Creating stories

Hi,
Is it OK to create small sized stories probably with only a single intent and response for all intents in my domain, and then train the model with a large augmentation factor? it should work when we think logically, right? combining these small stories to get almost every stories that one can generate manually. Just curious

Thanks,

Hi @faheemv,

In theory it would work sure, but having a large augmentation factor is going to make training extremely slow. I would favour using a non-ML policy for this kind of use case like the AugmentedMemoisationPolicy or MappingPolicy