Update the card: consistent title, project and phase, metric, related resources
Browse files
README.md
CHANGED
|
@@ -25,4 +25,24 @@ configs:
|
|
| 25 |
data_files:
|
| 26 |
- split: test
|
| 27 |
path: pilener/test-*
|
|
|
|
|
|
|
| 28 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
data_files:
|
| 26 |
- split: test
|
| 27 |
path: pilener/test-*
|
| 28 |
+
pretty_name: UniversalMedNER – Pile-NER In-Domain Evaluation Subset (First Phase)
|
| 29 |
+
language: [en]
|
| 30 |
---
|
| 31 |
+
|
| 32 |
+
# UniversalMedNER — Pile-NER In-Domain Evaluation Subset (First Phase)
|
| 33 |
+
|
| 34 |
+
Part of **UniversalMedNER**, the code, data and models of the bachelor thesis *Structuring Universal Chain-of-thought for Named Entity Recognition* (Department of Computer Science and Engineering, University of Bologna, 2026).
|
| 35 |
+
|
| 36 |
+
500 Pile-NER passages used as the **in-domain** test set of the first phase, in exactly the format of
|
| 37 |
+
`universalmedner/bioner-eval-subset`, so that the same evaluation code scores them.
|
| 38 |
+
|
| 39 |
+
- **No overlap with training**: passages are sampled only among those that were never candidates for fine-tuning
|
| 40 |
+
(the whole jury-filtered set is excluded, not only the examples actually used).
|
| 41 |
+
- **Gold**: the original Pile-NER annotation, i.e. the annotation style the first-phase students were trained on;
|
| 42 |
+
the set therefore measures what fine-tuning taught in its own domain.
|
| 43 |
+
- **Requested types**: up to six, a random number of the types present in the passage plus absent types sampled with
|
| 44 |
+
inverse frequency.
|
| 45 |
+
- **Fields**: `id`, `raw_text`, `all_gold`, `messages` (as in `universalmedner/bioner-eval-subset`).
|
| 46 |
+
|
| 47 |
+
Built with `04_evaluation/build_pilener_eval.py`. The second phase uses the test split of
|
| 48 |
+
`disi-unibo-nlp-students/pilener-teacher-cot-agreement-v1` instead.
|