Hugging Face's logo

PontifexMaximus
/
Arabic2

Text2Text Generation
Transformers
PyTorch
TensorBoard
marian
generated_from_trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
1
PontifexMaximus commited on
Commit
2236aac
•
1 Parent(s): c6438f0

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +44 -30
README.md CHANGED
@@ -3,7 +3,7 @@ license: apache-2.0
3
  tags:
4
  - generated_from_trainer
5
  datasets:
6
- - opus100
7
  metrics:
8
  - bleu
9
  model-index:
@@ -13,13 +13,13 @@ model-index:
13
  name: Sequence-to-sequence Language Modeling
14
  type: text2text-generation
15
  dataset:
16
- name: opus100
17
- type: opus100
18
  args: ar-en
19
  metrics:
20
  - name: Bleu
21
  type: bleu
22
- value: 46.8089
23
  ---
24
 
25
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -27,11 +27,11 @@ should probably proofread and complete it, then remove this comment. -->
27
 
28
  # opus-mt-ar-en-finetuned-ar-to-en
29
 
30
- This model is a fine-tuned version of [Helsinki-NLP/opus-mt-ar-en](https://huggingface.co/Helsinki-NLP/opus-mt-ar-en) on the opus100 dataset.
31
  It achieves the following results on the evaluation set:
32
- - Loss: 1.0713
33
- - Bleu: 46.8089
34
- - Gen Len: 14.1755
35
 
36
  ## Model description
37
 
@@ -51,39 +51,53 @@ More information needed
51
 
52
  The following hyperparameters were used during training:
53
  - learning_rate: 2e-06
54
- - train_batch_size: 32
55
- - eval_batch_size: 32
56
  - seed: 42
57
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
58
  - lr_scheduler_type: linear
59
- - num_epochs: 16
60
  - mixed_precision_training: Native AMP
61
 
62
  ### Training results
63
 
64
  | Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
65
  |:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|
66
- | No log | 1.0 | 312 | 1.2132 | 43.7663 | 14.4193 |
67
- | 1.3072 | 2.0 | 624 | 1.1869 | 44.1712 | 14.4054 |
68
- | 1.3072 | 3.0 | 936 | 1.1675 | 44.5448 | 14.2182 |
69
- | 1.2535 | 4.0 | 1248 | 1.1510 | 44.8762 | 14.2004 |
70
- | 1.2309 | 5.0 | 1560 | 1.1375 | 45.2067 | 14.1375 |
71
- | 1.2309 | 6.0 | 1872 | 1.1251 | 45.4479 | 14.1887 |
72
- | 1.21 | 7.0 | 2184 | 1.1145 | 45.7117 | 14.2103 |
73
- | 1.21 | 8.0 | 2496 | 1.1051 | 45.951 | 14.1665 |
74
- | 1.1896 | 9.0 | 2808 | 1.0968 | 46.1647 | 14.178 |
75
- | 1.1837 | 10.0 | 3120 | 1.0899 | 46.342 | 14.1819 |
76
- | 1.1837 | 11.0 | 3432 | 1.0842 | 46.4735 | 14.1672 |
77
- | 1.1589 | 12.0 | 3744 | 1.0795 | 46.561 | 14.1729 |
78
- | 1.1523 | 13.0 | 4056 | 1.0759 | 46.6884 | 14.1706 |
79
- | 1.1523 | 14.0 | 4368 | 1.0733 | 46.7542 | 14.1735 |
80
- | 1.1524 | 15.0 | 4680 | 1.0718 | 46.7835 | 14.1712 |
81
- | 1.1524 | 16.0 | 4992 | 1.0713 | 46.8089 | 14.1755 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
82
 
83
 
84
  ### Framework versions
85
 
86
- - Transformers 4.19.2
87
- - Pytorch 1.7.1+cu110
88
- - Datasets 2.2.2
89
  - Tokenizers 0.12.1
 
3
  tags:
4
  - generated_from_trainer
5
  datasets:
6
+ - opus_infopankki
7
  metrics:
8
  - bleu
9
  model-index:
 
13
  name: Sequence-to-sequence Language Modeling
14
  type: text2text-generation
15
  dataset:
16
+ name: opus_infopankki
17
+ type: opus_infopankki
18
  args: ar-en
19
  metrics:
20
  - name: Bleu
21
  type: bleu
22
+ value: 53.5086
23
  ---
24
 
