Add deep learning 2022 examples

[isclaireya]

No description provided.

[JohnGriffiths]

Hi Claire.

This is looking very promising.

Unfortunately I am having a few issues with this PR.

I tried running the code but I ran into issues with a .json file that is accessed in your personal folder, and then later a .npz file being accessed that I'm not sure where it came from.

These also can't be fully resolved in this one because you have interdependent code distributed across 2 PRs.

Please can you make a new PR fixing the following:

• Include both the examples file and the model code in the depr folder that it calls in the same PR
• Check that the code runs
• Check that the sphinx website builds and the page produced looks ok
• Add the file path to the example to the sphinx minigallery list in the 'brain cognition' index.rst folder. This is also necessary to check the sphinx website in prev point

LMK if you need any pointers

j

[JohnGriffiths]

ok actually bear with me - I think I can do some of this now and add some fixes. will follow up.

[isclaireya]

Hi John,
I am on the path to the "Add the file path to the example to the sphinx minigallery list in the 'brain cognition' index.rst folder". Sorry for the delay.
Another issue is that I attempted to store the data on GitHub, but due to the size of the files, the process failed. I am trying to find an alternative approach.

[JohnGriffiths]

ok update - I have nearly finished a PR adding a new data fetcher with all the files used in your example

when that's merged I will need you to do a new PR doing all of the above, but run the fetcher first and refer to those paths when loading files

look at the other examples for how that should look

[JohnGriffiths]

ok done - c.f. this PR, now merged to main

#297

so you can do things like

from whobpyt.datasets.fetchers import fetch_eggriffiths2022
eg_dir = fetch_eggriffiths2022()

remember that the above command will not re-download the data file if you have already done so; namely if the folder already exists. if it exists and does not have all the content in that would be an issue however

at the start, and then when loading data, use something like

f = os.path.join(eg_dir, 'hagmann', 'weights.txt')
W = np.loadtxt(f)

make sure you check that any files you want to load are in the folder and that they are correctly downloading