficient Foundation Language Models

SwiGLU: GLU Variants Improve Transformer (2020)

ReGLU, GEGLU, SwiGLU

GLU: Gated Linear Unit (2016)

Language Modeling with Gated Convolutional Networks

Adaptive Softmax (2016)

Efﬁcient softmax approximation for GPUs

Toolformer (2023-02)

Language Models Can Teach Themselves to Use Tools

Why Can GPT Learn In-Context?

Language Models Secretly Perform Gradient Descent as Meta-Optimizers

Hands-on Deep Learning with PyTorch

A Series Of YouTube Video Lectures

GPT-3: In-Context Few-Shot Learner (2020)

A generative language model with 175B params, 10x more than previous LMs

GPT-2: Too Dangerous To Release (2019)

The Difference between GPT-1 and GPT-2

GPT: Generative Pre-Trained Transformer (2018)

The First Version

Longformer: The Long-Document Transformer (2020)

How it processes thousands of tokens

Swin Transformer for Hierarchical Vision (2021)

A new Vision Transformer with Shifted Windows

ViT: Vision Transformer (2020)

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

DETR: Object Detection with Transformers (2020)

End-to-End Object Detection with Transformers

YOLOv3: An Incremental Improvement (2018)

The last version of YOLO with Joseph Redmon

RetinaNet: One-stage Detector + Focal Loss (2017)

Ross Girshick and Kaiming He developed RetinaNet

Mask R-CNN: Faster R-CNN + Mask Branch (2017)

How the mask branch works

FCN: Fully Convolutional Networks (2014)

Fully Convolutional Networks for Semantic Segmentation

DeepLab v3: Semantic Segmentation (2017)

Hands-on PyTorch Tutorial

FPN: Feature Pyramid Network (2016)

YOLO v2: Better, Faster, Stronger

YOLO v2 became better and faster than Faster R-CNN and SSD

SSD: Single Shot MultiBox Detector

YOLO: You Look Only Once (The 1st Version)

An object detection model that is very fast, pretty accurate, yet simple.

Faster R-CNN: Real-Time Object Detection with RPN

Faster R-CNN eliminated the last-boss bottleneck (Selective Search).

Fast R-CNN: Understanding why it’s 213 Times Faster than R-CNN and More Accurate

What were we really comparing?

R-CNN: Region-based Convolutional Neural Network

Extracting Features + SVM Classifier

Object Detection: Non-Maximum Suppression (NMS)

Understanding the Logic and Tricky Part of NMS

Object Detection: mean Average Precision (mAP)

Calculating mAP with Confidence

Object Detection: Intersection over Union (IoU)

A Quick Explanation

Object Detection vs. Image Classification

Simple Comparision

YOLOv5 Transfer Learning

In Simple Steps Without Losing Your Mind

YOLOv5

Object Detection Made Easy with PyTorch Hub

CycleGAN

Turn a Horse into a Zebra and vice versa with the Magic of Self-Supervised Learning

Pix2Pix

Powerful Image-to-image Translation with a Conditional Adversarial Networks

CGAN (Conditional GAN)

Specify What Images To Generate With 1 Simple Yet Powerful Change

Fashion GAN

Generating Colorful Fashion Images with DCGAN

DCGAN (Deep Convolutional GAN)

Generates MNIST-like Images with Dramatically Better Quality

GAN (Generative Adversarial Network)

Simple Implementation with PyTorch

DistilBERT — distilled version of BERT

A Tiny BERT with Great Performance – 97% of the Original

RoBERTa — Robustly optimized BERT approach

Better than XLNet without Architectural Changes to the Original BERT

BERT — Bidirectional Encoder Representation from Transformers

Pioneering Wonderful Large-Scale Pre-Trained Language Model Boom

Transformer’s Evaluation Details

Greedy and Beam Search Translators

Transformer Training Details

Optimizer, Scheduler, Loss Function

Transformer Data Loader

To Make Writing A Training Loop Simple

Transformer Coding Details

A Simple Implementation

Transformer’s Encoder-Decoder

Let’s Understand The Model Architecture

Transformer’s Self-Attention

Why Is Attention All You Need?

Transformer’s Positional Encoding

How Does It Know Word Positions Without Recurrence?

BLEU (Bi-Lingual Evaluation Understudy)

How do we evaluate a machine translation with reference sentences?

Beam Search for Machine Translation

How Greedy, Exhaustive, and Beam Search Algorithms Work

Word Embedding Lookup

Solving the curse of dimensionality problem

Neural Machine Translation with Attention Mechanism

Giving AI the Ability to Know Where To Look

Long Short-Term Memory

How LSTM Mitigated the Vanishing Gradients But Not the Exploding Gradients

Recurrent Neural Networks

Understanding RNN, Deeper RNN, Bidirectional RNN, and Seq2Seq

Temporal Convolutional Networks

Can CNNs handle sequential data and maintain history better than LSTM?

Matrix Decomposition Demystified

Eigen Decomposition, SVD, and Pseudo-inverse Matrix Made Easy

Gradient Descent Optimizers

Understanding SGD, Momentum, Nesterov Momentum, AdaGrad, RMSprop, AdaDelta, and ADAM – Made Easy

Vanishing Gradient Problem

One of the most critical problems in the history of deep learning development

Dartmouth Conference

AI in 1956 and Reflections After 50 Years

The famous Monty Hall Problem

It requires a non-intuitive solution — revisited in a story of three math fans

Bayes Theorem Demystified

Dr. Demystifier (Dr. D) vs. a High School Student Ken

Normal Distribution Demystified

Understanding the Maximum Entropy Principle

Calculus of Variations Demystified

How To Solve The Shortest Path Problem

Lagrange Multiplier Demystified

Why Does It Work?

Constrained Optimization Demystified

Explained with a Simple Example

Raspberry Pi 3 for the Second Time

Let’s Build a Vehicle

KL Divergence Demystified

What does measuring the similarity of two probability distributions mean?

Cross-Entropy Demystified

What is it? Is there any relation to the entropy concept?

Entropy Demystified

Is it a disorder, uncertainty, or surprise?

Having Fun with Deep Convolutional GANs

MNIST, SVHN, CelebA

Up-sampling with Transposed Convolution

Understanding Generative Adversarial Networks

Use A Simple GAN To Generate MNIST images

Autoencoder

How to Reduce Image Noises

Convolutional Layers for Artistic Style Transfer

Magically Turn Cats into Art

Do Filters Dream of Convolutional Cats?

Let’s Examine Convolutional Features

Pipelines, Mind Maps, and Convolutional Neural Networks

Design Neural Networks By Try and Error

Introduction to Udacity Self-Driving Car Simulator

How To Set Up The Simulator

Raspberry Pi 3 for the First Time

A Step-by-step Instruction

Lane Detection with Python OpenCV

Finding Lane Lines In The Road Images

Beautiful Sudoku Solver by Peter Norvig

Python code that solves any Sudoku puzzles systematically