File size: 1,034 Bytes
e1c0f32 b0ea199 e1c0f32 9144e1b e1c0f32 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 |
---
language:
- en
- vi
tags:
- translation
license: apache-2.0
datasets:
- ALT
metrics:
- sacrebleu
---
This is a finetuning of a MarianMT pretrained on Chinese-English. The target language pair is Vietnamese-English.
### Example
```
%%capture
!pip install transformers transformers[sentencepiece]
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
# Download the pretrained model for English-Vietnamese available on the hub
model = AutoModelForSeq2SeqLM.from_pretrained("CLAck/vi-en")
tokenizer = AutoTokenizer.from_pretrained("CLAck/vi-en")
sentence = your_vietnamese_sentence
# This token is needed to identify the source language
input_sentence = "<2vi> " + sentence
translated = model.generate(**tokenizer(input_sentence, return_tensors="pt", padding=True))
output_sentence = [tokenizer.decode(t, skip_special_tokens=True) for t in translated]
```
### Training results
| Epoch | Bleu |
|:-----:|:-------:|
| 1.0 | 21.3180 |
| 2.0 | 26.8012 |
| 3.0 | 29.3578 |
| 4.0 | 31.5178 |
| 5.0 | 32.8740 |
|