Update README.md
Browse files
README.md
CHANGED
@@ -109,30 +109,45 @@ The following hyperparameters were set on the Fairseq toolkit:
|
|
109 |
| Warmup updates | 8000 |
|
110 |
| Dropout | 0.1 |
|
111 |
| Label smoothing | 0.1 |
|
|
|
112 |
The model was trained for a total of 19.000 updates. Weights were saved every 1000 updates and reported results are the average of the last 4 checkpoints.
|
|
|
113 |
## Evaluation
|
|
|
114 |
### Variable and metrics
|
115 |
-
|
|
|
|
|
116 |
### Evaluation results
|
|
|
117 |
Below are the evaluation results on the machine translation from Catalan to Italian compared to [Softcatalà](https://www.softcatala.org/) and [Google Translate](https://translate.google.es/?hl=es):
|
|
|
118 |
| Test set | SoftCatalà | Google Translate |mt-aina-it-ca|
|
119 |
|----------------------|------------|------------------|---------------|
|
120 |
| Flores 101 dev | 25,4 | **30,4** | 27,5 |
|
121 |
| Flores 101 devtest |26,6 | **31,2** | 27,7 |
|
122 |
| NTREX | 29,3 | **33,5** | 30,7 |
|
123 |
| Average | 27,1 | **31,7** | 28,6 |
|
|
|
124 |
## Additional information
|
|
|
125 |
### Author
|
126 |
Language Technologies Unit (LangTech) at the Barcelona Supercomputing Center.
|
|
|
127 |
### Contact information
|
128 |
For further information, send an email to <langtech@bsc.es>
|
|
|
129 |
### Copyright
|
130 |
Copyright Language Technologies Unit at Barcelona Supercomputing Center (2023)
|
|
|
131 |
### Licensing information
|
132 |
This work is licensed under a [Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0)
|
|
|
133 |
### Funding
|
134 |
This work was funded by the Departament de la Vicepresidència i de Polítiques Digitals i Territori de la Generalitat de Catalunya within the framework of Projecte AINA.
|
|
|
135 |
### Disclaimer
|
|
|
136 |
<details>
|
137 |
<summary>Click to expand</summary>
|
138 |
The models published in this repository are intended for a generalist purpose and are available to third parties. These models may have bias and/or any other undesirable distortions.
|
|
|
109 |
| Warmup updates | 8000 |
|
110 |
| Dropout | 0.1 |
|
111 |
| Label smoothing | 0.1 |
|
112 |
+
|
113 |
The model was trained for a total of 19.000 updates. Weights were saved every 1000 updates and reported results are the average of the last 4 checkpoints.
|
114 |
+
|
115 |
## Evaluation
|
116 |
+
|
117 |
### Variable and metrics
|
118 |
+
|
119 |
+
We use the BLEU score for evaluation on the [Flores-101](https://github.com/facebookresearch/flores), and [NTREX](https://github.com/MicrosoftTranslator/NTREX) evaluation datasets.
|
120 |
+
|
121 |
### Evaluation results
|
122 |
+
|
123 |
Below are the evaluation results on the machine translation from Catalan to Italian compared to [Softcatalà](https://www.softcatala.org/) and [Google Translate](https://translate.google.es/?hl=es):
|
124 |
+
|
125 |
| Test set | SoftCatalà | Google Translate |mt-aina-it-ca|
|
126 |
|----------------------|------------|------------------|---------------|
|
127 |
| Flores 101 dev | 25,4 | **30,4** | 27,5 |
|
128 |
| Flores 101 devtest |26,6 | **31,2** | 27,7 |
|
129 |
| NTREX | 29,3 | **33,5** | 30,7 |
|
130 |
| Average | 27,1 | **31,7** | 28,6 |
|
131 |
+
|
132 |
## Additional information
|
133 |
+
|
134 |
### Author
|
135 |
Language Technologies Unit (LangTech) at the Barcelona Supercomputing Center.
|
136 |
+
|
137 |
### Contact information
|
138 |
For further information, send an email to <langtech@bsc.es>
|
139 |
+
|
140 |
### Copyright
|
141 |
Copyright Language Technologies Unit at Barcelona Supercomputing Center (2023)
|
142 |
+
|
143 |
### Licensing information
|
144 |
This work is licensed under a [Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0)
|
145 |
+
|
146 |
### Funding
|
147 |
This work was funded by the Departament de la Vicepresidència i de Polítiques Digitals i Territori de la Generalitat de Catalunya within the framework of Projecte AINA.
|
148 |
+
|
149 |
### Disclaimer
|
150 |
+
|
151 |
<details>
|
152 |
<summary>Click to expand</summary>
|
153 |
The models published in this repository are intended for a generalist purpose and are available to third parties. These models may have bias and/or any other undesirable distortions.
|