Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

task: cross-text-coref #110

Merged
merged 49 commits into from
Sep 16, 2024
Merged

task: cross-text-coref #110

merged 49 commits into from
Sep 16, 2024

Conversation

ArneBinder
Copy link
Owner

@ArneBinder ArneBinder commented Sep 11, 2024 •
edited
Loading

TODO:

  • tests for model
  • implement & test unbatch_output
  • implement & test create_annotations_from_output
  • long input handling (implement max_window parameter)
  • add documentation for taskmodule and model
  • test downstream: on a dataset created from semantically_same relations in SciArg

follow-up: use reasonable model loss

@codecov-commenter
Copy link

codecov-commenter commented Sep 11, 2024 •
edited
Loading

Codecov Report

Attention: Patch coverage is 94.08284% with 20 lines in your changes missing coverage. Please review.

Project coverage is 95.71%. Comparing base (2ddb8a5) to head (904fac9).

Files with missing lines Patch % Lines
src/pie_modules/document/processing/text_pair.py 90.29% 13 Missing ⚠️
...pie_modules/taskmodules/cross_text_binary_coref.py 95.00% 6 Missing ⚠️
src/pie_modules/utils/tokenization.py 96.00% 1 Missing ⚠️
Additional details and impacted files 
@@            Coverage Diff             @@
##             main     #110      +/-   ##
==========================================
- Coverage   95.83%   95.71%   -0.13%     
==========================================
  Files          57       60       +3     
  Lines        4656     4992     +336     
==========================================
+ Hits         4462     4778     +316     
- Misses        194      214      +20

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

@ArneBinder ArneBinder force-pushed the taskmodules/cross_text_binary_coref branch 2 times, most recently from cb04c11 to 508bac3 Compare September 12, 2024 16:17
@ArneBinder ArneBinder changed the title task: cross-text-binary-coref [WIP] task: cross-text-binary-coref Sep 12, 2024
@ArneBinder
implement CrossTextBinaryCorefTaskModule
3923dbe
@ArneBinder
call save_hyperparameters()
52a8b45
@ArneBinder
make taskmodule (future) model compliant
71fb6e1
@ArneBinder
implement SimpleSimilarityModel
abec896
@ArneBinder
use fixture data for documents_with_negatives
16caf81
@ArneBinder
disentangle tests
b2f739b
@ArneBinder
streamline test
119ef75
@ArneBinder
improve test
6b90f99
@ArneBinder
create negatives from text to itself (but different spans)
5067e16
@ArneBinder
restrict candidates by having same entity type
a93abde
@ArneBinder
make RelationStatisticsMixin ready for multi-label or binary
ddadf5e
@ArneBinder
use RelationStatisticsMixin
4768fe0
@ArneBinder
rename model to SpanSimilarityModel; add similarity_threshold paramet…
5d0c7fa 
…er; set num_indices when "mention_pooling" is used
@ArneBinder
remove SpanSimilarityModel in favor of new SequencePairSimilarityMode…
cc2e828 
…lWithPooler
@ArneBinder
add tests for SequencePairSimilarityModelWithPooler
64fa61e
@ArneBinder
use mention pooling per default
41af7ce
@ArneBinder
make pre-commit happy
ad89617
@ArneBinder
implement unbatch_output() and create_annotations_from_output()
2a974ea
@ArneBinder
implement long text handling
6a2e878
@ArneBinder
set default label_threshold for SequencePairSimilarityModelWithPooler…
617be19 
… to 0.9
@ArneBinder
fix missed index shift because of added special tokens
309346a
@ArneBinder
minor fixes
8681f24
@ArneBinder
add check for direction of coref relations (should point from text to…
4376d16 
… text_pair)
@ArneBinder
add documentation for SequencePairSimilarityModelWithPooler
ebd09b2
@ArneBinder
add model and taskmodule to readme
98cc08f
@ArneBinder
add short documenation for CrossTextBinaryCorefTaskModule
9599ee2
@ArneBinder
outsource add_negative_relations() to document.precessing.text_pair
0cb1708
@ArneBinder
update documents_with_negatives.json with current output
6b0c032
@ArneBinder
rename add_negative_relations() to add_negative_coref_relations() and…
1901779 
… remove from CrossTextBinaryCorefTaskModule
@ArneBinder
implement construct_text_pair_coref_documents_from_partitions_via_rel…
8093a58 
…ations and add dedicated test_text_pair.py
@ArneBinder
add tqdm to add_negative_coref_relations
c0c2875
@ArneBinder
move document and annotation types to documents and annotations modul…
81ea67c 
…es, respectively
@ArneBinder
fix tokenization in encode_input
b1ceac2
@ArneBinder
outsource get_aligned_token_span() and SpanNotAlignedWithTokenExcepti…
b5f0c8b 
…on to utils.tokenization module
@ArneBinder
implement construct_text_document_from_text_pair_coref_document()
7487fd7
@ArneBinder
add parameter relation_label_mapping to construct_text_document_from_…
961a2a0 
…text_pair_coref_document()
@ArneBinder
add parameter no_relation_label to construct_text_document_from_text_…
1e3b5d4 
…pair_coref_document()
@ArneBinder ArneBinder force-pushed the taskmodules/cross_text_binary_coref branch from d210bfd to 1e3b5d4 Compare September 15, 2024 18:25
@ArneBinder ArneBinder changed the title task: cross-text-binary-coref task: cross-text-coref Sep 16, 2024
@ArneBinder ArneBinder merged commit cccdd33 into main Sep 16, 2024
2 checks passed
@ArneBinder ArneBinder deleted the taskmodules/cross_text_binary_coref branch September 16, 2024 16:52
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
enhancement New feature or request new model new taskmodule
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@ArneBinder @codecov-commenter