| commit | 4f4008057e6c9f15db351a5b8edab733799d20ec | [log] [tgz] |
|---|---|---|
| author | Stella Laurenzo <stellaraccident@gmail.com> | Thu Apr 04 09:55:24 2024 -0700 |
| committer | GitHub <noreply@github.com> | Thu Apr 04 09:55:24 2024 -0700 |
| tree | dca7501782c38ac3f98957f2e4ed4b30b2567d08 | |
| parent | b82c1aa4e40112317af7aba07ecc0b0806aab1a1 [diff] |
[python] Flesh out more of the python parameters API. (#16957) The existing Python parameters API was sufficient to construct parameter indexes/archives but did not fully support introspecting them from Python. New APIs: * `FileHandle`: * Implements buffer protocol for accessing host allocations. * `is_host_allocation -> bool` property * `host_allocation -> memoryview` * `__repr__` * `ParameterIndexEntry`: * Class added. * Properties: `key`, `length`, `metadata`, `is_file`, `is_splat`, `file_storage`, `file_view`, `splat_pattern`, `__repr__` * `ParameterIndex`: * `__getitem__` for index based iteration * `items()` for `dict`-like access to a list of key/value tuples * `__repr__` Includes a version upgrade of nanobind to 1.9.2 as some new features were needed (just bumped to current vs finding the exact version). This requires a patch to the Linux CI to make it set up a venv and install versions of build requirements like all of the others do (vs leaving this to whatever was packaged in the base docker).
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.