Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,53 @@
|
|
1 |
---
|
|
|
|
|
|
|
|
|
|
|
2 |
license: cc-by-nc-4.0
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
+
|
3 |
+
language:
|
4 |
+
- hu
|
5 |
+
tags:
|
6 |
+
- fill-mask
|
7 |
license: cc-by-nc-4.0
|
8 |
+
widget:
|
9 |
+
- text: "Elmesélek egy történetet a nyelvtechnológiáról."
|
10 |
---
|
11 |
+
|
12 |
+
# PULI BERT-Large
|
13 |
+
|
14 |
+
For further details, see [our demo site](https://juniper.nytud.hu/demo/nlp).
|
15 |
+
|
16 |
+
- Hungarian BERT large model
|
17 |
+
- Trained with Megatron-DeepSpeed [github](https://github.com/microsoft/Megatron-DeepSpeed)
|
18 |
+
- Dataset: 36.3 billion words
|
19 |
+
- Checkpoint: 150 000 steps
|
20 |
+
|
21 |
+
## Limitations
|
22 |
+
|
23 |
+
- max_seq_length = 1024
|
24 |
+
|
25 |
+
|
26 |
+
## Citation
|
27 |
+
If you use this model, please cite the following paper:
|
28 |
+
|
29 |
+
```
|
30 |
+
@inproceedings {yang-gpt3,
|
31 |
+
title = {Jönnek a nagyok! GPT-3, GPT-2 és BERT large nyelvmodellek magyar nyelvre},
|
32 |
+
booktitle = {XIX. Magyar Számítógépes Nyelvészeti Konferencia (MSZNY 2023)},
|
33 |
+
year = {2023},
|
34 |
+
publisher = {Szegedi Tudományegyetem},
|
35 |
+
address = {Szeged, Hungary},
|
36 |
+
author = {Yang, Zijian Győző and Dodé, Réka and Ferenczi, Gergő and Héja, Enikő and Kőrös, Ádám and Laki, László János and Ligeti-Nagy, Noémi and Jelencsik-Mátyus, Kinga and Vadász, Noémi and Váradi, Tamás},
|
37 |
+
pages = {0}
|
38 |
+
}
|
39 |
+
|
40 |
+
```
|
41 |
+
|
42 |
+
## Usage
|
43 |
+
|
44 |
+
```python
|
45 |
+
from transformers import BertTokenizer, MegatronBertModel
|
46 |
+
|
47 |
+
tokenizer = BertTokenizer.from_pretrained('NYTK/PULI-BERT-large')
|
48 |
+
model = MegatronBertModel.from_pretrained('NYTK/PULI-BERT-large')
|
49 |
+
text = "Replace me by any text you'd like."
|
50 |
+
encoded_input = tokenizer(text, return_tensors='pt')
|
51 |
+
output = model(**encoded_input)
|
52 |
+
|
53 |
+
```
|