Fast Edit Mode: a breakthrough from Anthropic that we're piloting in Zed. It allows Claude 3.5 Sonnet to echo its input far faster than generating new text. The result? Near-instantaneous refactoring. We're collaborating with their team in Zed's open source codebase.

According to my calculations, to make adequate use of the 1.5B free tokens @google is offering on the Gemini API to each developer. With a qps limit of 15, each request would have to be ~70k tokens. Is that how large average folks' requests are usually?

Met up in person with Pratik Desai at the AI Tinkerers meetup at GitHub HQ!

Oh so that's how tech companies use the "reverse a string" program....

I'm hosting a Vector Search Party in the Mission in SF on Wednesday evening. There'll be free food, drinks, fun demos and a lot of interesting conversations. RSVP at lu.ma/1gsr9hvl via Luma ! If you're not in the bay area, tell your friends who are!

After creating 1-2M tokens of synthetic data over several months, I've compiled some practical strategies to improve your results. Check out my latest article for straightforward tips on enhancing your synthetic data generation. ivanleo.com/blog/experimen…

Watch this space for more interesting visualization tools!

My buddy is hosting Vector Search Party in the Mission! This is gonna be an awesome / educational / fun event for anyone who's invested time and effort into tools provided by vector databases like Weaviate! It's tomorrow evening and will be a fun night!

Where's the "download project folder zip file" button in Anthropic Claude Artifacts? Until they have a hosted fiddle service, it should at least be easy to export the Artifacts to use in real repositories? Downloading each file is not the solution, neither is publish as link

Good turnout at the Mission Vector Search Party at SouthernPacificBrew in SF. Come join us if you're in the area! We're going to be here for a while

LinkedIn's new reels tab is cancer on steroids

ICYMI — I'm teaching a course on Search for RAG for seasoned engineers and CTOs! Starts in 2 weeks - weekends only. Early bird discounts have ended, but open to considering group and corporate discounts.

Witnessing how wrong journalists are about stuff you know about really makes you think about the news you read on other topics

Walmart used LLMs to improve 850m piece of data in their product catalogs, something that would have taken them 100x the headcount / time to do without LLMs

Check out this awesome thread by Nirant, for a peek into what it means to be an AI Engineer and practitioner! The no-nonsense approach to problem-solving is precisely what I aim for as my Northstar

Google could clean up their ranks so easily by just firing people who post this shit, and anyone with the same job profile and similar work patterns. I don't get it. Do they *want* people to believe this is how all Googlers work?

"I actually weirdly now have a lot more time on my hands" Oh, so that's why he's going on a public yapping tour

I'd love to have Benjamin Clavié do a Byaldi 102 - Getting Started - Why and where instead of classical OCR - Indexing, storing images - How can folks contribute Maybe a Luma, just say the word Benjamin — I'll make a Luma page and Zoom and everything else :)

Check out this Maven course, Search For RAG in the LLM era, featuring a guest lecture by Jerry Liu! You get: 🔥 Live Code Walkthroughs: Engage with code examples and implementations by coding along in real-time 🚀 Expert-led Sessions: Learn from industry veterans with over

I am thrilled to announce Trieve's $3.5M funding round led by Root Ventures! Myself and cdxker founded Trieve because we felt like building AI applications should be easier. We are looking to power every industry building with AI. New generations of discovery technology