# How to combine NLU and Core model? **URL:** \ **Category:** Rasa Open Source\ **Created:** [September 22, 2020, 9:13am UTC](https://forum.rasa.com/t/how-to-combine-nlu-and-core-model/34003 "2020-09-22T09:13:45Z")\ **Posts on this page:** 3\ **Page:** 1
**Author:** ![gagangupt16](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/gagangupt16/32/9315_2.png) [@gagangupt16](https://forum.rasa.com/u/gagangupt16)\ **Post date:** [September 22, 2020, 9:13am UTC](https://forum.rasa.com/t/how-to-combine-nlu-and-core-model/34003/1 "2020-09-22T09:13:45Z")
In order to speed up training time, is it possible to train NLU and core separately and combine them? ---
**Author:** ![Juste](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/juste/32/15_2.png) [@Juste](https://forum.rasa.com/u/Juste)\ **Post date:** [September 22, 2020, 12:41pm UTC](https://forum.rasa.com/t/how-to-combine-nlu-and-core-model/34003/2 "2020-09-22T12:41:15Z")
@gagangupt16 To run Rasa NLU and Core models you would need a unified model which is created running `rasa run` (training both NLU and Core models at the same time). What could help you reduce the training time is using the flag `--augmentation 0` which would disable data augmentation when training. ---
**Author:** ![sysang](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/sysang/32/16719_2.png) [@sysang](https://forum.rasa.com/u/sysang)\ **Post date:** [June 23, 2022, 3:47am UTC](https://forum.rasa.com/t/how-to-combine-nlu-and-core-model/34003/3 "2022-06-23T03:47:27Z")
I also have the need to force ignoring nlu training because motst of the time I definitely know I do not need to re-train nlu, but the train command decides by itself when to train or not to train nlu. Most of the time it made me disapointed. So I have to come with the idea that is training core and nlu separately then merge them later, but it seems not feasible as well because there is not command to merge core and nlu.