it’s been almost a week, so i’ll try once more 🙂 what do people use for word embeddings (word2vec etc.?) for the moment i just want to use existing models, not train new ones. i tried dl4j, but it takes 5-6 minutes just to load the Google News model, vs. 45 seconds for python gensim.

Have you already tried Mallet, they have a skip-gram model. I haven’t used Mallet’s skip-gram model, but in general I had some good experience doing LDA models in Mallet.

i have the mallet page open, but haven’t tried it yet