Saving and recalling a dictionary of trained models

[alfredaita]

I know this has been answered before, but I have some questions?

Using Keras 2.8 I have trained a dictionary of trained models. I'm designing it intending to use backtest_ml.

I called each of the functions separately to test. The dictionary of the trained model took about an hour to train; this is probably too long for submission purposes at this point.
I saved the model to my area as a CSV by converting it into a data frame [Pandas].
When I shut down and want to recall it, will it function? I examined the data, and it seems like it should. But I have not tried.
I edited the init.ipynb file but each time I have to run it, no problem. would not be convenient.
Do I have access to GPU's or TPU's

Thanks in advance
Alfred

[support]

@alfredaita Hello, you are addressing several points here:

1. I called each of the functions separately to test. The dictionary of the trained model took about an hour to train; this is probably too long for submission purposes at this point.

--> You have at most 10 minutes per point in time on the server. Does the training take one hour for each point in time? Or globally once? That is a big difference.

2. I saved the model to my area as a CSV by converting it into a data frame [Pandas].

--> Good idea. Yes, the csv file will persist in your area.

3. I edited the init.ipynb file but each time I have to run it, no problem. would not be convenient.

--> Yes, you have to run it each time, it is a design choice we used. A bit annoying but there are advantages on the server side.

4. Do I have access to GPU's or TPU's

--> So far no. But you can use the toolbox locally:

https://quantiacs.com/documentation/en/user_guide/local_development.html#conda-environment

[antinomy]

@alfredaita
In case you don't want to run init.py every time in order to install external libraries, I came up with a solution for this. You basically install the library in a folder in your home directory and let the strategy create symlinks to the module path at runtime. More details in this post.