Yes. There are services that provide free environment to run jupyter notebooks for research purposes (with GPU included, which is crucial for neural networks) - such as Google Colaboratory and Kaggle Kernels. Although they limit how long your computation may run (12 and 6 hours accordingly), which adds some difficulties to the process, although I think it is possible to bypass these restrictions.
Take a look at using AWS You can spin up an instance with as much processing power as you need and by using their pre-built images it will already be preconfigured with a lot of the packages etc you might need for any kind of ML.
I see you wanted to use DeepFaceLab which I guess requires some kind of GUI so unsure if this is suitable for your requirements but check it out, it seems to be the best way to perform high-processing machine learning without the fancy computer