Commit
•
d2d3cd6
1
Parent(s):
c715352
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
- vi
|
| 5 |
+
tags:
|
| 6 |
+
- translation
|
| 7 |
+
license: apache-2.0
|
| 8 |
+
datasets:
|
| 9 |
+
- ALT
|
| 10 |
+
metrics:
|
| 11 |
+
- sacrebleu
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
This is a finetuning of a MarianMT pretrained on English-Chinese. The target language pair is English-Vietnamese.
|
| 15 |
+
The first phase of training (mixed) is performed on a dataset containing both English-Chinese and English-Vietnamese sentences.
|
| 16 |
+
The second phase of training (pure) is performed on a dataset containing only English-Vietnamese sentences.
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
### Training results
|
| 20 |
+
|
| 21 |
+
MIXED
|
| 22 |
+
| Epoch | Bleu |
|
| 23 |
+
|:-----:|:-------:|
|
| 24 |
+
| 1.0 | 26.2407 |
|
| 25 |
+
| 2.0 | 32.6016 |
|
| 26 |
+
| 3.0 | 35.4060 |
|
| 27 |
+
| 4.0 | 36.6737 |
|
| 28 |
+
| 5.0 | 37.3774 |
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
PURE
|
| 32 |
+
| Epoch | Bleu |
|
| 33 |
+
|:-----:|:-------:|
|
| 34 |
+
| 1.0 | 37.3169 |
|
| 35 |
+
| 2.0 | 37.4407 |
|
| 36 |
+
| 3.0 | 37.6696 |
|
| 37 |
+
| 4.0 | 37.8765 |
|
| 38 |
+
| 5.0 | 38.0105 |
|
| 39 |
+
|