You are viewing a plain text version of this content. The canonical link for it is here.
Posted to discuss-archive@tvm.apache.org by "locnd182644 (via GitHub)" <gi...@apache.org> on 2026/01/19 09:49:02 UTC
[PR] [Relax][Onnx] Support Multi Input Ops with Multidirectional Broadcasting [tvm]
locnd182644 opened a new pull request, #18673:
URL: https://github.com/apache/tvm/pull/18673
This PR support Multi Input Ops with Multidirectional Broadcasting
### Description
- Support Multi Input Ops with Multidirectional Broadcasting (Min, Max, Mean, Sum)
- Edit handle workflow for MultiInputBase:
+ Compute target shape for Multidirectional Broadcasting
+ Broadcast_to with target shape
+ Stack op
+ Reduce ops with axis same stack op
### Expected
- Example target shape:
<img width="700" height="183" alt="image" src="https://github.com/user-attachments/assets/f9569dff-588e-49c5-ae72-c5b6ea22b6f3" />
### Reference
- Multidirectional Broadcasting: https://onnx.ai/onnx/repo-docs/Broadcasting.html
- Fixed: #18592
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For additional commands, e-mail: discuss-archive-help@tvm.apache.org
Re: [PR] [Relax][Onnx] Support Multi Input Ops with Multidirectional Broadcasting [tvm]
Posted by "tlopex (via GitHub)" <gi...@apache.org>.
tlopex merged PR #18673:
URL: https://github.com/apache/tvm/pull/18673
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For additional commands, e-mail: discuss-archive-help@tvm.apache.org
Re: [PR] [Relax][Onnx] Support Multi Input Ops with Multidirectional Broadcasting [tvm]
Posted by "gemini-code-assist[bot] (via GitHub)" <gi...@apache.org>.
gemini-code-assist[bot] commented on PR #18673:
URL: https://github.com/apache/tvm/pull/18673#issuecomment-3767408921
## Summary of Changes
Hello @locnd182644, I'm Gemini Code Assist[^1]! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!
This pull request significantly enhances the ONNX frontend for Relax by introducing robust support for multidirectional broadcasting in multi-input operators. This ensures that models leveraging operations like Min, Max, Sum, and Mean with inputs of different but broadcast-compatible shapes are correctly translated and executed within the Relax framework, improving compatibility and functionality for a wider range of ONNX models.
### Highlights
* **Multidirectional Broadcasting Support**: Implemented support for multidirectional broadcasting in ONNX multi-input operators such as Min, Max, Sum, and Mean, allowing these operations to correctly handle inputs with varying shapes that require broadcasting.
* **New Broadcasting Utility**: Added a new utility function, `compute_broadcast_shape`, to determine the target shape for multidirectional broadcasting between two input shapes.
* **Updated MultiInputBase Conversion**: Modified the `MultiInputBase` conversion logic to first compute the target broadcast shape for all inputs, then apply `relax.op.broadcast_to` to each input before stacking them, and finally performing the reduction operation.
* **Comprehensive Test Cases**: Introduced new parameterized test cases (`test_multi_input_broadcasting`) to validate the correct behavior of multidirectional broadcasting for multi-input operators across various complex input shape combinations.
🧠**New Feature in Public Preview:** You can now enable **Memory** to help **Gemini Code Assist** learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. **Click [here](https://codeassist.google/code-review/login) to enable Memory in your admin console.**
<details>
<summary><b>Using Gemini Code Assist</b></summary>
<br>
The full guide for Gemini Code Assist can be found on our [documentation page](https://developers.google.com/gemini-code-assist/docs/review-github-code), here are some quick tips.
<b>Invoking Gemini</b>
You can request assistance from Gemini at any point by creating a comment using either `/gemini <command>` or `@gemini-code-assist <command>`. Below is a summary of the supported commands on the current page.
Feature | Command | Description
--- | --- | ---
Code Review | `/gemini review` | Performs a code review for the current pull request in its current state.
Pull Request Summary | `/gemini summary` | Provides a summary of the current pull request in its current state.
Comment | @gemini-code-assist | Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help | `/gemini help` | Displays a list of available commands.
<b>Customization</b>
To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a `.gemini/` folder in the base of the repository. Detailed instructions can be found [here](https://developers.google.com/gemini-code-assist/docs/customize-gemini-behavior-github).
<b>Limitations & Feedback</b>
Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with :thumbsup: and :thumbsdown: on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up [here](https://google.qualtrics.com/jfe/form/SV_2cyuGuTWsEw84yG).
<b>You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the [Gemini Code Assist IDE Extension](https://cloud.google.com/products/gemini/code-assist).</b>
</details>
[^1]: Review the [Privacy Notices](https://policies.google.com/privacy), [Generative AI Prohibited Use Policy](https://policies.google.com/terms/generative-ai/use-policy), [Terms of Service](https://policies.google.com/terms), and learn how to configure Gemini Code Assist in GitHub [here](https://developers.google.com/gemini-code-assist/docs/customize-gemini-behavior-github). Gemini can make mistakes, so double check it and [use code with caution](https://support.google.com/legal/answer/13505487).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For additional commands, e-mail: discuss-archive-help@tvm.apache.org
Re: [PR] [Relax][Onnx] Support Multi Input Ops with Multidirectional Broadcasting [tvm]
Posted by "gemini-code-assist[bot] (via GitHub)" <gi...@apache.org>.
gemini-code-assist[bot] commented on code in PR #18673:
URL: https://github.com/apache/tvm/pull/18673#discussion_r2704026659
##########
python/tvm/relax/frontend/onnx/onnx_frontend.py:
##########
@@ -1675,9 +1691,14 @@ def _impl_v1(cls, bb, inputs, attr, params):
output = cls.numpy_op(*np_inputs) # pylint: disable=not-callable
return relax.const(output, output.dtype)
- # Expand inputs, stack them, then perform minimum over the new axis.
- inputs = [bb.normalize(relax.op.expand_dims(i, axis=0)) for i in inputs]
- stacked_tensor = relax.op.concat(inputs, axis=0)
+ input_shapes = [inp.struct_info.shape for inp in inputs]
+ current_target_shape = input_shapes[0]
+ for next_shape in input_shapes[1:]:
+ current_target_shape = compute_broadcast_shape(current_target_shape, next_shape)
Review Comment:

For conciseness, you could use `functools.reduce` to compute the target broadcast shape, which would replace this loop with a single line.
For example:
```python
import functools
...
current_target_shape = functools.reduce(compute_broadcast_shape, input_shapes)
```
This would require adding `import functools` at the top of the file.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For additional commands, e-mail: discuss-archive-help@tvm.apache.org
Re: [PR] [Relax][Onnx] Support Multi Input Ops with Multidirectional Broadcasting [tvm]
Posted by "locnd182644 (via GitHub)" <gi...@apache.org>.
locnd182644 commented on PR #18673:
URL: https://github.com/apache/tvm/pull/18673#issuecomment-3817522769
@tlopex Thank you !
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For additional commands, e-mail: discuss-archive-help@tvm.apache.org
Re: [PR] [Relax][Onnx] Support Multi Input Ops with Multidirectional Broadcasting [tvm]
Posted by "locnd182644 (via GitHub)" <gi...@apache.org>.
locnd182644 commented on PR #18673:
URL: https://github.com/apache/tvm/pull/18673#issuecomment-3782195967
@tvm-bot rerun
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: discuss-archive-unsubscribe@tvm.apache.org
For additional commands, e-mail: discuss-archive-help@tvm.apache.org