asraa

OK the codegen for this relation is quick and simple and looks correct (minus why it's not just S(0, c1, 0, c1)) ``` or (int c1 = 0; c1

I think it might be this link now https://github.com/corelab-src/dacapo/tree/dacapo/include/hecate/Dialect/CKKS/IR ah they're just a port

I wanted to revisit this since torch-mlir actually outputs dense resources for the large constants in the IR. For large models, we know we may want to split the model...

I agree that the shapes of the application data can be inferred from the packing, but do we still need to hold on to the underlying data type? I would...

Thanks for the test case! I'll debug it today. It might be something to do with padding - the example https://github.com/google/heir/tree/aae0d07bb44a43e660e72c5a6567b1a8e70cfc83/tests/Examples/openfhe/ckks/conv_2dis 4x4 with a 3x3 kernel and seemed ok, so...

Hey! Thanks so much for posting this issue, it was indeed a bug - we treated the indexes to access the tensors in the IR as unsigned integers (index types...

> What do you mean by policy? Something only we would follow or also our users? Users should also probably not use patch versions that have superseding ones, right? Otherwise...

Might be possible to detect `self-hosted` label here: https://docs.github.com/en/actions/hosting-your-own-runners/using-self-hosted-runners-in-a-workflow#using-default-labels-to-route-jobs