How many stories to train a custom slot type that takes a range of values?

In the documentation about Custom Slot Types, the example given is that of a restaurant booking system that can only handle bookings for up to 6 people. If the booking is between 1 and 6, then it proceeds to book the table, otherwise the bot will explain the limit. Here is the relevant parts of the documentation:

The example only created two stories for training the bot on how to handle different values of the people slot. I don’t think this is enough for the model to know that 6 is point where it should call action_explain_table_limit.

So my question is, in practice, how should we go about deciding the number of stories to write and the values to include when we want to train this type of custom slots?

Hi @hsm207! I think the best way is to just write a few stories, test it yourself, and keep adding stories until it is capable of handling the different values you want it to be able to handle. From there, you can make it even more robust by having your friends/colleagues test it with the share your bot feature in Rasa X and once again improve things based on real conversations

I has encounter the same question, I will be very thanksful if you give some suggestion about hou you solve the above question

Hi @xiaolinpeter. I really recommend following conversation-driven development to figure out the number of stories to write and the values to include when we want to train this type of custom slots

@tyd thanks for your suggestion