https://rasa.com/docs/rasa/nlu/language-support/ Myself Pranta Biswas, Tech lead and developer from Bangladesh. I’m quite new to RASA. I want to train my model with Bengali unicode letters using Tensorflow_embeddings or supervised_embeddings. I’ve already done training with some bengali sentence e.g.
তুমি কেমন আছ?
But always get an intent result of ‘null0’ while interactive learning. How can I train my bot with bengali unicode?
Hey @prantabiswas how much amount of training data you have provided, if you have given less amount of training data you won’t be able to get the expected output 
Thank u for replying @JiteshGaikwad If I train with a data and tests with the same data,shouldn’t I get my expected result? Again, is RASA able to tokenize unicode characters?
Hey @prantabiswas you can read this blog it will answer your question
Thank u @JiteshGaikwad ,but you haven’t yet provided me the link. 
Hey @prantabiswas sorry for that I forgot to paste the link 