An error occurs in CPU-only environment due to hardcoded 'cuda' device

[kawabe-jiw]

Description

An error occurs when running the model in a CPU-only environment because the tensor device is hardcoded to "cuda" in sam3/model/position_encoding.py.

Location of the Issue

sam3/sam3/model/position_encoding.py

Line 55 in 5dd401d

tensors = torch.zeros((1, 1) + size, device="cuda")

Suggested Fix

We can resolve this by dynamically allocating the device based on an existing tensor or module parameters. For example:

tensors = torch.zeros((1, 1) + size, device="cuda" if torch.cuda.is_available() else "cpu")