# Does the model trains from start every time i use interactive learning?

**URL:** <https://forum.rasa.com/t/does-the-model-trains-from-start-every-time-i-use-interactive-learning/3739>\
**Category:** Rasa Open Source\
**Created:** [December 20, 2018, 7:15am UTC](https://forum.rasa.com/t/does-the-model-trains-from-start-every-time-i-use-interactive-learning/3739 "2018-12-20T07:15:50Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![tbhavnani](https://avatars.discourse-cdn.com/v4/letter/t/3ec8ea/32.png) [@tbhavnani](https://forum.rasa.com/u/tbhavnani)\
**Post date:** [December 20, 2018, 7:15am UTC](https://forum.rasa.com/t/does-the-model-trains-from-start-every-time-i-use-interactive-learning/3739/1 "2018-12-20T07:15:50Z")

</div>

I have this doubt. How does the interactive learning work. every time i tell the bot to do something else from what it predicts, it retrains. But when i start the interactive training the next time, does it just take the stories from the stories.md to retrain, or it starts from the pre trained level(checkpoint).

---

<div class="post-metadata">

**Author:** ![Akshit](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/akshit/32/1431_2.png) [@Akshit](https://forum.rasa.com/u/Akshit)\
**Post date:** [December 20, 2018, 7:53am UTC](https://forum.rasa.com/t/does-the-model-trains-from-start-every-time-i-use-interactive-learning/3739/2 "2018-12-20T07:53:30Z")

</div>

@tbhavnani I had the same question! Maybe @Juste or @akelad can answer this question better than me! 🙂

---

<div class="post-metadata">

**Author:** ![akelad](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/akelad/32/69_2.png) [@akelad](https://forum.rasa.com/u/akelad)\
**Post date:** [January 8, 2019, 10:19am UTC](https://forum.rasa.com/t/does-the-model-trains-from-start-every-time-i-use-interactive-learning/3739/3 "2019-01-08T10:19:49Z")

</div>

So once you’ve completed interactive learning, the bot will ask you where you want to export the stories to. They’ll then get saved to that file. The next time you start interactive learning it will train on whatever stories you provide it with, if the stories generated from the previous interactive learning session are part of that then it’ll include them in training
