I have a few very long training examples in ‘nlu.md’. When I train my NLU pipeline, it exhausts all the system resources(I guess size of model depends on longest sequence of nlu.md) and crashes my system. How to fix the input length in the NLU without worrying about the training data examples?
hi @sainimohit23 - we currently don’t have this as a parmeter to configure, but that would make a lot of sense!
Would you be up for creating a PR to add this? Would be a super nice contribution
@amn41 ok I will go through the code and try to implement the functionality if I could.
1 Like