Migrate model card from transformers-repo
Browse filesRead announcement at https://discuss.huggingface.co/t/announcement-all-model-cards-will-be-migrated-to-hf-co-model-repos/2755
Original file history: https://github.com/huggingface/transformers/commits/master/model_cards/keshan/SinhalaBERTo/README.md
README.md
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language: si
|
3 |
+
tags:
|
4 |
+
- SinhalaBERTo
|
5 |
+
- Sinhala
|
6 |
+
- roberta
|
7 |
+
datasets:
|
8 |
+
- oscar
|
9 |
+
---
|
10 |
+
### Overview
|
11 |
+
|
12 |
+
This is a slightly smaller model trained on [OSCAR](https://oscar-corpus.com/) Sinhala dedup dataset. As Sinhala is one of those low resource languages, there are only a handful of models been trained. So, this would be a great place to start training for more downstream tasks.
|
13 |
+
|
14 |
+
## Model Specification
|
15 |
+
|
16 |
+
|
17 |
+
The model chosen for training is [Roberta](https://arxiv.org/abs/1907.11692) with the following specifications:
|
18 |
+
1. vocab_size=52000
|
19 |
+
2. max_position_embeddings=514
|
20 |
+
3. num_attention_heads=12
|
21 |
+
4. num_hidden_layers=6
|
22 |
+
5. type_vocab_size=1
|
23 |
+
|
24 |
+
## How to Use
|
25 |
+
You can use this model directly with a pipeline for masked language modeling:
|
26 |
+
|
27 |
+
```py
|
28 |
+
from transformers import AutoTokenizer, AutoModelWithLMHead, pipeline
|
29 |
+
|
30 |
+
model = BertForMaskedLM.from_pretrained("keshan/SinhalaBERTo")
|
31 |
+
tokenizer = BertTokenizer.from_pretrained("keshan/SinhalaBERTo")
|
32 |
+
|
33 |
+
fill_mask = pipeline('fill-mask', model=model, tokenizer=tokenizer)
|
34 |
+
|
35 |
+
fill_mask("මම ගෙදර <mask>.")
|
36 |
+
|
37 |
+
```
|