Aleksa Gordić

😱 AlphaTensor. 😱 This. Is. Huge. Reinventing matrix multiplication using Deep RL! Faster than Strassen's algorithm that has stood the test of time for 50+ years! 10-20% efficiency increase!

Imagine what that translates to considering how fundamental matrix multiplication is for deep learning, graphics, and many other fields. Energy, compute, money, time, etc.

The only sad thing with research like this one is that it'll never become as popular as systems that generate nice-looking images and videos. Otherwise, IMHO, it would have probably a bigger ripple effect than many of the things that have happened this year.

And that guys, is a big claim, trust me I know! 😅😅😅

Blog: https://lnkd.in/e7Hb-FnD

🚀 I'm happy to announce that YugoGPT-Chat is now available: https://www.yugochat.com/

I recently announced that the model outperformed Mistral & LLaMA 2 for Serbian, Bosnian, Croatian, and Montenegrin and now you can play with it in our playground app.

Couple of things to bear in mind:
* It's still a research preview and only the first iteration of our model - it will get much better over the incoming months

* It's still not optimized for multi-turn conversations -> so if it gets stuck saying the same thing, start a new conversation

* We'll be monitoring usage, collecting feedback and we'll open-source the base model weights in January next year!

I believe that AI should serve every language, and I'm building multilingual GenAI/foundational models for enterprise.

Our next step will be to raise seed funding, in order to accelerate, get a GPU cluster and build an enterprise LLM platform.

If your company wants to use YugoGPT please reach out! https://lnkd.in/dMhWRuqr

Finally, happy xmas to those that celebrate - I'll just say that next year will be fun. :)