torch-mlir/test/Conversion/TorchOnnxToTorch
zjgarvey 1201babb9f
[ONNX] rework some reduction op lowerings (#3870)
- Refactors more "onnx.ReduceXXX" patterns through helper function.
- Fixes bug with iterating unconditionally on `output_dim == 1` during
`dimList` inference.

This change results in passes for the following 11 models:

crossvit_15_240
crossvit_15_dagger_240
crossvit_15_dagger_408
crossvit_18_240
crossvit_18_dagger_240
crossvit_18_dagger_408
crossvit_9_240
crossvit_9_dagger_240
crossvit_base_240
crossvit_small_240
crossvit_tiny_240

---------

Co-authored-by: Vinayak Dev <104419489+vinayakdsci@users.noreply.github.com>
2024-11-14 16:25:28 +00:00
..
ops Cast static/dynamic shape for onnx.If branches to match result type (#3828) 2024-11-01 12:10:59 -07:00
simple_ops_a_to_f.mlir Adds onnx ConvTranspose support for autopadding. (#3797) 2024-10-18 12:31:33 -05:00
simple_ops_g_to_p.mlir Added support for Maxpool (Autopad) (#3774) 2024-10-23 13:04:50 +00:00
simple_ops_q_to_z.mlir [ONNX] rework some reduction op lowerings (#3870) 2024-11-14 16:25:28 +00:00
unsupported_fb_opt_ops.mlir Add attributes support for onnx cumsum op (#3241) 2024-05-11 02:09:01 +08:00