patrick-kidger / patrick-kidger/jaxtyping
Support for NestedTensors
Open
Nobody has claimed this yet.
feature
question
- Dominant language
- Python
- Stars
- 1.9k
- Forks
- 96
- PR merge metrics
- No merged PRs in 30d
Description
Currently NestedTensors give errors (examples - https://pytorch.org/tutorials/prototype/nestedtensor.html and https://pytorch.org/docs/stable/nested.html)
File <@beartype(models.transformer.Attention.forward) at 0x7f7ac394d430>:23, in forward(__beartype_object_221882800, __beartype_get_violation, __beartype_conf, __beartype_object_140169208358272, __beartype_object_140169211732352, __beartype_func, *args, **kwargs)
...
RuntimeError: Internal error: NestedTensorImpl doesn't support sizes. Please file an issue.
Just an idea, could be helpful for more folks! :) this library has already saved me from many bugs btw 🙏
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the failure with the linked NestedTensor examples and the reported models.transformer.Attention.forward traceback. Trace how runtime checking handles the NestedTensor input, then define regression coverage showing that the documented examples no longer raise the reported NestedTensorImpl sizes error.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100