I've been hacking on a notebook using @observablehq and @d3js_org to visualize a @TensorFlow.js deep-neural network during training. This demo trains a model to call a virtual strike zone based on MLBAM PITCHf/x data: beta.observablehq.com/@nkree… ** Not supported on mobile devices

To take a step back, I really miss the days where we've been able to take our time to design software that's going to last because we knew what's going to last. It's not the case in this domain and we are all sloppy. Do I love it? No. Is it ok? Sure, we are going to rewrite.

Int4 and Int6 is a great way to reach new levels of efficiency and latency for big LLMs - particularly on CPU. MAX makes is super easy to adopt and move GenAI models to these formats. I'm very excited about this release - and yes, MAX for NVIDIA GPUs is coming later in summer!

Beyond a huge step for open source, @Modular released MAX 24.2: a tremendous step forward in Mojo🔥 (implicit trait conformance, python kwargs support, nightly builds!), even better PyTorch dynamic shapes + performance, and more. Check it out, the team is on fire! 🔥❤️‍🔥🏎

A big step forward for Mojo - it's not just open source, but open development and contribution model with a progressive Apache2 LLVM license. 🏆 We ❤️‍🔥 the Mojo community and are thrilled to continuously open the language and its development as little Mojo is growing up!

Excited to be a top company on @FastCompany most innovative AI companies list 🔥 Huge thanks to our incredible community, partners, team and investors! 🙏🏼 We're working hard to push the boundaries everyday, to rebuild AI infrastructure for the world 🌍🚀 fastcompany.com/91029555/art…

🚨📢 MAX Developer Edition Preview has officially launched! It's a drop-in replacement for running and serving @PyTorch, @tensorflow and @onnxai models much faster⚡️ and leverages the power Mojo🔥 for extensibility. Check out the blog post👇⬇️ modul.ar/max-launch-blog

“Coding” was never the source of value, and people shouldn’t get overly attached to it. Problem solving is the core skill. The discipline and precision demanded by traditional programming will remain valuable transferable attributes, but they won’t be a barrier to entry. Many times over the years I have thought about a great programmer I knew that loved assembly language to the point of not wanting to move to C. I have to fight some similar feelings of my own around using existing massive codebases and inefficient languages, but I push through. I had somewhat resigned myself to the fact that I might be missing out on the “final abstraction”, where you realize that managing people is more powerful than any personal tool. I just don’t like it, and I can live with the limitations that puts on me. I suspect that I will enjoy managing AIs more, even if they wind up being better programmers than I am.

BBQ > Buffalo wings

If you're in Oslo, Norway🇳🇴 on Feb 15th, come attend Modular engineer Fabian Tschopp's talk on accelerating Generative AI deployments using MAX and Mojo🔥. The talk is organized by SINTEF, register today: modul.ar/3tXhtRQ

ModCon is coming 😱 It’s our premier AI developer conference and will feature product announcements, lightning talks, workshops, and expert discussions. December 4th, 2023 🗓️ at the Palace Hotel in San Francisco 🌉 Early bird tickets now available! 🔥modular.com/modcon/23/start

The first crack at llama2.🔥 is here 🚀 A Mojo 🔥 community member - Mojician - did a simple port from Python to Mojo, and shows its already 20% faster than Karpathys llama.c implementation 😱 How much faster can it go? 📈

The Modular AI Engine and Mojo 🔥 are designed to be used together - supercharging the usability and performance of your AI pipelines like never before🚀 modular.com/blog/whats-the-d…

Modular's AI Engine is 5x faster on compile time and 2x faster at runtime with full dynamic shapes support 😱 Read more below 👇🏼🚀🔥 modular.com/blog/modular-nat…

We are unveiling Mojo🔥 early, but we expect it to become a full superset of 🐍 Python 3 over time… and open source as it matures. Go to modular.com to check out Mojo🔥

“Mojo 🔥 … is incredibly flexible, incredibly fast, can utilize the hardware you have … and is really understandable to Python programmers like you and me. I will actually enjoy writing neural networks again” - Jeremy Howard @jeremyphoward

Mojo🔥 combines the usability of Python with the performance of C, unlocking unparalleled programmability of AI hardware and extensibility of AI models. Also, it's up to 35000x faster than Python 🤯 and … deploys 🏎

Modular’s engine is built on next generation technology 🔥 that enables hardware innovators to unlock their AI differentiation with unprecedented usability, generality, performance and hackability 🚀

Modular’s Engine has … modularity … and drops into the infrastructure that you already use while enabling insanely low latency 👏🏼🔥 Save money or scale your models to improve your products! 🦾 See how easy below👇🏼

Tuesday is going to be wild. If you care about AI, developer tools, or software engineering, there is a good chance it will 🤯 your mind! Join us 9am pacific at modular.com

Announcing the world’s fastest unified matrix multiplication 🚀 A technical underpinning of the Modular stack 🔥 See our results now 👇🏼 and learn more about how it works at our event on May 2nd! 🍿 modular.com/blog/the-worlds-…