RisingPhoelix
RisingPhoelix
Have a look at #2427. Sounds like this will help
Hi, are you still working on this or is it planned to implement this feature in the future? I have a very similar workflow where I want to skip the...
Thanks, @albi3ro ! Sounds great, sorry I didnt find the PR before posting this.
Hi @albi3ro and @CatalinaAlbornoz ! Thank you and everyone else for your work on Pennylane 0.41. Unfortunately, this issue persists after updating Pennylane. However, the output of jax.checking_leaks is different...
Maybe it's a Heisenbug... :') I am currently running the function in a jitted loop for many iterations and memory usage seems to be stable. So maybe we ignore it....
Now I tried to use this in a much more convoluted project with jitted optimization loop and encountered this leaked tracer also without `jax.checking_leaks` but it's referred to as a...
Finding the minimal example was easier than expected: ```python import jax import pennylane as qml @jax.jit def sample_circuit(phi, theta, key): # Device construction should happen inside a `jax.jit` decorated #...
That's unfortunate.. So there is currently no possibility to use Pennylane with finite shots and parameter shift rule in a JAX-compiled optimization loop? Or should this be implemented in a...
Hi all, just to let you and others know, I now implemented a custom parameter shift rule and shot-based measurement for qml.probs. Unfortunately, I am not allowed to share the...
Hey @CatalinaAlbornoz, feature request is created (see above), I hope it helps! :)