|
--- |
|
language: id |
|
tags: |
|
- tiny-roberta-indonesia |
|
license: mit |
|
datasets: |
|
- wikipedia |
|
widget: |
|
- text: "ikiryo adalah <mask> hantu dalam mitologi jepang." |
|
--- |
|
|
|
# Indonesian tiny-RoBERTa |
|
## How to Use |
|
|
|
### As Masked Language Model |
|
```python |
|
from transformers import pipeline |
|
|
|
pretrained_name = "akahana/tiny-roberta-indonesia" |
|
|
|
fill_mask = pipeline( |
|
"fill-mask", |
|
model=pretrained_name, |
|
tokenizer=pretrained_name |
|
) |
|
|
|
fill_mask("ikiryo adalah <mask> hantu dalam mitologi jepang.") |
|
``` |
|
|
|
### Feature Extraction in PyTorch |
|
```python |
|
from transformers import RobertaModel, RobertaTokenizerFast |
|
|
|
pretrained_name = "akahana/tiny-roberta-indonesia" |
|
model = RobertaModel.from_pretrained(pretrained_name) |
|
tokenizer = RobertaTokenizerFast.from_pretrained(pretrained_name) |
|
|
|
prompt = "ikiryo adalah <mask> hantu dalam mitologi jepang." |
|
encoded_input = tokenizer(prompt, return_tensors='pt') |
|
output = model(**encoded_input) |
|
``` |
