Closed ScottTodd closed 4 months ago
Hmm still need to iterate on the CPU models xfails.
Thanks, that worked. When we refactor those jobs next, we could make those steps continue to show all errors instead of just the first. (I had a branch refactoring the jobs into a matrix, can pull that in...)
This pulls in https://github.com/iree-org/iree/pull/17476 which now runs shape inference:
onnx.shape_inference.infer_shapes(model, data_prop=True)
.Generated from an already configured venv with: