Update README.md
Browse files
README.md
CHANGED
@@ -21,17 +21,26 @@ It achieves the following results on the evaluation set:
|
|
21 |
- Loss: 1.1731
|
22 |
- Bleu: 42.9597
|
23 |
|
24 |
-
## Model
|
25 |
|
26 |
-
|
27 |
|
28 |
-
## Intended
|
29 |
|
30 |
-
|
31 |
|
32 |
-
## Training and
|
33 |
|
34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
35 |
|
36 |
## Training procedure
|
37 |
|
|
|
21 |
- Loss: 1.1731
|
22 |
- Bleu: 42.9597
|
23 |
|
24 |
+
## Model Description
|
25 |
|
26 |
+
The model has been fine-tuned to perform machine translation from Kinyarwanda to English.
|
27 |
|
28 |
+
## Intended Uses & Limitations
|
29 |
|
30 |
+
The primary intended use of this model is for research purposes.
|
31 |
|
32 |
+
## Training and Evaluation Data
|
33 |
|
34 |
+
The model was fine-tuned using a combination of datasets from the following sources:
|
35 |
+
- [Digital Umuganda](https://huggingface.co/datasets/DigitalUmuganda/kinyarwanda-english-machine-translation-dataset/tree/main)
|
36 |
+
- [Masakhane](https://huggingface.co/datasets/masakhane/mafand/viewer/en-kin/validation)
|
37 |
+
- [Muennighoff](https://huggingface.co/datasets/Muennighoff/flores200)
|
38 |
+
|
39 |
+
For the training of the machine translation model, the dataset underwent the following preprocessing steps:
|
40 |
+
- Text was converted to lowercase
|
41 |
+
- Digits were removed
|
42 |
+
|
43 |
+
The combined dataset was divided into training and validation sets, with a split of 90% for training and 10% for validation.
|
44 |
|
45 |
## Training procedure
|
46 |
|