commit | ae286ab16cef95cd40731a2b0c1f78f7c73cdf92 | [log] [tgz] |
---|---|---|
author | hertschuh <1091026+hertschuh@users.noreply.github.com> | Thu Sep 22 22:43:28 2022 +0000 |
committer | GitHub <noreply@github.com> | Thu Sep 22 22:43:28 2022 +0000 |
tree | 4dd765f27d427750fe0cb94c79772bbd1e808a42 | |
parent | 9540b6d382410e232f3985c48c0583c187646854 [diff] |
Don't use fully dynamic shapes for Keras layers that don't support them (#10521) Keras layers that do not support fully dynamic shapes are tested with a dynamic batch size only. Keras currently calls `set_shape` based on the shape of the `Input` that is defined. However, this logic in Keras is being removed as it is incorrect in some scenarios. Because of this, the IREE "dynamic_dims" unit tests were actually running with a fully static shape and not exercising operations with a dynamic shape. Testing with dynamic shapes uncovered the fact that some layers do not support it.
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.