Skip to content

Conversation

@ArneBinder
Copy link
Owner

@ArneBinder ArneBinder commented Sep 24, 2025

This PR registers all models with @pie_core.Model instead of @PyTorchIEModel to make them work with pie_core.AutoAnnotationPipeline.from_config in the future.

Additional changes:

  • remove AutoPyTorchIEModel (just use AutoModel instead)
    • re-export AutoModel from pie_core. NOTE: Previously AutoPyTorchIEModel was exported as AutoModel! But if used correctly, it should not make a difference.
  • register PyTorchIEPipeline at name "pytorch-ie"
  • remove backwards compatibility: Pipeline and AutoPipeline(use PyTorchIEPipeline in either case)
  • adjust the README.md

Note: It is not yet possible to load an "old" pipeline (e.g. pie/example-ner-spanclf-conll03) via AutoAnnotationPipeline since its config does not has a pipeline_type key. In the future, it should be possible to provide that as argument in the following manner, but this still needs a minor adjustment in pie_core.Auto.from_config (see ArneBinder/pie-core#95):

pipeline = AutoAnnotationPipeline.from_pretrained("pie/example-ner-spanclf-conll03", pipeline_type="pytorch-ie")

@ArneBinder ArneBinder added the breaking Breaking Changes label Sep 24, 2025
@codecov
Copy link

codecov bot commented Sep 24, 2025

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.68%. Comparing base (a6bb91d) to head (c52d28d).
⚠️ Report is 4 commits behind head on main.

Additional details and impacted files
@@            Coverage Diff             @@
##             main     #503      +/-   ##
==========================================
+ Coverage   76.98%   82.68%   +5.69%     
==========================================
  Files          31       64      +33     
  Lines        1803     5358    +3555     
  Branches      350     1129     +779     
==========================================
+ Hits         1388     4430    +3042     
- Misses        337      675     +338     
- Partials       78      253     +175     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

@ArneBinder ArneBinder linked an issue Sep 24, 2025 that may be closed by this pull request
@ArneBinder ArneBinder self-assigned this Sep 24, 2025
…ter PyTorchIEPipeline at name "pytorch-ie"; remove backwards compatibility (Pipeline, AutoPipeline) and re-export AutoModel from pie_core instead of AutoPyTorchIEModel
@ArneBinder ArneBinder force-pushed the annotate-models-with-model branch from 3feb3b3 to c52d28d Compare September 24, 2025 15:14
@ArneBinder ArneBinder merged commit 3f6fab0 into main Sep 24, 2025
4 checks passed
@ArneBinder ArneBinder deleted the annotate-models-with-model branch September 24, 2025 15:24
ArneBinder added a commit to ArneBinder/pytorch-ie-hydra-template-1 that referenced this pull request Sep 24, 2025
ArneBinder added a commit to ArneBinder/pytorch-ie-hydra-template-1 that referenced this pull request Sep 24, 2025
@ArneBinder ArneBinder changed the title register all models @pie_core.Model register all models with @pie_core.Model Sep 24, 2025
ArneBinder added a commit to ArneBinder/pytorch-ie-hydra-template-1 that referenced this pull request Sep 24, 2025
* remove pie-modules in code (in favor of pie-documents, but just implicitly)

* remove pie-modules in configs (in favor of pie-documents)

* switch to python 3.10 in github CI workflows

* use pytorch-ie from main branch

* update local conll2003 dataset script

* make pre-commit happy

* update revisions in dataset configs

* use latest version of pytorch-ie (including ArneBinder/pytorch-ie#503); several fixes

* fix sequence_classification_with_pooler model config

* update pie-core to 0.3.0

* update deprecated imports and class names

* use pytorch-ie release 0.33.0
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

breaking Breaking Changes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

make AutoAnnotationPipeline work

2 participants