Contact
We welcome correspondence from those who operate models at scale, from those who study inference, and from those who wish to follow the work.
It is helpful to state
- The model and its size
- The hardware on which it runs
- The speed obtained at present, and the speed required