I am working on the MNIST data on my own. The idea is to use different values for the number of hidden layers, number of nodes in a given layer, etc. How do you organize these things while you are working on creating a model for a problem? DO you do everything in one code file or you use different code files for choosing the best?
1
-
$\begingroup$ Hi and welcome to AI SE! What do you mean by "code file"? Is this just a programming question? If it's just a programming question, it's off-topic here. See ai.stackexchange.com/help/on-topic for more details. Please, edit your post to clarify this. $\endgroup$– nbroApr 27, 2020 at 14:12
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
3
it seems to me that you are talking about hyperparameter tuning and effect of hyperparameters on the network in general. If you are working with tensorflow, I recommend you to look into tensorboard.
Hands-on TensorBoard can be a good starting point.
-
$\begingroup$ I wanted to know how do you work with different models for a given problem. Say you have a model with 10 hidden layers and another with 7 hidden layers. How do you organize and set up different models that are meant for a project in your computer or cloud? $\endgroup$ Apr 28, 2020 at 3:48
-
$\begingroup$ You can write function where you pass input parameter as number of layers and other hyperparameters and run that model.Then you can save that model which can be reused. It also depends what do you want to do with the model. What do you do with different models for same problem? $\endgroup$ Apr 28, 2020 at 19:37
-
$\begingroup$ For a given classification problem, you check out a different number of hidden layers or you check out for a different number of nodes in each layer. I am using Jupyter Notebooks and it gets hard to do everything in just one .ipynb file. What can I do to organize my code for doing the above mentioned without making everything cumbersome. Thanks. $\endgroup$ Apr 29, 2020 at 3:30