qiskit-machine-learning icon indicating copy to clipboard operation
qiskit-machine-learning copied to clipboard

Add `SamplerQNN` class using terra primitives

Open ElePT opened this issue 3 years ago • 3 comments
trafficstars

Summary

This PR presents a reference implementation of a Qiskit NeuralNetwork that leverages the newly introduced primitives (so far, this class is based on the terra interface). It closes issue #400.

Details and comments

This PR is blocked by the release of Qiskit-terra that will include gradients and changes in the primitives interface.

Open Points of Discussion

  • [ ] Where to set keyword args -> all keyword?
  • [ ] Hoy to define typehints: list vs sequence vs numpy array...

TO-DO

  • [ ] Add reno
  • [ ] Add deprecation warnings

ElePT avatar Jul 11 '22 11:07 ElePT

Why would this not be in neural networks (same comment for the kernel PR but in kernels of course) and subclass NeuralNetwork. I would think we would place this for what it is not what its built from. I realize its all draft and work in progress, so I am not going to comment beyond this aspect for now. Maybe because its all work in progress it just did not get that far yet.

woodsp-ibm avatar Jul 13 '22 20:07 woodsp-ibm

Why would this not be in neural networks (same comment for the kernel PR but in kernels of course) and subclass NeuralNetwork. I would think we would place this for what it is not what its built from. I realize its all draft and work in progress, so I am not going to comment beyond this aspect for now. Maybe because its all work in progress it just did not get that far yet.

Yes, I totally agree, I just moved my primitives PoC directory as-is from a different repo, but the class definitely fits better into neural_networks.

ElePT avatar Jul 14 '22 12:07 ElePT

Pull Request Test Coverage Report for Build 3373322648

  • 157 of 170 (92.35%) changed or added relevant lines in 5 files are covered.
  • 2 unchanged lines in 1 file lost coverage.
  • Overall coverage increased (+0.07%) to 88.925%

Changes Missing Coverage Covered Lines Changed/Added Lines %
qiskit_machine_learning/neural_networks/sampler_qnn.py 130 143 90.91%
<!-- Total: 157 170
Files with Coverage Reduction New Missed Lines %
qiskit_machine_learning/neural_networks/neural_network.py 2 92.08%
<!-- Total: 2
Totals Coverage Status
Change from base Build 3366556544: 0.07%
Covered Lines: 3541
Relevant Lines: 3982

💛 - Coveralls

coveralls avatar Oct 12 '22 16:10 coveralls