pytorch / pytorch/vision

Replace uses of `_is_tracing` with `torch.onnx.is_in_onnx_export()`

Open
#3,487 0 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement module: models module: onnx
Dominant language
Python
Stars
17.9k
Forks
7.3k
Avg merge
1d 15h
Merged PRs (30d)
13

Description

Through the codebase we have special implementations for ONNX guarded by torchvision._is_tracing(). We should see if it would be possible to replace it with torch.onnx.is_in_onnx_export() so that the meaning of those code-blocks is simpler, and so that we don't depend on internal PyTorch APIs.

I would start by checking if replacing https://github.com/pytorch/vision/blob/5b663c8eb691d540d064197a681a3a0fa924599c/torchvision/__init__.py#L92 with the aforementioned function passes ONNX tests, and then replace all remaining uses of it in the codebase.

cc @neginraoof

Contributor guide

Open the contributing guide

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 with the replacement at torchvision/init.py line 92 and run the ONNX tests to check whether the behavior remains correct. Then search the codebase for the remaining uses of torchvision._is_tracing and evaluate each ONNX-specific guard. Done means all applicable uses rely on torch.onnx.is_in_onnx_export() without breaking the ONNX tests.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Refactor
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.