Open
Remove onnxruntime-easy dependency, inline its code#32
Conversation
…e.py Co-authored-by: xadupre <22452781+xadupre@users.noreply.github.com> Agent-Logs-Url: https://github.com/onnxruntime/mobius/sessions/2a25ff2f-35b5-4acf-b7e5-4f45dbca8aee
Copilot
AI
changed the title
[WIP] Remove dependency on onnxruntime-easy and import code if necessary
Remove onnxruntime-easy dependency, inline its code
Mar 24, 2026
xadupre
reviewed
Mar 24, 2026
xadupre
reviewed
Mar 24, 2026
xadupre
reviewed
Mar 24, 2026
Co-authored-by: xadupre <22452781+xadupre@users.noreply.github.com> Agent-Logs-Url: https://github.com/onnxruntime/mobius/sessions/60380e38-d669-4ed2-b8cd-d74d9ee42c9c
justinchuby
reviewed
Mar 24, 2026
…nditional Co-authored-by: justinchuby <11205048+justinchuby@users.noreply.github.com> Agent-Logs-Url: https://github.com/onnxruntime/mobius/sessions/feb5cf18-6cf6-432e-8ef5-ac0224beb18a
Contributor
Author
|
Just as a heads up, I was blocked by some firewall rules while working on your feedback. Expand below for details. Warning Firewall rules blocked me from connecting to one or more addresses (expand for details)I tried to connect to the following addresses, but was blocked by firewall rules:
If you need me to access, download, or install something from one of these locations, you can either:
|
🏗️ Architecture Diff
No architecture changes detected. ✅ Legend: ⚪ No change · 🔵 Minor (attrs/inits) · 🟡 Moderate (nodes added/removed) · 🔴 Major (interface changed) |
Performance Comparison
|
xadupre
marked this pull request as ready for review
March 25, 2026 10:05
Contributor
There was a problem hiding this comment.
Pull request overview
This PR removes the onnxruntime-easy test dependency by inlining equivalent ORT session/dtype handling logic into the repo’s internal testing utilities, simplifying the dependency graph while keeping ONNX Runtime inference available for tests/examples.
Changes:
- Replaced
onnxruntime-easyusage with a directonnxruntime.InferenceSessionwrapper and OrtValue conversion utilities. - Added
onnxruntimeandml_dtypesto thetestingoptional dependencies and removedonnxruntime-easy.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| src/mobius/_testing/ort_inference.py | Inlines ORT session creation + OrtValue conversion to replace onnxruntime-easy. |
| pyproject.toml | Updates testing extras to depend on onnxruntime + ml_dtypes instead of onnxruntime-easy. |
Comment on lines
+22
to
+38
| def _to_ort_value(value: np.ndarray, device: str = "cpu") -> ort.OrtValue: | ||
| """Convert a numpy array to an OrtValue, handling special ml_dtypes dtypes.""" | ||
| # Use DLPack when available (e.g. torch tensors or non-zero-size arrays) | ||
| if hasattr(value, "__dlpack__"): | ||
| is_zero_size = hasattr(value, "size") and value.size == 0 | ||
| if not is_zero_size: | ||
| return ort.OrtValue(_ort_c.OrtValue.from_dlpack(value.__dlpack__(), False), value) | ||
| if isinstance(value, np.ndarray): | ||
| try: | ||
| onnx_type = ir.DataType.from_numpy(value.dtype) | ||
| except TypeError: | ||
| pass | ||
| else: | ||
| return ort.OrtValue.ortvalue_from_numpy_with_onnx_type( | ||
| value, onnx_element_type=onnx_type.value | ||
| ) | ||
| return ort.OrtValue.ortvalue_from_numpy(np.asarray(value), device) |
There was a problem hiding this comment.
_to_ort_value(..., device=...) only uses device in the final fallback path; the DLPack and ortvalue_from_numpy_with_onnx_type branches ignore it. Either remove the parameter to avoid a misleading API, or ensure the chosen OrtValue creation path respects the requested device (especially if this is intended to support CUDA inputs).
gramalingam
approved these changes
Mar 25, 2026
Member
|
We need to run the tests locally because the CIs are currently broken. Otherwise lgtm. |
Member
|
Is this ready? |
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
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.
onnxruntime-easyintosrc/mobius/_testing/ort_inference.pyonnxruntime-easyfrompyproject.tomltesting dependencies, addonnxruntimeandml_dtypesinsteadonnx.TensorProtoconstants instead of hardcoded ONNX element type integersOriginal prompt
⚡ Quickly spin up Copilot coding agent tasks from anywhere on your macOS or Windows machine with Raycast.