Open
Description
Lately in a lot of our CI runs we have been running into issues with the gradients_test -> bazel test //tensorflow_quantum/python/differentiators/gradient_test running for a very long time and sometimes flaking. We should fix this to ideally cut down the runtime and make it more reliable. Perhaps move away from fuzz testing against cirq and instead test against each gate we support individually followed by minimal fuzz testing ?