ML framework
The software library, such as PyTorch or JAX, in which AI models are written, trained and run.
Also: PyTorch, JAX
Machine learning frameworks provide the building blocks of models and handle running them on GPUs or other accelerators. PyTorch, created at Meta and since 2022 governed by the PyTorch Foundation, is the most widely used. JAX, from Google, is favoured on TPUs and by some labs. Frameworks sit on top of hardware libraries such as CUDA and ROCm, and their support for a chip largely decides whether customers can use it, which is why every accelerator maker invests heavily in PyTorch compatibility.