gpu

Python Modal: Parallelize class method .remote() calls for bulk inference with multiple kwargs

Modal jobs killed when local process terminates, wasting GPU time