Check out weights and biases. I believe it can help you manage multiple experiments. As for speed you may be able to test them concurrently once you have them all set up separately. And I think someone already mentioned you can use a smaller dataset to make the process faster