commit | 47693b58d955c541e2b9fc7aa977efd18be62e1e | [log] [tgz] |
---|---|---|
author | Scott Todd <scott.todd0@gmail.com> | Fri May 03 17:05:02 2024 -0700 |
committer | GitHub <noreply@github.com> | Fri May 03 17:05:02 2024 -0700 |
tree | ffc6a811309975cba98a7d4afe5d5d40b9503c28 | |
parent | 30aa5e5db7ba678e3238a97809bfa5e82ab65d1b [diff] |
Bump SHARK-TestSuite commit to include npy->bin output comparison. (#17281) This fixes https://github.com/iree-org/iree/issues/16674 by switching from comparing against .npy files to comparing against .bin files thanks to https://github.com/nod-ai/SHARK-TestSuite/pull/212. For example with https://github.com/nod-ai/SHARK-TestSuite/tree/main/iree_tests/onnx/node/generated/test_add_uint8: * model.mlir: ```mlir module { func.func @test_add_uint8(%arg0: !torch.vtensor<[3,4,5],ui8>, %arg1: !torch.vtensor<[3,4,5],ui8>) -> !torch.vtensor<[3,4,5],ui8> attributes {torch.onnx_meta.ir_version = 7 : si64, torch.onnx_meta.opset_version = 17 : si64, torch.onnx_meta.producer_name = "backend-test", torch.onnx_meta.producer_version = ""} { %none = torch.constant.none %0 = torch.operator "onnx.Add"(%arg0, %arg1) : (!torch.vtensor<[3,4,5],ui8>, !torch.vtensor<[3,4,5],ui8>) -> !torch.vtensor<[3,4,5],ui8> return %0 : !torch.vtensor<[3,4,5],ui8> } } ``` * `test_data_flags.txt` before: ``` --input=@input_0.npy --input=@input_1.npy --expected_output=@output_0.npy ``` with these errors: ``` [FAILED] result[0]: metadata is 3x4x5xi8; expected that the view matches 3x4x5xui8; expected that the view is equal to contents of a view of 3x4x5xui8 expected: 3x4x5xui8=[[18 24 22 22 20][14 38 44 14 27][30 14 20 12 20][36 25 31 6 8]][[28 30 19 7 18][26 33 23 34 19][29 39 14 32 44][26 25 18 28 24]][[24 36 16 28 25][16 23 22 20 39][27 20 34 19 26][16 17 8 16 11]] actual: 3x4x5xi8=[[18 24 22 22 20][14 38 44 14 27][30 14 20 12 20][36 25 31 6 8]][[28 30 19 7 18][26 33 23 34 19][29 39 14 32 44][26 25 18 28 24]][[24 36 16 28 25][16 23 22 20 39][27 20 34 19 26][16 17 8 16 11]] ``` * `test_data_flags.txt` after: ``` --input=3x4x5xui8=@input_0.bin --input=3x4x5xui8=@input_1.bin --expected_output=3x4x5xi8=@output_0.bin ``` (that still seems a bit mismatched? it does pass the test though)
IREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.