onnx / onnx/models

deprecated attributes in the models from the zoo

Open
#167 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Jupyter Notebook
Stars
9.8k
Forks
1.6k
PR merge metrics
No merged PRs in 30d

Description

Hello,

TL;DR: is there a plan to remove the deprecated attributes? Shouldn't we ensure that all new uploaded models do not use any deprecated attributes?

Longer: In the IR, the auto_padding is deprecated in most of the operators.
The problem is that a lot of backends is still using it, and some of the models from the zoo are using this attribute for maxpool or convolution.

We may have some backends that don't implement this behavior and yet be compatible with ONNX.
As an example, I am implementing an ONNX decoder in Go and I am using the Gorgonia library as a backend.
The problem I encounter is that I have to implement those attributes if I want to actually execute those models, and from times to time it is not trivial.

  • Shouldn't we ensure that all new uploaded models do not use any deprecated attributes?
  • What is the plan to manage the lifecycle of the models in the zoo?

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reviewing the model zoo's upload process and the models that use the deprecated auto_padding attribute, especially for maxpool and convolution. Determine whether the repository has a way to validate new uploads and how existing models are maintained. Done would require an agreed lifecycle or validation plan, but the issue does not identify files or tests.

Written by the indexing model from the issue text.

Assessment

Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.