Deeponet

[ayushinav]

Test Summary:         | Pass  Error  Total     Time
LuxNeuralOperators.jl |   43      1     44  5m07.3s
  Layers              |   32            32  2m49.3s
  FNO                 |    8             8  2m14.4s
  DeepONet            |    3      1      4     3.6s
    DeepONet: CPU     |    3             3     1.1s
    DeepONet: CUDA    |           1      1     2.3s

The failing test has the following main lines in the error message.
ERROR: Method overwriting is not permitted during Module precompilation. Use ``__precompile__(false)`` to opt-out of precompilation.

and ArgumentError: Illegal conversion of a CUDA.DeviceMemory to a Ptr{Float32}.
Wasn't completely sure about this and thought about putting it here.

The first DeepONet function provides the feature compatibility with NeuralOperatorrs.jl and provides the more commonly used Dense layers to construct the branch and the trunk nets. Its documentation can be accessed by help?> DeepONet(;) or any other matching dispatch.
The second function connects any two architectures provided they have compatible last layers. Its documentation pops up when someone types in help?> DeepONet.
Also, had to do a small workaround to get the printing of architectures better.
We can make tutorials for both implementations. Putting the ideas down here until we make docs and tutorials and before we merge them into NeuralOperators.jl.

[avik-pal]

Also cc @ChrisRackauckas

[avik-pal]

Also can you open a PR setting up CI workflows. Copy the ones over from https://github.com/LuxDL/Boltz.jl/. Don't have to setup the documentation workflows, we will put the documentation of the package in the main lux repo https://lux.csail.mit.edu/stable/api/Domain_Specific_Modeling/Boltz

[avik-pal]

Rebase I have added CI (CPU, CUDA, and AMDGPU).

[ayushinav]

Rebase I have added CI (CPU, CUDA, and AMDGPU).

Thanks, that helps!

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.53%. Comparing base (5b30d54) to head (ebf9e87).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main       #5      +/-   ##
==========================================
+ Coverage   97.05%   97.53%   +0.47%     
==========================================
  Files           7        8       +1     
  Lines          68       81      +13     
==========================================
+ Hits           66       79      +13     
  Misses          2        2              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ayushinav]

Looks like it's all good? @avik-pal

[avik-pal]

Format check is failing, run the formatter and it should be good.