Make PyTorch models up to 40% faster! Thunder is a source to source compiler for PyTorch. It enables using different hardware executors at once; across one or thousands of GPUs.
So now that the test has run, do we need this?
It would be cool to leave things that are not to be looked at for merging in draft state (but preferably closing them after they have done what they need, too).
Testing whether indeed we do not have Nones in the
saved_tensors_list
across all our CIs.