25
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
27
 
28
  # opus-mt-ar-en-finetuned-ar-to-en
29
 
30
+ This model is a fine-tuned version of [Helsinki-NLP/opus-mt-ar-en](https://huggingface.co/Helsinki-NLP/opus-mt-ar-en) on the opus_infopankki dataset.
31
  It achieves the following results on the evaluation set:
32
+ - Loss: 0.7636
33
+ - Bleu: 53.5086
34
+ - Gen Len: 13.5728
35
 
36
  ## Model description
37
 
 
51
 
52
  The following hyperparameters were used during training:
53
  - learning_rate: 2e-06
54
+ - train_batch_size: 64
55
+ - eval_batch_size: 64
56
  - seed: 42
57
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
58
  - lr_scheduler_type: linear
59
+ - num_epochs: 30
60
  - mixed_precision_training: Native AMP
61
 
62
  ### Training results
63
 
64
  | Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
65
  |:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|
66
+ | No log | 1.0 | 278 | 1.5114 | 35.2767 | 14.2084 |
67
+ | 1.6677 | 2.0 | 556 | 1.4025 | 37.5243 | 14.0245 |
68
+ | 1.6677 | 3.0 | 834 | 1.3223 | 39.4262 | 13.8101 |
69
+ | 1.4743 | 4.0 | 1112 | 1.2567 | 40.7045 | 13.8533 |
70
+ | 1.4743 | 5.0 | 1390 | 1.2001 | 41.8356 | 13.8083 |
71
+ | 1.3428 | 6.0 | 1668 | 1.1504 | 43.2448 | 13.6958 |
72
+ | 1.3428 | 7.0 | 1946 | 1.1072 | 44.177 | 13.6783 |
73
+ | 1.2595 | 8.0 | 2224 | 1.0701 | 45.17 | 13.6587 |
74
+ | 1.1829 | 9.0 | 2502 | 1.0345 | 45.9612 | 13.6706 |
75
+ | 1.1829 | 10.0 | 2780 | 1.0042 | 46.9009 | 13.6236 |
76
+ | 1.1188 | 11.0 | 3058 | 0.9760 | 47.7478 | 13.6205 |
77
+ | 1.1188 | 12.0 | 3336 | 0.9505 | 48.3082 | 13.6283 |
78
+ | 1.0735 | 13.0 | 3614 | 0.9270 | 48.9782 | 13.6203 |
79
+ | 1.0735 | 14.0 | 3892 | 0.9060 | 49.5541 | 13.6311 |
80
+ | 1.0269 | 15.0 | 4170 | 0.8869 | 49.9905 | 13.6222 |
81
+ | 1.0269 | 16.0 | 4448 | 0.8700 | 50.4806 | 13.6047 |
82
+ | 0.9983 | 17.0 | 4726 | 0.8538 | 50.9186 | 13.6159 |
83
+ | 0.9647 | 18.0 | 5004 | 0.8398 | 51.3492 | 13.6146 |
84
+ | 0.9647 | 19.0 | 5282 | 0.8271 | 51.7219 | 13.5275 |
85
+ | 0.9398 | 20.0 | 5560 | 0.8156 | 52.0177 | 13.5756 |
86
+ | 0.9398 | 21.0 | 5838 | 0.8053 | 52.3619 | 13.5807 |
87
+ | 0.9206 | 22.0 | 6116 | 0.7963 | 52.6051 | 13.5652 |
88
+ | 0.9206 | 23.0 | 6394 | 0.7885 | 52.8322 | 13.5669 |
89
+ | 0.9012 | 24.0 | 6672 | 0.7818 | 52.9402 | 13.5701 |
90
+ | 0.9012 | 25.0 | 6950 | 0.7762 | 53.1182 | 13.5695 |
91
+ | 0.8965 | 26.0 | 7228 | 0.7717 | 53.1596 | 13.5612 |
92
+ | 0.8836 | 27.0 | 7506 | 0.7681 | 53.3116 | 13.5719 |
93
+ | 0.8836 | 28.0 | 7784 | 0.7656 | 53.4399 | 13.5758 |
94
+ | 0.8777 | 29.0 | 8062 | 0.7642 | 53.4805 | 13.5737 |
95
+ | 0.8777 | 30.0 | 8340 | 0.7636 | 53.5086 | 13.5728 |
96
 
97
 
98
  ### Framework versions
99
 
100
+ - Transformers 4.20.1
101
+ - Pytorch 1.12.0
102
+ - Datasets 2.3.2
103
  - Tokenizers 0.12.1