# Memory Issue

**URL:** <https://forum.rasa.com/t/memory-issue/22143>\
**Category:** Rasa Open Source\
**Created:** [December 6, 2019, 6:42am UTC](https://forum.rasa.com/t/memory-issue/22143 "2019-12-06T06:42:17Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![shashank34](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/shashank34/32/5110_2.png) [@shashank34](https://forum.rasa.com/u/shashank34)\
**Post date:** [December 6, 2019, 6:42am UTC](https://forum.rasa.com/t/memory-issue/22143/1 "2019-12-06T06:42:17Z")

</div>

Using this pipeline

- name: “SpacyNLP”
- name: “WhitespaceTokenizer”
- name: “SpacyFeaturizer”
- name: “RegexFeaturizer”
- name: “EntitySynonymMapper”
- name: “SklearnIntentClassifier”
- name: “CRFEntityExtractor”
- name: “DucklingHTTPExtractor” url: “[http://localhost:8000](http://localhost:8000)” dimensions: [“time”,“number”,“distance”,“email” , “amount-of-money”] locale: “en\_GB” timezone: “Europe/London” policies:

Memory consumption increases as size of nlu data increases. Please recommend the best pipeline so that memory consumption handled

Thanks

---

<div class="post-metadata">

**Author:** ![Tobias\_Wochinger](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/tobias_wochinger/32/4580_2.png) [@Tobias\_Wochinger](https://forum.rasa.com/u/Tobias_Wochinger)\
**Post date:** [December 6, 2019, 8:51am UTC](https://forum.rasa.com/t/memory-issue/22143/2 "2019-12-06T08:51:18Z")

</div>

> Memory consumption increases as size of nlu data increases. Please recommend the best pipeline so that memory consumption handled

What would you consider as “memory consumption is handled”?

---

<div class="post-metadata">

**Author:** ![shashank34](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/shashank34/32/5110_2.png) [@shashank34](https://forum.rasa.com/u/shashank34)\
**Post date:** [December 6, 2019, 10:32am UTC](https://forum.rasa.com/t/memory-issue/22143/3 "2019-12-06T10:32:41Z")

</div>

@Tobias_Wochinger @dakshvar22 Here crf entity extranctor using more memory than other entity extractors . how we can handle the memory consumption in aws servers if nlu data increases as it blocks the instance ?

---

<div class="post-metadata">

**Author:** ![Tobias\_Wochinger](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/tobias_wochinger/32/4580_2.png) [@Tobias\_Wochinger](https://forum.rasa.com/u/Tobias_Wochinger)\
**Post date:** [December 10, 2019, 9:08am UTC](https://forum.rasa.com/t/memory-issue/22143/4 "2019-12-10T09:08:29Z")

</div>

> than other entity extractors

Can you name a few examples? CRF is an actually machine learning based algorithm which has to be trained from scratch in contrast to others which are either rule-based or pretrained.

Apart from giving the machines more power, you could try lowering the amount of features you feed in the component [ner\_crf docs](https://rasa.com/docs/rasa/nlu/components/#crfentityextractor). Also lowering `max_iterations` might help.