Benchmark speed

[sebffischer]

What we want to compare:

Deep Learning Libraries

• R: torch
• R: mlr3torch
• python: pytorch

Dataset: guess the corellation -> regression problem

Device: cpu and cuda

[cxzhang4]

TODO:

Verify that the timing strategies in Python and R are equivalent.

Verify that each framework uses the same number of cores and threads (set to 1 for simplicity)

• mlr3torch uses 1 thread by default

run the experiments on some GPU