Understanding LLMs in Simple Terms

 Everyone talks about ChatGPT and revolution of GenAI. Ever wondered what powers the chatbot you use or how your favourite AI assistant seems to know just the right thing to say? It’s all thanks to LLMs, or Large Language Models. Let me break it down in plain English and make it easy to grasp.

Key things to know about LLMs:

  1. Think of them like super-smart text predictors! (imagine writing "Hello, how are..." and it knows to complete with "you"... but way more advanced!)
  2. They work like a magical book predictor (they read MILLIONS of books, articles, websites... if you read non-stop, it would take 2600+ years to read what GPT-3 learned from!)
  3. Don't expect them to be human-like!!! (they're like a chef who memorised recipes but never tasted food)

Cool Metaphors that helped me understand:

  1. The Library Detective
  • Has read EVERY book in existence
  • Can find info super fast
  • Connects different facts together (like having the world's smartest librarian at your service 24/7!)
  1. The Time-Frozen Librarian
  • Knows everything until their "freeze date"
  • Can't learn new things after conversations
  • Memory resets like in "50 First Dates" movie! (important to remember when using ChatGPT or similar tools)

Things to Remember:

  1. They work by predicting what comes next (just like autocomplete in your phone, but MUCH smarter)
  2. They use something called "transformers" (they don't read word by word, they look at EVERYTHING at once!)
  3. They're super resource-heavy (even if you could do a BILLION calculations per second, training would take 100 million years!!!)

Don't!!!

  1. Don't assume they understand like humans (they're pattern matching machines, not conscious beings)
  2. Don't trust everything they say (they can mix up facts and create "hallucinations" - wrong but convincing info)
  3. Don't expect them to know recent events (they're frozen in time at their last training date)

My Experience:

After playing around with these tools, I've realised they're AMAZING at:

  • Writing help
  • Summarising stuff
  • Answering questions
  • Translation
  • Creative tasks

But remember... they're tools, not magic! they’re not human and don’t think like us. They’re statistical geniuses—not philosophers. Use them wisely and double-check important facts!!!

That's all for now! Hope this helps you understand LLMs better. Drop your thoughts in comments!

PS: Clean your AI workspace everyday!!! (just kidding... but seriously, verify the outputs! 😉)

Also you can read my old post -  Principles and Tactics for Working with Language Models



Comments

Post a Comment

Popular posts from this blog

MDK Hospital opened in Horana

Image
Horana is getting a brand new private hospital! The M.D.K Healthcare Center & Hospital has just opened its doors on Anguruwathota Road, right next to the HNB bank branch. For years, many had to travel to Colombo for certain medical treatments, but hopes with new private hospital this will change. This new hospital, built by local businessman Mr. M.D. Piyananda Kularathna, comes with all the modern medical equipment we've been hoping for. It's wonderful to see such a major investment in Horana community's health. To makes this even better, I hope they have brought in highly skilled doctors and medical staff. No more long trips to Colombo for routine treatments! The hospital looks quite impressive too - modern and clean, just what you'd expect from a proper healthcare centre. The location is perfect as well. Being on Anguruwathota Road means it's easy to find and get to, whether you're coming from the town centre or surrounding areas. I've already notice...
Read more

Disable single quote Within Forms

This is a good method taught to me by my friend to disable any character in the keyboard Within Forms in your web page/ web application. A simple and effective method to disable specific characters (like single quotes) within form inputs on your webpage or web application. The Code <!DOCTYPE html> <html> <head> <script type="text/javascript"> function stopRKey(evt) { // Normalize the event object evt = evt || window.event; // For debugging - shows the keyCode console.log(evt.keyCode); // Block single quote (39 or 222 depending on browser) if (evt.keyCode == 39 || evt.keyCode == 222) { return false; } } // Attach the event handler to the document document.onkeypress = stopRKey; </script> </head> <body> <h3>Test Area</h3> <textarea rows="5"></text...
Read more

My Docker learning short notes

I had this note from long time in my Evernotes, I thought I should have post it here, so someone might find it useful. This is just Docker basics and commonly used commands that I've collected over time. Why Docker? Docker enables containerization - running applications in containers instead of virtual machines. Important notes: Containers are not VMs - they have different benefits Provides standardization across environments Enables continuous deployment from development to production Ensures consistency using the same container image throughout the deployment process Key Benefits: Self-contained environments Isolated from other applications Runs almost everywhere (especially in the Cloud) Small and lightweight Highly scalable Fast deployment Core Concepts Images An image is an executable package containing everything needed to run an application: ...
Read more