GoMLX project provides growing support for 1st-class machine learning (ML) for the Go language.

The project includes several repositories under github.com/gomlx, but also some of its dependencies or collaborating projects. Among them:

• github.com/gomlx/gomlx - GoMLX is a machine learning framework for Go, with batteries included. A rich set of ML functionality and XLA backends (CPU, CUDA, TPU, and likely other compute supported), and a pure Go backend (slower but very portable).
• github.com/gomlx/gopjrt - XLA/PJRT bindings for Go. XLA/PJRT is one of the fastest ML engines, and it is also used by TensorFlow, Jax and even PyTorch. This repository makes it available to GoMLX. It is paired with github.com/gomlx/stablehlo to generate the IR (intermediary representation) consumed by PJRT.
• github.com/gx-org/go-stablehlo: New home for the stablehlo Go library. GX and GoMLX will share backends, and it will be hosted under gx-org.
• github.com/gomlx/go-huggingface: Conveniently download files, models, tokenizers from HuggingFace Hub.
• gnn, bsplines: other supporting ML-related functionality.
• github.com/janpfeifer/gonb: a Jupyter notebooks kernel for Go. Very useful to quick demo and test machine learning model, plot graphs with the model metrics, data analysis, reports, etc.

This sponsorship program aims at supporting the maintenance and further development of these projects and new related ones. Another important aspect that donations can help is with access (buying or renting on cloud) to new hardware, to expand portability.

Most importantly, this sponsorship aims at fueling our coffee ☕ addiction!

Don't hesitate in reaching out to inquire and/or suggest future plans. Development can be tracked on the CHANGELOG files.