PyTorch Foundation Spotlight: Simon Mo

Key Takeaways

VR has [music] become a PyTorch Foundation project as of May 2025 and this actually has been a long time in the making in the sense that VR has been built on top [music] of PyTorch from day one. It is fundamentally an ecosystem where we work [music] with people who build the model like Meta, Quinn, OpenAI as well as the accelerator companies like [music] Nvidia, AMD, Intel, many of them are PyTorch sponsors that directly match making sure that all the model can be run on [music] VM efficiently on all the hardware. VM's primary focus is to drive ease of [music] use and efficiency. One of the key barrier to adopting large language models in enterprise and organization today is how hard it is and how expensive it is to use. By leveraging [music] open source systems, a lot of users can use this on day [music] one to be able to get value out of it. And additionally, VLM keep driving the frontier of efficiency forward so that you can always achieve the lowest cost possible on a given hardware to serve [music] the model. VM's goal is to become the easiest to use and most efficient inference engine that will help [music] driving the progress of AI forward.