Comments (6)
I think GridSample has been updated to support ND inputs (https://onnx.ai/onnx/operators/onnx__GridSample.html). To help me understand better, was your intention to support a range of input ranks during shape validation? What would this enable?
from onnx.
I think GridSample has been updated to support ND inputs (https://onnx.ai/onnx/operators/onnx__GridSample.html). To help me understand better, was your intention to support a range of input ranks during shape validation? What would this enable?
Yes your understanding is correct.
If you look here, the logic is that it'll throw an error if you try to pass it input with anything but 4 dimensions. ORT is interested into extending this feature to support 5D, so (if my understanding is correct), this will need to change a bit to account for both 4D and 5D being valid inputs. If an operator is interested in supporting ND, would it not be helpful to have this validation?
Keep in mind this is a really small detail and I figured it was worth making a PR. I am still learning both ORT and ONNX - this isn't required functionality but I was mostly curious to see if it offered any utility. If I'm missing some context, please lmk and I can close both this PR and issue.
from onnx.
Thanks for the clarification. The new grid sample is going to support dimensions >5 as well, at least according to the spec. So specifying a range seems to be runtime specific. I would recommend adding this function only when it is needed for inferring shapes for onnx domain operations defined in this repository, to keep the APIs lean. Please correct me if there's anything I overlooked.
from onnx.
Thanks for the clarification. The new grid sample is going to support dimensions >5 as well, at least according to the spec. So specifying a range seems to be runtime specific. I would recommend adding this function only when it is needed for inferring shapes for onnx domain operations defined in this repository, to keep the APIs lean. Please correct me if there's anything I overlooked.
Sounds good, I don't think you've missed anything. Would you prefer this PR closed then?
(Tangential) can you point to some cxx related issues I can contribute towards if any exist? :)
from onnx.
Sounds good, I don't think you've missed anything. Would you prefer this PR closed then?
Yes. I think it's good to close it for now. Your contribution is appreciated still :)
(Tangential) can you point to some cxx related issues I can contribute towards if any exist? :)
Absolutely! Here is a list of issues that would welcome contributions: https://github.com/onnx/onnx/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-desc+label%3A%22contributions+welcome%22
Some c++ related ones are:
Etc.
from onnx.
Sounds good, I don't think you've missed anything. Would you prefer this PR closed then?
Yes. I think it's good to close it for now. Your contribution is appreciated still :)
(Tangential) can you point to some cxx related issues I can contribute towards if any exist? :)
Absolutely! Here is a list of issues that would welcome contributions: https://github.com/onnx/onnx/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-desc+label%3A%22contributions+welcome%22
Some c++ related ones are:
- Convert Split from opset11 to 18 failed. #5222
- onnx.checker.check_model() does not report mismatch input/output shapes #5974
- Create down conversion for GroupNormalization-21 #5843
- Inliner fails on the attached model #5816
Etc.
Much appreciated! I'll look into one of those :D
from onnx.
Related Issues (20)
- Equivalent Implementations to ScatterElements operator HOT 3
- onnx.load() -> Can't Open this page HOT 1
- Square Activation is not supported by onnxruntime HOT 1
- Inference Layer by Layer or feature extraction HOT 2
- Support for collective operations / MPI primitives HOT 4
- When is the next release, 1.16.0? HOT 3
- RUNTIME_EXCEPTION : Non-zero status code returned while running Reshape node. HOT 6
- INT8 quantization model from torch to onnx slower than FP32 model on cpu HOT 1
- When exporting from torch, the result of the input varies depending on the dummy input. HOT 2
- [Feature request] Use protobuf's generator to generate pyi stubs HOT 1
- Does inline_local_functions preserve ir_version? HOT 3
- 1.16.0 built from source on RHEL8 fails with: undefined symbol: _ZNSt10filesystem7__cxx114path14_M_split_cmptsEv
- Include IRv10 information in the proto spec file
- Clarify default value of `axis` for `DequantizeLinear` when input rank is 1 HOT 1
- [Spec] DepthToSpace `mode` attribute is counter-intuitive HOT 1
- [Feature request] Bump Conv to accept bfloat16
- Migrate CI pipelines from AzureDevOps to Github Actions
- Pause all PR merges HOT 1
- Result accuracy is different with PyTorch HOT 2
- Is TfIdfVectorizer safe to use? HOT 6
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from onnx.