Update README.md
Browse files
README.md
CHANGED
@@ -48,21 +48,6 @@ It's a straightforward and fast solution that creates a positive impact in the c
|
|
48 |
|
49 |
By using gender inclusive models we can help reducing gender bias in a language corpus by, for instance, adding data augmentation and creating different examples
|
50 |
|
51 |
-
## Model specs
|
52 |
-
|
53 |
-
This model is a fine-tuned version of [spanish-t5-small](https://huggingface.co/flax-community/spanish-t5-small) on the data described below.
|
54 |
-
It achieves the following results on the evaluation set:
|
55 |
-
- 'eval_bleu': 93.8347,
|
56 |
-
- 'eval_f1': 0.9904,
|
57 |
-
|
58 |
-
## Training procedure
|
59 |
-
### Training hyperparameters
|
60 |
-
The following hyperparameters were used during training:
|
61 |
-
- learning_rate: 1e-04
|
62 |
-
- train_batch_size: 32
|
63 |
-
- seed: 42
|
64 |
-
- num_epochs: 10
|
65 |
-
- weight_decay: 0,01
|
66 |
|
67 |
## Training and evaluation data
|
68 |
|
@@ -97,6 +82,23 @@ The data used for the model training has been manually created form a compilatio
|
|
97 |
[Nombra.en.red. En femenino y en masculino](https://www.inmujeres.gob.es/areasTematicas/educacion/publicaciones/serieLenguaje/docs/Nombra_en_red.pdf)
|
98 |
|
99 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
100 |
## Metrics
|
101 |
|
102 |
For training, we used both Blue (sacrebleu implementation in HF) and BertScore. The first one, a standard in Machine Translation processes, has been added for ensuring robustness of the newly generated data, while the second one is kept for keeping the expected semantic similarity.
|
|
|
48 |
|
49 |
By using gender inclusive models we can help reducing gender bias in a language corpus by, for instance, adding data augmentation and creating different examples
|
50 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
51 |
|
52 |
## Training and evaluation data
|
53 |
|
|
|
82 |
[Nombra.en.red. En femenino y en masculino](https://www.inmujeres.gob.es/areasTematicas/educacion/publicaciones/serieLenguaje/docs/Nombra_en_red.pdf)
|
83 |
|
84 |
|
85 |
+
## Model specs
|
86 |
+
|
87 |
+
This model is a fine-tuned version of [spanish-t5-small](https://huggingface.co/flax-community/spanish-t5-small) on the data described below.
|
88 |
+
It achieves the following results on the evaluation set:
|
89 |
+
- 'eval_bleu': 93.8347,
|
90 |
+
- 'eval_f1': 0.9904,
|
91 |
+
|
92 |
+
## Training procedure
|
93 |
+
### Training hyperparameters
|
94 |
+
The following hyperparameters were used during training:
|
95 |
+
- learning_rate: 1e-04
|
96 |
+
- train_batch_size: 32
|
97 |
+
- seed: 42
|
98 |
+
- num_epochs: 10
|
99 |
+
- weight_decay: 0,01
|
100 |
+
|
101 |
+
|
102 |
## Metrics
|
103 |
|
104 |
For training, we used both Blue (sacrebleu implementation in HF) and BertScore. The first one, a standard in Machine Translation processes, has been added for ensuring robustness of the newly generated data, while the second one is kept for keeping the expected semantic similarity.
|