Binder -- Enable online viewing/running of Jupyter examples

[lyndond]

Changes:

• Fixed bug w/ eigendistortions when k=1
• from tqdm.auto import tqdm for nicer notebook formatting when running synthesis with eigendistortions
• Re-ran 02_Eigendistortions.ipynb and Demo_Eigendistortion.ipynb w/ fixed code
• removed pt.imshow from notebooks
• Added Binder badges to README.md that will always point to master branch examples folder
• Added Binder badges to individual ipynb for 02_Eigendistortions.ipynb and Demo_Eigendistortion.ipynb so people will be able to open them online from readthedocs.io or from the github repo
• Added packages and subpackages to setup.py in order for Binder to pip install properly. Otherwise, it just throws package not found errors.
• Limit setup.py to torch<1.8 because steerpyr is breaking tons of steerpyr related tests bc of the torch.fft module, but Nikhil's future PR should fix it.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[lyndond]

added simple in-line comments explaining changes

[codecov-io]

Codecov Report

Merging #107 (92590fb) into master (a3d031d) will decrease coverage by 6.35%.
The diff coverage is 25.00%.

@@            Coverage Diff             @@
##           master     #107      +/-   ##
==========================================
- Coverage   76.63%   70.28%   -6.36%     
==========================================
  Files          36       36              
  Lines        3463     3463              
  Branches      741      741              
==========================================
- Hits         2654     2434     -220     
- Misses        618      843     +225     
+ Partials      191      186       -5     

Impacted Files	Coverage Δ
plenoptic/synthesize/eigendistortion.py	17.68% <25.00%> (-71.43%)	⬇️
plenoptic/metric/model_metric.py	28.57% <0.00%> (-71.43%)	⬇️
plenoptic/synthesize/autodiff.py	20.00% <0.00%> (-66.67%)	⬇️
...simulate/canonical_computations/non_linearities.py	20.51% <0.00%> (-64.11%)	⬇️
plenoptic/metric/perceptual_distance.py	67.92% <0.00%> (-17.93%)	⬇️
plenoptic/tools/signal.py	33.72% <0.00%> (-15.70%)	⬇️
plenoptic/simulate/models/frontend.py	89.79% <0.00%> (-10.21%)	⬇️
plenoptic/tools/data.py	70.76% <0.00%> (-9.24%)	⬇️
plenoptic/simulate/models/ventral_stream.py	78.54% <0.00%> (-0.70%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a3d031d...92590fb. Read the comment docs.

[billbrod]

The only change I think we need is the author/author_email fields of setup.py, but there are a couple of places I'd like to start a conversation (and probably make issues for)

[billbrod]

Looking at the notebooks, it looks like the synthesis didn't actually run? (all the progress bars say 0/X) But the images are there, so I'm confused.

[lyndond]

Yeah I see that too... strange. Maybe it's tqdm.notebook bugging out. Can't have our cake and eat it too?

[billbrod]

Ah yeah, that could definitely be it. Very strange.

[billbrod]

I'm not sure that util import is necessary, but otherwise I think this is good to go.

[billbrod]

Oh one other requested change: can you document somewhere the steps to get tqdm working in notebooks? with tqdm.auto, it won't fallback to the ascii version if it can't use the fancy one -- it will raise confusing exceptions (speaking from experience). So our requirements should point out that, in addition to installing jupyter, they'll also need ipywidgets (I think that's the only extra requirement)

[billbrod]

All looks good to me!

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

ThomasYerxa commented on 2021-03-18T19:11:01Z
----------------------------------------------------------------

could be just me but I couldn't see the most noticeable distortion at first, I guess that is the whole point but may want to consider cranking up the distortion factor?

[ThomasYerxa]

looks good to me, left one comment on the tutorial notebook that isn't related to this PR.