Right now, lda is the only project. It contains main.clj, and if you load that it has a simple example of running the program. Which just consists of making a random corpus, and then resampling the topics for each word. And yes, I do realize that using a random corpus for this defeats the whole point of LDA.