[Python] Support bfloat16 in Python bindings via ml_dtypes - #31677
Draft
bhushan23 wants to merge 1 commit into
Draft
[Python] Support bfloat16 in Python bindings via ml_dtypes#31677bhushan23 wants to merge 1 commit into
bhushan23 wants to merge 1 commit into
Conversation
Marshal bfloat16 tensors through OrtValue and session.run using the ml_dtypes.bfloat16 numpy extension dtype, since numpy has no native bfloat16. The dtype's type_num is resolved lazily at first use and cached, then wired into IsNumericDType, OnnxRuntimeTensorToNumpyType, and NumpyTypeToOnnxRuntimeTensorType so bfloat16 arrays flow through the standard input/output paths. Adds ml_dtypes as a base runtime dependency and covers the plumbing with OrtValue round-trip and Identity/Add session.run tests.
|
Azure Pipelines: There may be pipelines that require an authorized user to comment /azp run to run. |
Contributor
|
See #22306 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Marshal bfloat16 tensors through OrtValue and session.run using the ml_dtypes.bfloat16 numpy extension dtype, since numpy has no native bfloat16. The dtype's type_num is resolved lazily at first use and cached, then wired into IsNumericDType, OnnxRuntimeTensorToNumpyType, and NumpyTypeToOnnxRuntimeTensorType so bfloat16 arrays flow through the standard input/output paths.
Adds ml_dtypes as a base runtime dependency and covers the plumbing with OrtValue round-trip and Identity/Add session.run tests.
Motivation and Context
onnx-runtime has support for bfloat16 for CUDA EP but is not usable via python APIs e.g. session.run due to lack of support for bfloat16 dtype in numpy.
Projects such as aimet-onnx depends on onnx-runtime for quantization and with this change calibrate with bf16