Oyedokun Oluwatobiloba (@devaladey) profile picture

Full-stack dev learning AI in public ๐Ÿง  | HTML CSS JS React Next.js Python & more | Documenting my AI journey daily | Come learn with me ๐Ÿ‘‡

Lagos, Nigeria
Joined May 2022
Photo by Oyedokun Oluwatobiloba (@devaladey)
I went quiet for a few days Not because I stopped learning But because I was heads down studying ๐Ÿ“š Back now and we're not slowing down Activation Functions, Backpropagation, NLP, LLMs and more coming ๐Ÿ”ฅ Did you miss the AI breakdowns? ๐Ÿ‘‡ #AI #LearnInPublic
10
ChatGPT, self driving cars, face recognition They all have one thing in common Deep Learning ๐Ÿงต Let me show you how it actually works #AI #LearnInPublic
1
16
First โ€” remember Neural Networks? A Neural Network is like a brain with layers Deep Learning is just a Neural Network with MANY more layers The word "deep" literally means depth of layers More layers = more understanding ๐Ÿง 
1
3
2 weeks of learning AI in public done โœ… What I've covered so far: ๐Ÿค– What AI actually is ๐Ÿง  Neural Networks ๐Ÿ“Š Supervised vs Unsupervised โš–๏ธ Bias vs Variance ๐Ÿ“‰ Loss Functions & more And I'm just getting started Drop a ๐Ÿ”ฅ if you've been following along #AI #LearnInPublic
17
How does AI know when it's wrong? One word โ€” Loss Function This is what makes AI actually learn ๐Ÿงต #AI #LearnInPublic
1
33
Remember y = mx + b? The model makes a prediction But that prediction is rarely perfect at first Loss Function measures HOW WRONG the prediction is The bigger the mistake โ†’ the higher the loss The smaller the mistake โ†’ the lower the loss ๐ŸŽฏ
1
4
Why do AI models fail even after training? Two reasons โ€” Bias and Variance This one changed how I think about ML ๐Ÿงต #AI #LearnInPublic
1
42
Let's start with Bias Imagine an archer shooting at a target ๐ŸŽฏ Every arrow lands in the same spot But that spot is far from the bullseye The archer is consistent But consistently wrong That's High Bias
1
4
Overfitting is like a student who memorizes answers but doesn't understand the subject They ace practice tests Fail the real exam AI does the same thing The goal is always to learn patterns not memorize data ๐ŸŽฏ #AI #LearnInPublic
52
Two things that can ruin your AI model Overfitting and Underfitting Most beginners don't know the difference ๐Ÿงต #AI #LearnInPublic
1
34
Let's start with Overfitting Remember our student studying for an exam? Imagine they memorized every past question word for word But never understood the concept They ace the past questions But fail the real exam completely That's Overfitting ๐Ÿ“
1
8
The more I learn about AI the more I realize It's not magic It's just math, data and patterns Anyone can understand it ๐ŸŽฏ #AI #LearnInPublic
24
How do you know if AI actually learned something? Two words โ€” Training and Testing Data ๐Ÿงต #AI #LearnInPublic
1
35
Think about studying for an exam ๐Ÿ“ You spend weeks going through past questions You learn the patterns You understand the concepts Those past questions = Training Data That's what the model learns from
1
5
Two words every ML engineer uses daily Features and Labels Let me explain with something you already know ๐Ÿงต #AI #LearnInPublic
1
41
Imagine you're trying to predict if someone is a good footballer โšฝ You look at: ๐Ÿ“Œ Speed ๐Ÿ“Œ Goals scored ๐Ÿ“Œ Dribbling ability ๐Ÿ“Œ Fitness These are the things you observe to make a judgment In ML these are called Features
1
14
One week of learning AI in public done โœ… What I expected: robots and rocket science What it actually is: patterns, data and a lot of googling ๐Ÿ˜ญ Best decision I made was just starting #AI #LearnInPublic
32
Everyone talks about AI models But nobody explains what a model actually is ๐Ÿงต Let me break it down simply #AI #LearnInPublic
1
31
A model is just a function that takes input and gives output ๐Ÿ“ธ Input: photo of a dog ๐Ÿค” Model: processes it ๐Ÿถ Output: "that's a dog" That's it A model is just a very smart function
1
7
AI didn't start with ChatGPT ๐Ÿคฏ 1950 โ€” Can machines think? 1997 โ€” AI beats world chess champion 2017 โ€” Transformers invented 2022 โ€” ChatGPT changes everything 70 years of work behind that one chat box #AI #LearnInPublic
30
AI is only as smart as the data it learns from ๐Ÿงต Let me explain Training Data in the simplest way possible #AI #LearnInPublic
1
31
Think about how you recognize your mum's cooking You've eaten it hundreds of times Your brain knows the taste, the smell, the texture AI learns the same way The hundreds of examples = Training Data ๐Ÿฒ
1
4
There are 2 main ways AI learns from data ๐Ÿงต Most people don't know the difference Let me break it down simply #AI #LearnInPublic
1
27
First โ€” Supervised Learning You give AI data WITH the answers ๐Ÿฑ Here are 1000 cat pictures โ€” they are cats ๐Ÿถ Here are 1000 dog pictures โ€” they are dogs AI learns by studying the labeled examples Like a student studying with an answer key ๐Ÿ“
1
10
You make decisions like an AI every day and don't even know it ๐Ÿงต Let me explain Decision Trees in the simplest way possible #AI #LearnInPublic
1
37
Imagine you wake up and ask yourself: "Should I carry an umbrella today?" โ˜๏ธ Is it cloudy? โ†’ Yes โ†’ Is rain forecasted? โ†’ Yes โ†’ Take umbrella That's literally a Decision Tree ๐ŸŒณ
1
25
A Neural Network sounds scary. It's not. Here's the simplest explanation: ๐Ÿง  Your brain has neurons that send signals ๐Ÿ’ป AI copies that idea with numbers Input โ†’ process โ†’ output That's it. That's a Neural Network. Mind = blown when I learned this ๐Ÿคฏ #AI #LearnInPublic
24
AI feels overwhelming until you see the map: ๐Ÿค– AI โ†’ broad concept ๐Ÿ“Š ML โ†’ learns from data ๐Ÿง  Deep Learning โ†’ neural networks ๐Ÿ’ฌ NLP โ†’ understands language ๐Ÿ”ข Embeddings โ†’ words as numbers Posting my journey daily ๐Ÿ‘‡ #AI #LearnInPublic
36
Bullish on shrapnel and whole team
๐Ÿ”ฅ #Bybit x Shrapnel: 35,000,000 $SHRAP Up for Grabs! Follow @Bybit_official and @playSHRAPNEL, quote retweet this tweet, deposit and complete the form to win big! ๐Ÿ’ซ Join Event: i.bybit.com/ab13zFvx #BybitTrading #TheCryptoArk
41
React or Vuejs which? I think I lyk Vuejs because of the fact that the markup, styles and logic of a component are written in a single file (SFC) compared to React where you write your styles in a separate file and willโ€™ve to navigate back and forth the file system of your IDE.
3