# Deep learning architecture

**URL:** <https://help.galaxyproject.org/t/deep-learning-architecture/7673>\
**Category:** usegalaxy.eu support\
**Tags:** gtn-tutorial, galaxy-local\
**Created:** [March 13, 2022, 7:29am UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673 "2022-03-13T07:29:08Z")\
**Posts on this page:** 9\
**Page:** 1

<div class="post-metadata">

**Author:** ![amir](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/amir/32/539_2.png) [@amir](https://help.galaxyproject.org/u/amir)\
**Post date:** [March 13, 2022, 7:29am UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/1 "2022-03-13T07:29:08Z")

</div>

Hi  
I just finished deep learning tutorials and still some points are not clarified for me. for example : How do i should determine the number of neurons, batch size and epochs?

Regards  
Amir

---

<div class="post-metadata">

**Author:** ![gallardoalba](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/gallardoalba/32/1903_2.png) [@gallardoalba](https://help.galaxyproject.org/u/gallardoalba)\
**Post date:** [March 13, 2022, 5:20pm UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/2 "2022-03-13T17:20:36Z")

</div>

Hi @amir,  
I informed the responsible for the training; he will probably answer you in the next few days.

Regards

---

<div class="post-metadata">

**Author:** ![amir](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/amir/32/539_2.png) [@amir](https://help.galaxyproject.org/u/amir)\
**Post date:** [March 23, 2022, 8:43am UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/3 "2022-03-23T08:43:43Z")

</div>

Thank you. I think still there is a lot of problems in tools related to deep learning. it would be great if some one consider to solving those problems.

Best  
Amir

---

<div class="post-metadata">

**Author:** ![gallardoalba](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/gallardoalba/32/1903_2.png) [@gallardoalba](https://help.galaxyproject.org/u/gallardoalba)\
**Post date:** [March 29, 2022, 5:01pm UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/4 "2022-03-29T17:01:25Z")

</div>

Could you provide me with some details about additional problems? Thank you so much for your feedback. Regards!

---

<div class="post-metadata">

**Author:** ![amir](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/amir/32/539_2.png) [@amir](https://help.galaxyproject.org/u/amir)\
**Post date:** [March 30, 2022, 9:09am UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/5 "2022-03-30T09:09:02Z")

</div>

Hi  
Thank you for your response.  
You can see the first problem in attached file.

 ![Untitled](https://us1.discourse-cdn.com/flex020/uploads/galaxy/original/2X/3/365642d5c1bd74bce5f894e7791ce97f2f2f27bf.png)  
it seems the deep learning tool still has some main problems, I tried some other options but still no progress.  
also, this is galaxy training datasets.  
Best  
Amir

---

<div class="post-metadata">

**Author:** ![anuprulez](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/anuprulez/32/2877_2.png) [@anuprulez](https://help.galaxyproject.org/u/anuprulez)\
**Post date:** [March 30, 2022, 12:46pm UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/6 "2022-03-30T12:46:38Z")

</div>

Hi Amir,

Thanks for using the deep learning tutorials. I agree these “jobs” shown in grey did not start because they require GPUs to process and none of them is free at the moment. Once, at least one or more become free, these jobs would start. We are currently working on finding an ideal job scheduling scheme for such tools to maximize the usage of CPUs and GPUs that we currently have.

---

<div class="post-metadata">

**Author:** ![anuprulez](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/anuprulez/32/2877_2.png) [@anuprulez](https://help.galaxyproject.org/u/anuprulez)\
**Post date:** [March 30, 2022, 12:53pm UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/7 "2022-03-30T12:53:46Z")

</div>

Hi Amir,

When you use this tool " **Create deep learning model** with an optimizer, loss function and fit parameters", you can find several endpoints for changing parameters you mentioned such as the number of epochs, batch size, optimizer, learning rate and so on. In this tool " **Create a deep learning model architecture** using Keras", you can set up your layers defining their types, the number of neurons and so on.

 ![model_builder](https://us1.discourse-cdn.com/flex020/uploads/galaxy/original/2X/b/b88cd878fafb78bd6fa5d58b1d08193950d8aa1d.png)

---

<div class="post-metadata">

**Author:** ![anuprulez](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/anuprulez/32/2877_2.png) [@anuprulez](https://help.galaxyproject.org/u/anuprulez)\
**Post date:** [March 30, 2022, 1:12pm UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/8 "2022-03-30T13:12:47Z")

</div>

Also, you will need to restart these jobs

---

<div class="post-metadata">

**Author:** ![amir](https://sea2.discourse-cdn.com/flex020/user_avatar/help.galaxyproject.org/amir/32/539_2.png) [@amir](https://help.galaxyproject.org/u/amir)\
**Post date:** [March 30, 2022, 2:04pm UTC](https://help.galaxyproject.org/t/deep-learning-architecture/7673/9 "2022-03-30T14:04:23Z")

</div>

Thanks for your response.  
In tutorial, didnt mentioned anywhere choosing batch size … etc base on what. some of options were explained but unfortunately these points were missed