Update README.md
Browse files
README.md
CHANGED
@@ -54,6 +54,25 @@ please refer to our paper and the annotation guidelines for the
|
|
54 |
|
55 |
This is an HF friendly dataset loader for training LLMs on the NER tags.
|
56 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
57 |
|
58 |
### Acknowledgements
|
59 |
|
|
|
54 |
|
55 |
This is an HF friendly dataset loader for training LLMs on the NER tags.
|
56 |
|
57 |
+
### Usage Example
|
58 |
+
|
59 |
+
```
|
60 |
+
from datasets import load_dataset
|
61 |
+
|
62 |
+
annodb = load_dataset("priamai/AnnoCTR","all_tags",trust_remote_code=True)
|
63 |
+
|
64 |
+
label_list = annodb["train"].features[f"all_tags"].feature.names
|
65 |
+
|
66 |
+
|
67 |
+
```
|
68 |
+
The letter that prefixes each `tag` indicates the token position of the entity:
|
69 |
+
|
70 |
+
- `B-` indicates the beginning of an entity.
|
71 |
+
- `I-` indicates a token is contained inside the same entity (for example, the `State` token is a part of an entity like
|
72 |
+
`Empire State Building`).
|
73 |
+
- `0` indicates the token doesn't correspond to any entity.
|
74 |
+
|
75 |
+
There are 23 tags currently.
|
76 |
|
77 |
### Acknowledgements
|
78 |
|