Update README.md
Browse files
README.md
CHANGED
@@ -1,5 +1,7 @@
|
|
1 |
-
|
2 |
|
|
|
|
|
3 |
from transformers import RobertaTokenizerFast, AutoModelForTokenClassification
|
4 |
from datasets import load_dataset
|
5 |
|
@@ -37,12 +39,22 @@ def tokenize_and_align_labels(examples):
|
|
37 |
tokenized_inputs["labels"] = labels
|
38 |
return tokenized_inputs
|
39 |
|
40 |
-
model = AutoModelForTokenClassification.from_pretrained('HeTree/HeConE')
|
41 |
tokenizer = RobertaTokenizerFast.from_pretrained('HeTree/HeConE')
|
42 |
raw_dataset = load_dataset('HeTree/MevakerConcSen')
|
43 |
window_size = 5
|
44 |
raw_dataset_window = raw_dataset.map(split_into_windows, batched=True, batch_size=window_size, remove_columns=raw_dataset['train'].column_names)
|
45 |
tokenized_dataset = raw_dataset_window.map(tokenize_and_align_labels, batched=False)
|
|
|
46 |
|
|
|
47 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
```
|
|
|
1 |
+
## Hebrew Conclusion Extraction Model (based on token classification)
|
2 |
|
3 |
+
#### How to use
|
4 |
+
```python
|
5 |
from transformers import RobertaTokenizerFast, AutoModelForTokenClassification
|
6 |
from datasets import load_dataset
|
7 |
|
|
|
39 |
tokenized_inputs["labels"] = labels
|
40 |
return tokenized_inputs
|
41 |
|
42 |
+
model = AutoModelForTokenClassification.from_pretrained('HeTree/HeConE')
|
43 |
tokenizer = RobertaTokenizerFast.from_pretrained('HeTree/HeConE')
|
44 |
raw_dataset = load_dataset('HeTree/MevakerConcSen')
|
45 |
window_size = 5
|
46 |
raw_dataset_window = raw_dataset.map(split_into_windows, batched=True, batch_size=window_size, remove_columns=raw_dataset['train'].column_names)
|
47 |
tokenized_dataset = raw_dataset_window.map(tokenize_and_align_labels, batched=False)
|
48 |
+
```
|
49 |
|
50 |
+
### Citing
|
51 |
|
52 |
+
If you use HeConE in your research, please cite [HeRo: RoBERTa and Longformer Hebrew Language Models](http://arxiv.org/abs/2304.11077).
|
53 |
+
```
|
54 |
+
@article{shalumov2023hero,
|
55 |
+
title={HeRo: RoBERTa and Longformer Hebrew Language Models},
|
56 |
+
author={Vitaly Shalumov and Harel Haskey},
|
57 |
+
year={2023},
|
58 |
+
journal={arXiv:2304.11077},
|
59 |
+
}
|
60 |
```
|