I grew up with a Brownie Camera as one of many in my family's collection. Today, I decided to try some new camera techniques with AI through the decades using my "AI Brownie Camera" Here are the results (alts in images)

Easy way to reverse engineer most plugins

I've been trying Microsoft's newly-released Guidance language for formally describing LLM prompt interactions. First impression is that it's going to significantly help LLMs interacting with agents and APIs. github.com/microsoft/guid…

Worked on LangChain new plan-and-execute update and chatted with Harrison Chase to tweak and improve the design of 'AutoGPT' style agent to allow better planning and better memory management, will do a demo later today on NexusGPT !

Y Combinator has a HUGE portfolio of 55+ AI startups. Here is EACH of their breakdown & what they do (W23):

Mother of all LLM benchmarks! - Use GPT-4 if you need best quality - Use claude-instant-v1 for everything else - Google PaLM2 is nowhere near OpenAI/Anthropic - OpenAI models are painfully slow compared to competition - Open-source models next source: github.com/kagisearch/pyl…

Tree of Thought significantly increases performance of LLMs over Chain of Thought for reasoning tasks. arxiv.org/abs/2305.10601

Happy to announce the public beta of superagent.sh. Our vision is to make production ready LLM powered agents accessible to developers independent of their current stack or environment. Github: github.com/homanp/superag… Docs: docs.superagent.sh

AI Software Developer - the most in-demand job of the future. Start today and prepare yourself for an AI-equipped world. Your skills will power revolutionary products and services.

I can't believe I've just fine-tuned a 33B-parameter LLM on Google Colab in a few hours.😱 Insane announcement for any of you using open-source LLMs on normal GPUs! 🤯 A new paper has been released, QLoRA, which is nothing short of game-changing for the ability to train and

AI coding doesn't have to be overwhelming. Break it down into small steps. What's the smallest usable program that can accomplish something? Build that first. Then, build on top of it until you have something you are proud of.

AI has the potential to revolutionize education. It can create personalized learning experiences tailored to individual students' needs, providing them with the best possible education.

AI-powered education is transforming how we learn. It's no longer one-size-fits-all. By leveraging personalized data and adaptive technology, it can tailor the experience to fit each individual's unique needs.

LlamaIndex has advanced capabilities for data management/retrieval/querying. We’ve started a “Discover LlamaIndex” YouTube tutorial series to help you make use of all this tooling! 🎬📗 First up - Simon Suo shows you how to ask q’s over multiple docs 👨‍🏫 youtube.com/watch?v=GT_Lsj…

Every year, AI technology advances exponentially. Investing in AI now is like investing in the internet in 1995. Start looking for opportunities and get ahead of the curve.

AI developers don't need to understand all the frameworks. They need to understand the *why* behind them and the *how* to use them effectively. The rest is just research & experimentation.

If you want to benefit from the unstoppable trend of AI, invest in the technologies that will amplify your abilities rather than replace them.

AI will make it easier to learn more of what we want, when we want it. It will recommend more tailored interests and personalized mentorship. The classrooms of the future will be more student-driven.