I have written a program using the Qiskit-Pennylane plugin that creates a hybrid NN that I would like to train using the IBM Qiskit backend, specifically with the Pay-As-You-Go model.
I was wondering if there was any way I can predict the total amount of IBMQ runtime I will need to train my model. I would like to approximate how much the job will cost before I send it.
Hi @Raghav_Ramki, unfortunately we don’t have a way to approximate these costs. I would make sure to be careful about the number of shots that you set for your device (if you don’t know how to set this please let me know), and also make sure that you know how many iterations and epochs you need. Batching doesn’t work well on Qiskit so you should also be careful about that. My first recommendation would be to try everything on a simulator and look for their cost-calculation formula so that you can approximate the costs before you actually pay them.
I hope this helps!