Evaluating the likelihood of data points in a LLM's training set - View it on GitHub