search

PennyLaneAI / comments

0 stars 0 forks source link

qml/demos/tutorial_post-variational_quantum_neural_networks/ #24

Open utterances-bot opened 1 month ago

utterances-bot commented 1 month ago

Post-variational quantum neural networks | PennyLane Demos

Learn about post-variational quantum neural networks

https://pennylane.ai/qml/demos/tutorial_post-variational_quantum_neural_networks/

mehrnia commented 1 month ago

Hi, Thanks for great context I was wondering why you did not compute the result based on all qubits. you may call 8-local?

georgepwhuang commented 1 month ago

Hi mehrnia! Yes, indeed an 8-local computation is possible, but in consideration of the cost of measurement on all possible 8-local measurements, we have opted not to do so, but to instead use the hybrid method to attempt to introduce some factors of global cost to the network.

supreethmv commented 2 weeks ago

Hi, for tracking the optimization process and avoid over-fitting the train set, how would you track also the test loss and accuracy at each update step. I tried with callback, but the callback (or debug) functionality available in JAX explicitly suggests that we shouldn't be doing any compute intensive tasks, like finding the test loss and accuracy. I can get the performance on the test set only at the end of the training, and there's no way I can get the test losses in the middle of the training or is there a workaround for this?