Commit
•
a5582fc
1
Parent(s):
1f9be24
Update README.md
Browse files
README.md
CHANGED
@@ -39,25 +39,31 @@ should probably proofread and complete it, then remove this comment. -->
|
|
39 |
|
40 |
# distilroberta-base-ner-wikiann-conll2003-4-class
|
41 |
|
42 |
-
This model is a fine-tuned version of [distilroberta-base](https://huggingface.co/distilroberta-base) on the wikiann
|
43 |
-
|
44 |
-
-
|
45 |
-
|
46 |
-
-
|
47 |
-
|
48 |
-
|
|
|
|
|
49 |
|
50 |
-
## Model
|
51 |
|
52 |
-
|
|
|
|
|
53 |
|
54 |
-
|
|
|
55 |
|
56 |
-
|
|
|
57 |
|
58 |
-
|
59 |
|
60 |
-
|
61 |
|
62 |
## Training procedure
|
63 |
|
@@ -75,6 +81,20 @@ The following hyperparameters were used during training:
|
|
75 |
|
76 |
### Training results
|
77 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
78 |
|
79 |
|
80 |
### Framework versions
|
|
|
39 |
|
40 |
# distilroberta-base-ner-wikiann-conll2003-4-class
|
41 |
|
42 |
+
This model is a fine-tuned version of [distilroberta-base](https://huggingface.co/distilroberta-base) on the wikiann and conll2003 dataset. It consists out of the classes of conll2003.
|
43 |
+
|
44 |
+
O (0), B-PER (1), I-PER (2), B-ORG (3), I-ORG (4) B-LOC (5), I-LOC (6) B-MISC (7), I-MISC (8).
|
45 |
+
|
46 |
+
eval F1-Score: **95,39** (merged dataset)
|
47 |
+
test F1-Score: **90,75** (merged dataset)
|
48 |
+
|
49 |
+
eval F1-Score: **95,39** (CoNLL-03)
|
50 |
+
test F1-Score: **90,75** (CoNLL-03)
|
51 |
|
52 |
+
## Model Usage
|
53 |
|
54 |
+
```python
|
55 |
+
from transformers import AutoTokenizer, AutoModelForTokenClassification
|
56 |
+
from transformers import pipeline
|
57 |
|
58 |
+
tokenizer = AutoTokenizer.from_pretrained("philschmid/distilroberta-base-ner-wikiann-conll2003-4-class")
|
59 |
+
model = AutoModelForTokenClassification.from_pretrained("philschmid/distilroberta-base-ner-wikiann-conll2003-4-class")
|
60 |
|
61 |
+
nlp = pipeline("ner", model=model, tokenizer=tokenizer, grouped_entities=True)
|
62 |
+
example = "My name is Philipp and live in Germany"
|
63 |
|
64 |
+
nlp(example)
|
65 |
|
66 |
+
```
|
67 |
|
68 |
## Training procedure
|
69 |
|
|
|
81 |
|
82 |
### Training results
|
83 |
|
84 |
+
It achieves the following results on the evaluation set:
|
85 |
+
- Loss: 0.0705
|
86 |
+
- Precision: 0.9492
|
87 |
+
- Recall: 0.9585
|
88 |
+
- F1: 0.9539
|
89 |
+
- Accuracy: 0.9882
|
90 |
+
|
91 |
+
It achieves the following results on the test set:
|
92 |
+
- Loss: 0.239
|
93 |
+
- Precision: 0.8984
|
94 |
+
- Recall: 0.9168
|
95 |
+
- F1: 0.9075
|
96 |
+
- Accuracy: 0.9741
|
97 |
+
|
98 |
|
99 |
|
100 |
### Framework versions
|