But I feel that this is not correct as the first slot has no arrays? As product is a list slot what would be the correct way? Or does it not matters for story prediction?
Though I have examples of a short story, I get right after beginning at the second turn wrong action prediction. The debug says the correct state but then it says:
DEBUG:rasa_core.policies.memoization:There is no memorised next action
although it is in story but maybe due to the different notaion in slots it is not used?
Then it predict with keras and low confidence the wrong action…
that looks like the right story format, the product in the intent line is the entity, and the line after that contains the slot, which if it’s a list slot that should be correct.
can you post the debug log from when you entered the data?