Raghuram Shankar

icon indicating a website link https://raghuramshankar.github.io

icon location Senior Data Scientist at @elysia-bi

Results 1 comments of Raghuram Shankar

[Feature Request] gpytorch on mps

I am interested in contributing to this! Instead of using `train_x = train_x.cuda()`, using the following seems to work for me: `gpu = torch.device("mps:0")` `train_x = train_x.to(gpu, dtype=torch.float32)` However, when...