[R] Fine-tuning services report

About this article

If you have some data and want to train or run a small custom model but don't have powerful enough hardware for training, fine-tuning services can be a good solution. Once training (requiring more resources than inference) is done, the custom model can then run locally. For larger models, there is also (for some providers) the option to run inference with the custom model using their services. To get a better overview of the currently existing landscape, I did some benchmarking and experiment...