Commits · 5a9403db010f889b7ec9e72fe4097e746f3a7fa8 · nlp / ahisto-modules / Named Entity Experiments · GitLab

Feb 06, 2023
- Add command-line interface · 5a9403db
  Vít Starý Novotný authored Feb 06, 2023
  
  5a9403db
Jan 19, 2023
- Add parameter `displayed_labels` to `NerConfusionMatrix` · 24013b97
  Vít Starý Novotný authored Jan 19, 2023
  
  24013b97
- Repeat evaluation in `05_train_additional_ner_models_for_paper.ipynb` · f57437b0
  Vít Starý Novotný authored Jan 19, 2023
  
  f57437b0
- Add parameter `use_bio_tags` to `TaggedSentence.confusion_matrix()` · be0859f6
  Vít Starý Novotný authored Jan 19, 2023
  
  be0859f6
- Fix handling of adjacent entities in `TaggedSentence.possible_hypotheses` · 4aee4ab9
  Vít Starý Novotný authored Jan 19, 2023
  
  4aee4ab9
- Revert "Allow the use of `*-PER-or-LOC` tags in `NerModel.test()`" · 02cda378
  Vít Starý Novotný authored Jan 19, 2023
```
This reverts commit cc3e380e.
```
  02cda378
- Consider all possible hypotheses in `TaggedSentence.confusion_matrix()` · 9364f902
  Vít Starý Novotný authored Jan 19, 2023
  
  9364f902
- Make `TaggedSentence.evaluate()` independently maximize all results · 05852013
  Vít Starý Novotný authored Jan 19, 2023
  
  05852013
- Merge branch 'fix/per-or-loc' into 'main' · 8654313b
  Vít Starý Novotný authored Jan 19, 2023
```
Use `PER-or-LOC` labels during evaluation

Closes #1

See merge request nlp/ahisto-modules/named-entity-search!1
```
  8654313b
- Consider all possible hypotheses in `TaggedSentence.evaluate()` · ffcb6828
  Vít Starý Novotný authored Jan 19, 2023
  
  ffcb6828
- Consider trailing named entities in `NerEvaluationResult.from_tagged_sentences()` · 99ec4f5d
  Vít Starý Novotný authored Jan 19, 2023
  
  99ec4f5d
- Allow the use of `*-PER-or-LOC` tags in `NerModel.test()` · cc3e380e
  Vít Starý Novotný authored Jan 19, 2023
  
  cc3e380e
Jan 18, 2023
- Evaluate additional NER models for paper · 8b3c30ca
  Vít Starý Novotný authored Jan 18, 2023
  
  8b3c30ca
Jan 17, 2023
- Train additional NER models for paper · 597fd669
  Vít Starý Novotný authored Jan 17, 2023
  
  597fd669
Jan 15, 2023
- Only load JSON document once in `Document.relevant_pages` · 44862646
  Vít Starý Novotný authored Jan 15, 2023
  
  44862646
- List the number of (confirmed) canonical persons and locations · 6222c3c0
  Vít Starý Novotný authored Jan 15, 2023
  
  6222c3c0
Jan 12, 2023
- Fix incorrect reference · 14da619f
  Vít Starý Novotný authored Jan 12, 2023
  
  14da619f
- Reevaluate best models using micro-strict and micro-fuzzy measures · 0652ceb9
  Vít Starý Novotný authored Jan 12, 2023
```
See Ehrmann et al. (2020, Section 5.2)
Also available at WWW: <https://aclanthology.org/2020.lrec-1.121/>
```
  0652ceb9
- Finish evaluation of sixth-generation of models · f58bfc3c
  Vít Starý Novotný authored Jan 12, 2023
  
  f58bfc3c
- Add `search.evaluation.NEREvaluationResult` · ce15d30d
  Vít Starý Novotný authored Jan 11, 2023
  
  ce15d30d
Jan 05, 2023
- Use seventh-generation models to tag MLM datasets · 4b44e99d
  Vít Starý Novotný authored Jan 05, 2023
  
  4b44e99d
Jan 04, 2023
- Update to Python 3.8 · 3698733e
  Vít Starý Novotný authored Jan 04, 2023
  
  3698733e
- Extend quantitative evaluation with confusion matrices · ceb57116
  Vít Starý Novotný authored Jan 04, 2023
  
  ceb57116
- Extend qualitative evaluation to Latin and German example sentences · 3dc65ece
  Vít Starý Novotný authored Jan 04, 2023
  
  3dc65ece
Jan 03, 2023
- Evaluate seventh-generation models · 720fce41
  Vít Starý Novotný authored Jan 03, 2023
  
  720fce41
Dec 30, 2022
- Make checkpoint directory name unique to a Git commit · c7568c43
  Vít Starý Novotný authored Dec 30, 2022
  
  c7568c43
- Increase `warmup_ratio` to 0.1 · e87e66ec
  Vít Starý Novotný authored Dec 30, 2022
  
  e87e66ec
- Add `warmup_ratio` to `config.ini` · 90fee260
  Vít Starý Novotný authored Dec 30, 2022
  
  90fee260
- Set `max_steps` and `number_of_training_epochs` to approximately 200 epochs · 661c6101
  Vít Starý Novotný authored Dec 30, 2022
  
  661c6101
- Work around https://github.com/gaussalgo/adaptor/issues/15 · 3b533d46
  Vít Starý Novotný authored Dec 30, 2022
  
  3b533d46
Dec 29, 2022
- Train seventh-generation models · 037f8531
  Vít Starý Novotný authored Dec 07, 2022
  
  037f8531
Dec 28, 2022
- Evaluate sixth-generation models · d1b65369
  Vít Starý Novotný authored Dec 28, 2022
  
  d1b65369
Dec 27, 2022
- Reduce the number of saved checkpoints · 12ba560d
  Vít Starý Novotný authored Dec 27, 2022
  
  12ba560d
Dec 18, 2022
- Pin `transformers` package · 18c590c4
  Vít Starý Novotný authored Dec 18, 2022
  
  18c590c4
- Update `04_extract_dataset_statistics.ipynb` · 3297b04e
  Vít Starý Novotný authored Dec 18, 2022
  
  3297b04e
- Train sixth-generation models · 58ee3ea0
  Vít Starý Novotný authored Dec 18, 2022
  
  58ee3ea0
- Evaluate fifth-generation models · 8c95e445
  Vít Starý Novotný authored Dec 18, 2022
  
  8c95e445
Dec 16, 2022
- Increase delay between model training · 76e2de6e
  Vít Starý Novotný authored Dec 16, 2022
  
  76e2de6e
- Revert "Load best model at the end of training" · f7c4d3f6
  Vít Starý Novotný authored Dec 16, 2022
```
This reverts commit 44a6597e.

I received the following error when training:
https://discuss.huggingface.co/t/distributed-training-keyerror-eval-f1-in-questionansweringtrainer-taken-from-trainer-qa-py-in-examples/26308
```
  f7c4d3f6
- Train fifth-generation models · a944e451
  Vít Starý Novotný authored Dec 16, 2022
  
  a944e451