Sharing quick notes/links for some great AI videos I've watched over the past few days

Just watched this video from Jerry @LlamaIndex earlier, great overview for the features LlamaIndex gives devs to build(and monitor) production-ready RAG apps

Sharing quick notes/links for some great AI videos I've watched over the past few days

Just watched this video from Jerry @LlamaIndex earlier, great overview for the features LlamaIndex gives devs to build(and monitor) production-ready RAG apps

https://www.youtube.com/watch?v=TRjq7t2Ms5I

This video is one of the talks that I saw live at OpenAI DevDay a few weeks ago, which goes over maximizing LLM performance

It talks about optimizing prompts, best practices for RAG, model evals, fine-tuning and more

bringing mini apps to 1B users @base              dylansteck.com

There's one particular story from this talk that stood out

A customer had a setup where a query would decide one of two knowledge bases to pull on for RAG -- turned out that "simpler" methods of prompt engineering, tool use, and query expansion yielded in higher accuracy than more complex processes like HyDE retrieval

I've long admired Jason Yuan's work and am impressed again with what new.computer has been building

This video goes over different examples for how context-aware LLMs can form interfaces that feel much more dynamic and intelligent

Each demo is much different than the other

I've long admired Jason Yuan's work and am impressed again with what new.computer has been building

This video goes over different examples for how context-aware LLMs can form interfaces that feel much more dynamic and intelligent

Each demo is much different than the other

https://www.youtube.com/watch?v=ieWT6X2Yh_g

Intro to Large Language Models by Karpathy

I'm sure this video has already been shared a ton, but I can't overstate how well he explains the processes behind LLMs.

Intro to Large Language Models by Karpathy

I'm sure this video has already been shared a ton, but I can't overstate how well he explains the processes behind LLMs.

https://www.youtube.com/watch?v=zjkBMFhNj_g

This video put many things into perspective for me, especially:
- the training process
- what makes up the parameters 
- how you go from base model to a chat model

The last two videos are all about Ilya -- I watched these the day Sam got fired

Not only is this video very cinematic(views of SF + dramatic music), it also lets Ilya really speak his mind on how AI can change our futures

You can tell how much he thinks about these things

The last two videos are all about Ilya -- I watched these the day Sam got fired

Not only is this video very cinematic(views of SF + dramatic music), it also lets Ilya really speak his mind on how AI can change our futures

You can tell how much he thinks about these things

https://www.youtube.com/watch?v=9iqn1HhFJ6c

And last of all, probably the most I've learned from an AI video in a while has to be this video where Ilya explains GPT-2 and goes over transformers and attention(this was 2019 so only two years after Attention Is All You Need)

And last of all, probably the most I've learned from an AI video in a while has to be this video where Ilya explains GPT-2 and goes over transformers and attention(this was 2019 so only two years after Attention Is All You Need)

https://www.youtube.com/watch?v=T0I88NhR_9M

This video has to be the clearest yet most in depth explanation I've seen of transformers and attention -- the two key breakthroughs that not only power ChatGPT(and similar models) but changed the ways neural networks were architected

This explanation made me want to learn even more about transformers

If you have time to even watch a few minutes, I *highly* recommend watching the "Attention" section(starts around 11:00)

One of the most important innovations in neural networks that powers the large language models we're all using today

This video is one of the talks that I saw live at OpenAI DevDay a few weeks ago, which goes over maximizing LLM performance

It talks about optimizing prompts, best practices for RAG, model evals, fine-tuning and more

https://www.youtube.com/watch?v=ahnGLM-RC1Y

🐵 ← 🙈🙉🙊 → /phil →  phil.cockfield.net, db.team →  
ƒ(🌳) →  🧫 /sys  ←  @cell <T> 🧠🐦‍⬛, 100% human cartoon

Yeah that video is awesome -- and LlamaIndex/LangChain are both great as well, very helpful for managing your E2E LLM processes and having adapters for multiple models(open and closed source)

Day zero to say the least!

LlamaIndex looks super interesting.  I was really buoyed by Andrej Karpathy's recent summary of LLMs ("A busy person's guide") by the view on the OSS models, the analogs to prior close-vs-open system epochs (in PC history), and how we are seeing the same play out now.

Day freak'in ZERO, eh?!

👁️👁️ ... 🙏 🙇🏻‍♀️