Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,32 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- ur
|
| 4 |
+
- en
|
| 5 |
license: apache-2.0
|
| 6 |
+
datasets:
|
| 7 |
+
- iwslt14
|
| 8 |
+
metrics:
|
| 9 |
+
- bleu
|
| 10 |
+
library_name: fairseq
|
| 11 |
+
pipeline_tag: translation
|
| 12 |
---
|
| 13 |
+
|
| 14 |
+
### English to Urdu Translation
|
| 15 |
+
Urdu to English translation model is a Transformer model trained on IWSLT back-translated data using Faireq.
|
| 16 |
+
This model is produced during the experimentation related to building Context-Aware NMT models for low-resourced languages such as Urdu, Hindi, Sindhi, Pashtu and Punjabi. This particular model does not contains any contextual information and it is baseline sentence-level transformer model.
|
| 17 |
+
The evaluation is done on WMT2017 standard test set.
|
| 18 |
+
|
| 19 |
+
* source group: Urdu
|
| 20 |
+
* target group: English
|
| 21 |
+
|
| 22 |
+
* model: transformer
|
| 23 |
+
* Contextual
|
| 24 |
+
* Test Set: WMT2017
|
| 25 |
+
* pre-processing: Moses + Indic Tokenizer
|
| 26 |
+
* Dataset + Libray Details: [DLNMT](https://github.com/sami-haq99/nrpu-dlnmt)
|
| 27 |
+
|
| 28 |
+
## Benchmarks
|
| 29 |
+
|
| 30 |
+
| testset | BLEU |
|
| 31 |
+
|-----------------------|-------|
|
| 32 |
+
| Wmt2017 | 57.95 |
|