When I bought M2 Ultra nearly 1 year ago, I tried Pytorch and I was so depressed by result that I was planning to return it, then all of sudden, out of nowhere: Apple MLX! The Apple Silicon AI unlocker! 🔝🔝🔝

<🧌> Some notes I took during OpenAI dev day: realtime api (speech-to-speech): pip install moshi_mlx vision in the fine-tuning api: pip install mlx-vlm prompt caching (50% discounts and faster processing for recently-seen input tokens): pip install mlx-lm model

Ok, I'm 100% convinced, time is the real life currency and we need more and more to be spent on what we like and love most! 💪

ChatMLX 1.1.1 Release !! ⬆️ Bump Jinja Swift 1.0.5、MLX Swift 0.18.0 - Perf improvements!!! - Support Llama 3.2 github.com/maiqingqiang/C…

Awni Hannun ChatMLX 1.1.1 has been upgraded to MLX Swift 0.18.0! 0.18.0 is too fast~ 🚀🚀🚀 x.com/johnmai_it/sta…

Another key player leaving OpenAI (BTW before releasing its creature Sora to the public). I'm wondering if this is for the money or for th culture within OpenAI. Something is clearly not working somewhere.

"Why do you do it? Why do you keep working?" Here’s my answer and remember: Never give up, my friends. Life is a single, precious gift that deserves to be lived to the fullest! 🚀 medium.com/@ivanfioravant…

Sora from OpenAI reminds me of Hololens from Microsoft: lot of hype, no real release.

The Llama 3.2 1B and 3B models are my favorite LLMs -- small but very capable. If you want to understand how the architectures look like under the hood, I implemented them from scratch (one of the best ways to learn): github.com/rasbt/LLMs-fro…

Humanity will win in the long-term. We have always found a way.

I use c# less and less during prototyping, main reasons: - too much code and classes even for trivial tasks - AI models are weaker on c# 🤷‍♂️