I am blown away by RAGAS With 10 lines of code, I created a question + answer dataset of Airbnb's latest annual report (10-K). The dataset has 3 parts: • questions • contexts • ground truth answers Next step: Evaluate how well various LLMs perform RAG on financial

I've seen many reports of people struggling to finetune Gemma. We are using Alpaca-style instruction formats over conversational style and we are now getting superior performance out of Gemma-2b compared with Mistral-7b for our NER task (see screenshot to see our task). The

100% sure I am the only AI Engineer that knows this trick. The json schema examples in your Pydantic model is visible to the model. It is in fact the right way to include examples in your prompt, without touching a string. repost this if you think someone else needs to know.

"Value your time. It is all you have. It’s more important than your money. It’s more important than your friends. It is more important than anything. Your time is all you have. Do not waste your time." Naval

⚔️RAG Arena There's so many RAG strategies out there - how do you decide which one is best? Run them in battle mode: the same question with 2 different strategies. Compare and vote for winner Awesome project by Mendable! Inspired by LMSYS Org RAGArena.com

* Language is low bandwidth: less than 12 bytes/second. A person can read 270 words/minutes, or 4.5 words/second, which is 12 bytes/s (assuming 2 bytes per token and 0.75 words per token). A modern LLM is typically trained with 1x10^13 two-byte tokens, which is 2x10^13 bytes.

This is 💯.

Action produces information. Just keep doing things

“Action produces information. Just keep doing stuff.” — Brian Armstrong

Pre-seed is back. Seeing many more companies raise a $1M - $2M pre-seed followed by a $4M - $5M seed. More traction is needed to raise an A - and this is finally trickling through the pre-seed / seed world.

How build a simple iphone ios app with Cursor in in 84 seconds. Short form platform engines starting back up.

Pay the world whatever it takes to leave you alone to think.

A prompt that helps Claude 3.5 Sonnet beat OpenAI's o1 model in reasoning!

Every now and then I refresh the books on the kids nightstands and bookshelves. I never tell them what to read, but I strategically place some books. This is the pile going in today.

Humans are mimetic. Your peer group has an outsized impact on your mindset and aspirations. If you surround yourself with status-seekers and credentialists, you’ll aspire to get a job at McKinsey or Goldman. Surround yourself with wildly ambitious founders and you’ll aspire to

Y Combinator's 2024 request for startups:

C.S. Lewis on the value of starting where you are: “You can’t go back and change the beginning, but you can start where you are and change the ending.”

A few tailwind classes later and Apple is once again an innovative company

Finally got around to setting up my mac and iphone so that I can vibe code with Codex / Claude Code on the go > tmux - persistent sessions > tailscale - private secure network > termius - ssh client Left: iOS Right: macOS Always in sync. Pick up where you left off easily.

Expectation: the age of the IDE is over Reality: we’re going to need a bigger IDE (imo). It just looks very different because humans now move upwards and program at a higher level - the basic unit of interest is not one file but one agent. It’s still programming.