Add tests for cross attention

[jackzhxng]

Summary

Add tests for the true branch of the torch.cond, for testing the behavior of y = None, when the encoder cache is populated and no additional encoder inputs are coming in.

Note: torch.export where the same tensor is used for multiple args in the example inputs, e.g. (self.x, self.x) results in a weird graph where the two parameters seem to be treated as the same node. As a basic example, when I would pass in something like ep.module()(torch.zeros(10), torch.ones(10)), when tracing through the interpreter, for ops where I am expecting the arg to be the first parameter, torch.zeros(10), it takes the second parameter, torch.ones(10). @angelayi is this expected behavior?

Test plan

Added tests

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/7609

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

⏳ No Failures, 1 Pending

As of commit 57f7414 with merge base 73f50df ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.