New language (and more) from Chris Lattner and team.
Write everything in one language: Mojo combines the parts of Python that researchers love with the systems programming features that require the use of C, C++ and CUDA. Mojo is built on top of next-generation compiler technologies that unlock significant performance gains when you add types to your programs, enables you to define zero-cost abstractions, benefit from Rust-like memory safety, and that powers unique autotuning and compile-time metaprogramming capabilities.
Iāve just learned about this great idea. I mean, it was so sad that S4TF didnāt work out. I really liked using a performant, type-safe language like Swift for ML/AI development. But it seems like now there is a second chance! I wonder if we see fastai as one of the first big projects ported to mojo
Reading the docs, I like how nicely it exposes ālow-levelā concepts like copying/cloning semantics, borrow checker (!) and other MLIR features. Big news, indeed.
Iām trying to figure out how Mojo will change things for practitioners, which is why Iām trying to follow Jeremyās approach in part 2, but, with Mojo in this nbdev documentation website: slomojo/
At the moment I canāt really say that I know what Iām doing but I hope to put it on mojodojo.dev at some point.
I also wonder whether the future of the fastai course will be in Mojo. Wouldnāt that be cool? And if so maybe that would mean support for fastai on macs further down the line.
Iāve been wanting to learn Mojo and finally have some time. I looked at their website and saw their max-engine, which seems to have the most important features for AI, isnāt free to commercial use.
Iām wondering, what does the AI community think about Mojo these days? Is it still considered a good choice for AI development?