[Add] Dockerfile

[ravinkohli]

This resolves issue #231. But it is an incomplete implementation. The rest has been raised as another issue #315.

[codecov]

Codecov Report

Merging #314 (cd57a92) into development (4a29852) will decrease coverage by 0.23%.
The diff coverage is n/a.

@@ Coverage Diff @@ ## development #314 +/- ## =============================================== - Coverage 82.09% 81.86% -0.24%  =============================================== Files 151 151 Lines 8651 8651 Branches 1322 1322 =============================================== - Hits 7102 7082 -20  - Misses 1084 1104 +20  Partials 465 465 

Impacted Files	Coverage Δ
...eline/components/setup/optimizer/base_optimizer.py	71.42% <0.00%> (-19.05%)	⬇️
...abular_preprocessing/base_tabular_preprocessing.py	66.66% <0.00%> (-13.34%)	⬇️
...ne/components/setup/lr_scheduler/base_scheduler.py	86.20% <0.00%> (-6.90%)	⬇️
...up/network_initializer/base_network_initializer.py	93.54% <0.00%> (-6.46%)	⬇️
.../pipeline/components/setup/network/base_network.py	88.05% <0.00%> (-5.98%)	⬇️
...nts/setup/early_preprocessor/EarlyPreprocessing.py	85.71% <0.00%> (-5.72%)	⬇️
...reprocessing/feature_preprocessing/TruncatedSVD.py	89.65% <0.00%> (-3.45%)	⬇️
...peline/components/training/trainer/MixUpTrainer.py	97.14% <0.00%> (-2.86%)	⬇️
...omponents/training/data_loader/base_data_loader.py	90.54% <0.00%> (-2.71%)	⬇️
autoPyTorch/pipeline/components/base_component.py	80.95% <0.00%> (-1.91%)	⬇️
... and 2 more

---

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 4a29852...cd57a92. Read the comment docs.

[nabenabe0928]

I think we should use any one of the docker images from here because some codes are assuming that we can use torch on CUDA. FYI, basically people use xxx-devel-xxx.

Additional reference here

[ravinkohli]

It's fine to use this ubuntu container for the dockerfile that we upload to the docker hub. I think the most general use case is to use CPU. If someone would like to use it with CUDA, it is quite easy for them to edit this file.

[nabenabe0928]

Btw, i don't think pynisher works properly when we use CUDA, but have you ever tried any runs on CUDA?

[ravinkohli]

Actually thats a good point. I'll create an issue and a subsequent PR where we can use docker to run all our tests on cuda. What do you think?

[nabenabe0928]

Sounds great:)