DL4J/Text Processing: Using existing word2vec models

Hi there,

I would like to play around with the word embeddings from the DL4J/Textprocessing plugins in KNIME. It would be awesome if I could use existing models, such as these. Currently, there's obviously no way to supply external models. Do you guys have any plans for such a functionality? Looking at the DL4J docs., this seems to be supported. In case it's not on your backlog anyways, I could probably try to create a node for this myself.

And here's one issue I just discovered: When cancelling the "Word Vector Learner" node, the process still seems to be running (KNIME remains at full CPU usage). I need to quit KNIME completely to get back to normal CPU usage.


Hi qqilihq,

we plan to add a node which is able to read such models.

Thanks for reporting the second issue, I will have a look.



Great, looking forward!

-- Philipp