| commit | 60a2566f26b82fb8dd736d5468623b9b7a898a67 | [log] [tgz] |
|---|---|---|
| author | Kunwar Grover <51270680+Groverkss@users.noreply.github.com> | Fri Sep 29 03:30:21 2023 +0530 |
| committer | GitHub <noreply@github.com> | Thu Sep 28 15:00:21 2023 -0700 |
| tree | 69e48931d7abbd8ba5914098d1d31eef3006bf2b | |
| parent | 1a635643583e8787030b9fe57fae5d0543644f65 [diff] |
Add iree-reduce tool (#14892) This patch adds an `iree-reduce` tool, a test case reduction tool which reduces the size of a given test case, producing the same error as the original test case. The tool implementation is split into 2 parts: Framework and Strategies. "Framework" is independent of IREE and implements the Delta Debugging Algorithm. "Strategies" implements IREE-specific delta passes for test case reduction. Currently, 3 delta passes are implemented: 1. ReduceLinalgOnTensorsDelta: Replaces linalg operations with a fill operation 2. ReduceFlowDispatchOperandToResultDelta: Replace the result of a flow.dispatch operation with it's operand, if they have the same shape 3. ReduceFlowDispatchResultBySplatDelta: Replace a flow.dispatch operation with a splat operation. The delta passes are initial deltas that worked well. They can be upgraded as we need arises.
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.