| commit | caacdf60c4de914b750e0ca027a9917cb45bc45e | [log] [tgz] |
|---|---|---|
| author | Jakub Kuderski <kubak@google.com> | Tue Apr 11 13:24:32 2023 -0400 |
| committer | GitHub <noreply@github.com> | Tue Apr 11 13:24:32 2023 -0400 |
| tree | 1656fa53e63cba4f686fffe5b181519258629563 | |
| parent | 5853b758dbadccd77246987e38f12acefbcc1a2b [diff] |
Add initial StableHLO to Linalg lowering pass files (#12957) This is a port of the MHLO to Linalg lowering from https://github.com/tensorflow/mlir-hlo. The iree-mhlo-fork commit used to port the conversion: 1f096a793ab7f73ae8f62deb8b6502c543763ca1. The imported files are relicensed under the [Google CLA](https://cla.developers.google.com/about/google-individual) from the Apache 2.0 license (Tensorflow) to the nearly-identical Apache 2.0 with the LLVM exceptions license (IREE). The initial import covers the lowering of StableHLO ops that can be trivially mapped to their MHLO counterparts. More complicated ops, like convolutions, gather, or rng, are not ported yet. In porting MHLO conversions and tests to operate on StableHLO ops, I changed all namespaces, header guards, and copyright headers, and formatted all files to match the conventions used by IREE. Any addition modifications were a non-goal. I plan to reorganizanize and clean this up further after the initial porting. Issue: https://github.com/openxla/iree/issues/12678
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.