Lightning AI的动态

Backed by $200 million in funding, 28-year-old Scott Wu and his team of competitive coders at Cognition are building an AI tool that can program entirely on its own, like an “army of junior engineers.”

?? New Hugging Face kinda episode alert! ?? Many of us are aware of and might even use the models from the Hugging Face (HF) hub. However, if you don’t know what this company is about and are just getting started in AI, listen to my conversation with Andrea Francis, a software engineer in the dataset viewer team at HF. We spoke about her career, what HF is about, the different teams, and the strong sense of "community". Here are some takeaways for you: ? HF’s main areas of focus: datasets, models, and spaces ? Terabytes worth of datasets available for free for you to get creative and build innovative AI solutions ? Making sure the datasets are well-curated and follow the safety guidelines ? Various technologies/tools (DuckDb, Kubernetes, etc.) used by the dataset viewer team ? You could get started in different ways (contribute to the open source models, documentation, etc.) Link in comments ?

Last Thursday I had the pleasure to talk about optimizing LLMs for production inference at the AI Developer Meetup hosted by Koyeb We had some more than 150 people and awesome talks by Estelle Scifo, Yann Leger, Steeve Morin and Sophia Yang, Ph.D. My core message was to highlight how different inference and training are. And that they have very different requirements. If you've just trained a cool new model here are some of the things to keep in mind when moving to production: - don't package code you're not using. For example the pytorch/pytorch docker image starts at 7.6BG, are you really using all of that? - utilize your hardware: flashattention, batching, prefill etc. - understand your use case and measure in production your throughput and latency Thanks to everyone who came by and for all the interesting conversations I had at the event ?? And most importantly thanks to Alisdair Broshar who's the mastermind organizing these events!

We’re kicking today off with TechCrunch Disrupt AI Agent Hack Day! We’re bringing together experts from the industry to share their insights and help teams push the boundaries of AI through lightning talks in the morning, workshops in the afternoon, and then we’ll be dazzled by project demos by hackers in the evening. Here's the speaker lineup: - Philippe Brulé of Restack - Raouf Chebri of Neon - Adam Chan of Weaviate - Sandro Okropiridze of STORI AI - Daniele Bernardi of Toolhouse - Chenhe Gu of Dify

I created Socrates along with one my mate, we had in our mind to change the way of traditional learning with the concept of Socratic method of teaching. We recently participated in the Google Gen AI hackathon on Devfolio where we met with the problem statement of Blume Ventures. Today it solved my doubt about the O(n^3) concept. You can try this at https://lnkd.in/gqPtsqfA

Earlier this week, we at Parity (YC S24) released a small competition on our new benchmark called SREBench, challenging people to root cause Kubernetes issues faster than our AI agent. I'm excited to share this blog post on exactly how and why we created SREBench! Check it out ??

"Automated code generation is one of the biggest use cases for generative AI. Meta gave SF’s top hackers 24 hours, thousands of dollars in prizes, and a ton of compute to see what’s possible." -Alex Reibman Sharing Alex's summary of the Codegen Hackathon in SF. See attached pdf. Or review on thread reader app here: https://lnkd.in/ggQVygwu

Join us at AI Engineer World's Fair today (Booth G1), and catch a super interesting talk from our very own Phil Nash on 'ColBERT at scale with RAGStack'. Make your way to *YBB Salon 12* for 1.12pm, and discover: ◆ What ColBERT is ◆ How it can improve simple vector searches ◆ How you can use it with the open-source RAGStack See live demos of ColBERT in action and learn how you can implement ColBERT embeddings, index them in Astra DB, and build better retrieval for your GenAI applications. ???? More info → https://dtsx.io/4eGEVFc #DataStax #RAGStack #AIEWF

Great to see everyone who came along to see Dom Winsor and I talk about Exploring Data Engineering using free tools. If you missed it (or just want to remind yourself) the slides and demos are available in my GitHub repository: https://lnkd.in/ejEFkYFH

Just posted about my first AI conference experience!