Why LLMs get dumb (Context Windows Explained)
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
i didn't want to like this....
NetworkChuck
145.7k views
Why can’t LLMs just LEARN the context window?
bycloud
30.9k views
Dark Web Expert gave me his AI Tool
NetworkChuck
8.2k views
DEFCON is Not What I Expected...
NetworkChuck
14.6k views
you need to learn MCP RIGHT NOW!! (Model Context Protocol)
NetworkChuck
788.6k views
THIS Is Where the Internet Lives
NetworkChuck
753.6k views
I'm done with the AI hype
NetworkChuck
200.5k views
I'm worried about Network Engineers.....
NetworkChuck
323.7k views
You're going to get Hacked in 2025
NetworkChuck
405.2k views
let's have some coffee
NetworkChuck
34.9k views
Top Comments (10)
Any time I feel like a LLM is getting off topic or has made multiple wrong turns, trying a new chat absolutely helps! Great video! I like your AI content!
I feel attacked for forgetting the entire conversation 1:08
💻 Get fast, secure remote access with Twingate (it’s FREE): https://ntck.co/twingate_contextwindows No, ChatGPT doesn’t have memory loss… it has a context window. Have you ever felt like ChatGPT forgets what you just said? Gets dumber the longer you talk? Starts hallucinating and making up stuff? There’s a reason for that, and it’s one of the weirdest things about how AI works. In this video, I’ll break down what a context window is and how it limits what AI can remember. Grab some coffee, and prepare to unravel your AI’s goldfish brain. 🔥🔥Join the NetworkChuck Academy!: https://ntck.co/NCAcademy **Sponsored by Twingate TIMESTAMPS: 0:00 - Intro - What is a Context Window? 1:14 - LM Studio Demo: How Much Can AI Remember? 5:39 - Local Models vs. Cloud Models 8:13 - Why LLMs are bad at paying attention 11:47 - How to Max Out your Context Windows RESOURCES/LINKS: 🧠 LM Studio (Run AI models locally on your computer): https://lmstudio.ai/ 🔢 OpenAI Tokenizer (See how many tokens your text uses): https://platform.openai.com/tokenizer
Dude! I had no idea. I always had it summarize our conversation every so often, so that when it started to go bannanas, I could start a new chat with what we've done and our goals. It works decent.
This is why I always open a new chat for a new question.
yeah when i run my local llama3 instance it sometimes forgets the old messages, now i know why , thanks
I wish we could make a long-term memory for LLMS, that way they can store their knowledge outside of the memory and recall them later.
0:50 news flash... I still remember that dumb thing you said 👿
Yeah my adhd brain wants to talk about several things at once and then i noticed that it starts to get mixed up which is understandable that this was a good explanation. I'm very very new to ai and coding so this was a good video for me to watch thank you for sharing.
Watching this video after they added the branch, new chat feature solved a lot of this problem. Effectively being able to branch and pull earlier context as needed.
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
Any time I feel like a LLM is getting off topic or has made multiple wrong turns, trying a new chat absolutely helps! Great video! I like your AI content!
I feel attacked for forgetting the entire conversation 1:08
💻 Get fast, secure remote access with Twingate (it’s FREE): https://ntck.co/twingate_contextwindows No, ChatGPT doesn’t have memory loss… it has a context window. Have you ever felt like ChatGPT forgets what you just said? Gets dumber the longer you talk? Starts hallucinating and making up stuff? There’s a reason for that, and it’s one of the weirdest things about how AI works. In this video, I’ll break down what a context window is and how it limits what AI can remember. Grab some coffee, and prepare to unravel your AI’s goldfish brain. 🔥🔥Join the NetworkChuck Academy!: https://ntck.co/NCAcademy **Sponsored by Twingate TIMESTAMPS: 0:00 - Intro - What is a Context Window? 1:14 - LM Studio Demo: How Much Can AI Remember? 5:39 - Local Models vs. Cloud Models 8:13 - Why LLMs are bad at paying attention 11:47 - How to Max Out your Context Windows RESOURCES/LINKS: 🧠 LM Studio (Run AI models locally on your computer): https://lmstudio.ai/ 🔢 OpenAI Tokenizer (See how many tokens your text uses): https://platform.openai.com/tokenizer
Dude! I had no idea. I always had it summarize our conversation every so often, so that when it started to go bannanas, I could start a new chat with what we've done and our goals. It works decent.
This is why I always open a new chat for a new question.
yeah when i run my local llama3 instance it sometimes forgets the old messages, now i know why , thanks
I wish we could make a long-term memory for LLMS, that way they can store their knowledge outside of the memory and recall them later.
0:50 news flash... I still remember that dumb thing you said 👿
Yeah my adhd brain wants to talk about several things at once and then i noticed that it starts to get mixed up which is understandable that this was a good explanation. I'm very very new to ai and coding so this was a good video for me to watch thank you for sharing.
Watching this video after they added the branch, new chat feature solved a lot of this problem. Effectively being able to branch and pull earlier context as needed.