Commit
•
5cd8050
1
Parent(s):
7b6ccdc
Update README.md
Browse files
README.md
CHANGED
|
@@ -11,17 +11,11 @@ license: apache-2.0
|
|
| 11 |
|
| 12 |
# HelsinkiNLP-FineTuned-Legal-es-zh
|
| 13 |
|
| 14 |
-
This model is a fine-tuned version of [Helsinki-NLP/opus-tatoeba-es-zh](https://huggingface.co/Helsinki-NLP/opus-tatoeba-es-zh) on a dataset constructed by the author himself.
|
| 15 |
-
It achieves the following results on the evaluation set:
|
| 16 |
-
- Loss: 2.0905
|
| 17 |
-
|
| 18 |
-
## Model description
|
| 19 |
-
|
| 20 |
-
Transformer-based NMT model to translate from Spanish to Simplified Chinese, fine-tuned for legal domain.
|
| 21 |
|
| 22 |
## Intended uses & limitations
|
| 23 |
|
| 24 |
-
This model is the result of the master graduation thesis for the Tradumatics: Translation Technologies program at the Autonomous University of Barcelona.
|
| 25 |
|
| 26 |
The thesis intends to explain various theories and certain algorithm details about neural machine translation, thus this fine-tuned model only serves as a hands-on practice example for that objective, without any intention of productive usage.
|
| 27 |
|
|
@@ -31,9 +25,7 @@ The dataset is constructed from the Chinese translation of Spanish Civil Code, S
|
|
| 31 |
|
| 32 |
There are 9972 sentence pairs constructed. 1000 are used for evaluation and the rest for training.
|
| 33 |
|
| 34 |
-
## Training
|
| 35 |
-
|
| 36 |
-
### Training hyperparameters
|
| 37 |
|
| 38 |
The following hyperparameters were used during training:
|
| 39 |
- learning_rate: 2e-05
|
|
@@ -48,7 +40,7 @@ The following hyperparameters were used during training:
|
|
| 48 |
- weight_decay: 0.01
|
| 49 |
- early_stopping_patience: 8
|
| 50 |
|
| 51 |
-
|
| 52 |
|
| 53 |
| Training Loss | Epoch | Step | Validation Loss |
|
| 54 |
|:-------------:|:-----:|:----:|:---------------:|
|
|
@@ -75,8 +67,7 @@ The following hyperparameters were used during training:
|
|
| 75 |
| 1.1238 | 7.49 | 8400 | 2.1102 |
|
| 76 |
| 1.1417 | 7.84 | 8800 | 2.1078 |
|
| 77 |
|
| 78 |
-
|
| 79 |
-
### Framework versions
|
| 80 |
|
| 81 |
- Transformers 4.7.0
|
| 82 |
- Pytorch 1.8.1+cu101
|
|
|
|
| 11 |
|
| 12 |
# HelsinkiNLP-FineTuned-Legal-es-zh
|
| 13 |
|
| 14 |
+
This model is a fine-tuned version of [Helsinki-NLP/opus-tatoeba-es-zh](https://huggingface.co/Helsinki-NLP/opus-tatoeba-es-zh) on a dataset of legal domain constructed by the author himself.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
## Intended uses & limitations
|
| 17 |
|
| 18 |
+
This model is the result of the master graduation thesis for the Tradumatics: Translation Technologies program at the Autonomous University of Barcelona. Please refer to GitHub repo created for this thesis for full-text and relative open-sourced materials: https://github.com/guocheng98/MUTTT2020_TFM_ZGC
|
| 19 |
|
| 20 |
The thesis intends to explain various theories and certain algorithm details about neural machine translation, thus this fine-tuned model only serves as a hands-on practice example for that objective, without any intention of productive usage.
|
| 21 |
|
|
|
|
| 25 |
|
| 26 |
There are 9972 sentence pairs constructed. 1000 are used for evaluation and the rest for training.
|
| 27 |
|
| 28 |
+
## Training hyperparameters
|
|
|
|
|
|
|
| 29 |
|
| 30 |
The following hyperparameters were used during training:
|
| 31 |
- learning_rate: 2e-05
|
|
|
|
| 40 |
- weight_decay: 0.01
|
| 41 |
- early_stopping_patience: 8
|
| 42 |
|
| 43 |
+
## Training results
|
| 44 |
|
| 45 |
| Training Loss | Epoch | Step | Validation Loss |
|
| 46 |
|:-------------:|:-----:|:----:|:---------------:|
|
|
|
|
| 67 |
| 1.1238 | 7.49 | 8400 | 2.1102 |
|
| 68 |
| 1.1417 | 7.84 | 8800 | 2.1078 |
|
| 69 |
|
| 70 |
+
## Framework versions
|
|
|
|
| 71 |
|
| 72 |
- Transformers 4.7.0
|
| 73 |
- Pytorch 1.8.1+cu101
|