Hugging Face's logo

mahwizzzz
/
urdu_text_correction

Text2Text Generation
Transformers
Safetensors
mbart
Generated from Trainer
Model card Files Community
mahwizzzz commited on
Commit
d049f15
·
verified ·
1 Parent(s): fc202c3

Model save

Browse files
Files changed (2) hide show
  1. README.md +98 -0
  2. generation_config.json +0 -1
README.md ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: transformers
3
+ license: mit
4
+ base_model: facebook/mbart-large-50
5
+ tags:
6
+ - generated_from_trainer
7
+ metrics:
8
+ - wer
9
+ - bleu
10
+ - rouge
11
+ model-index:
12
+ - name: urdu_text_correction
13
+ results: []
14
+ ---
15
+
16
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
+ should probably proofread and complete it, then remove this comment. -->
18
+
19
+ # urdu_text_correction
20
+
21
+ This model is a fine-tuned version of [facebook/mbart-large-50](https://huggingface.co/facebook/mbart-large-50) on an unknown dataset.
22
+ It achieves the following results on the evaluation set:
23
+ - Loss: 0.4305
24
+ - Wer: 0.1795
25
+ - Cer: 0.0761
26
+ - Bleu: 0.6996
27
+ - Rouge1: 0.2025
28
+ - Rouge2: 0.0699
29
+ - Rougel: 0.2023
30
+ - Meteor: 0.8296
31
+ - Gen Len: 28.4033
32
+ - Exact Match: 0.1096
33
+
34
+ ## Model description
35
+
36
+ More information needed
37
+
38
+ ## Intended uses & limitations
39
+
40
+ More information needed
41
+
42
+ ## Training and evaluation data
43
+
44
+ More information needed
45
+
46
+ ## Training procedure
47
+
48
+ ### Training hyperparameters
49
+
50
+ The following hyperparameters were used during training:
51
+ - learning_rate: 3e-05
52
+ - train_batch_size: 32
53
+ - eval_batch_size: 32
54
+ - seed: 42
55
+ - gradient_accumulation_steps: 4
56
+ - total_train_batch_size: 128
57
+ - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
58
+ - lr_scheduler_type: cosine
59
+ - lr_scheduler_warmup_steps: 500
60
+ - num_epochs: 3
61
+ - mixed_precision_training: Native AMP
62
+
63
+ ### Training results
64
+
65
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Cer | Bleu | Rouge1 | Rouge2 | Rougel | Meteor | Gen Len | Exact Match |
66
+ |:-------------:|:------:|:-----:|:---------------:|:------:|:------:|:------:|:------:|:------:|:------:|:------:|:-------:|:-----------:|
67
+ | 1.472 | 0.1209 | 500 | 1.2427 | 0.4582 | 0.2987 | 0.3494 | 0.1559 | 0.041 | 0.1562 | 0.5446 | 27.6629 | 0.0032 |
68
+ | 0.9869 | 0.2419 | 1000 | 0.8662 | 0.3212 | 0.1755 | 0.5057 | 0.1779 | 0.0541 | 0.1778 | 0.6863 | 28.232 | 0.0227 |
69
+ | 0.8582 | 0.3628 | 1500 | 0.7816 | 0.2878 | 0.1529 | 0.555 | 0.1837 | 0.0586 | 0.1838 | 0.7262 | 28.4877 | 0.0327 |
70
+ | 0.7774 | 0.4837 | 2000 | 0.6885 | 0.257 | 0.1289 | 0.5881 | 0.1866 | 0.0603 | 0.1865 | 0.7504 | 27.897 | 0.0478 |
71
+ | 0.6964 | 0.6047 | 2500 | 0.6298 | 0.2442 | 0.1172 | 0.6074 | 0.1896 | 0.0612 | 0.1894 | 0.7662 | 28.4579 | 0.0548 |
72
+ | 0.6468 | 0.7256 | 3000 | 0.5851 | 0.224 | 0.1037 | 0.6326 | 0.1951 | 0.068 | 0.1952 | 0.7852 | 28.107 | 0.0676 |
73
+ | 0.6148 | 0.8465 | 3500 | 0.5557 | 0.2224 | 0.1025 | 0.639 | 0.1935 | 0.0648 | 0.1935 | 0.7871 | 28.1589 | 0.0678 |
74
+ | 0.5834 | 0.9675 | 4000 | 0.5342 | 0.2112 | 0.096 | 0.6535 | 0.1959 | 0.0638 | 0.1959 | 0.7989 | 28.1429 | 0.0769 |
75
+ | 0.5252 | 1.0883 | 4500 | 0.5173 | 0.2035 | 0.091 | 0.662 | 0.197 | 0.068 | 0.1971 | 0.8044 | 28.2387 | 0.083 |
76
+ | 0.5176 | 1.2092 | 5000 | 0.5023 | 0.2032 | 0.0911 | 0.6637 | 0.1982 | 0.0691 | 0.1985 | 0.8047 | 28.2411 | 0.0807 |
77
+ | 0.5031 | 1.3301 | 5500 | 0.4873 | 0.1958 | 0.0846 | 0.6754 | 0.1969 | 0.0691 | 0.1969 | 0.8146 | 28.3568 | 0.0911 |
78
+ | 0.4887 | 1.4511 | 6000 | 0.4771 | 0.1917 | 0.0836 | 0.6807 | 0.2003 | 0.0698 | 0.2002 | 0.8164 | 28.3507 | 0.0941 |
79
+ | 0.4797 | 1.5720 | 6500 | 0.4696 | 0.1912 | 0.0833 | 0.6822 | 0.1998 | 0.0685 | 0.2002 | 0.8183 | 28.3144 | 0.0975 |
80
+ | 0.4724 | 1.6929 | 7000 | 0.4599 | 0.1868 | 0.0802 | 0.6881 | 0.1992 | 0.0692 | 0.1992 | 0.8231 | 28.3751 | 0.1024 |
81
+ | 0.4674 | 1.8139 | 7500 | 0.4532 | 0.1867 | 0.0804 | 0.6889 | 0.1998 | 0.0715 | 0.1999 | 0.823 | 28.4065 | 0.0996 |
82
+ | 0.4548 | 1.9348 | 8000 | 0.4459 | 0.1826 | 0.0775 | 0.6952 | 0.2016 | 0.0704 | 0.2016 | 0.8268 | 28.3558 | 0.1071 |
83
+ | 0.4109 | 2.0556 | 8500 | 0.4430 | 0.184 | 0.0783 | 0.6925 | 0.2016 | 0.0711 | 0.2016 | 0.8252 | 28.4034 | 0.1036 |
84
+ | 0.4085 | 2.1766 | 9000 | 0.4400 | 0.1841 | 0.0789 | 0.6929 | 0.2016 | 0.0702 | 0.2015 | 0.8249 | 28.3683 | 0.1053 |
85
+ | 0.4056 | 2.2975 | 9500 | 0.4370 | 0.1819 | 0.0771 | 0.6968 | 0.2006 | 0.0699 | 0.2005 | 0.8282 | 28.417 | 0.1077 |
86
+ | 0.4005 | 2.4184 | 10000 | 0.4352 | 0.1823 | 0.0775 | 0.6968 | 0.2024 | 0.0704 | 0.2024 | 0.828 | 28.4263 | 0.1096 |
87
+ | 0.4031 | 2.5394 | 10500 | 0.4324 | 0.1802 | 0.0762 | 0.6994 | 0.2031 | 0.0705 | 0.203 | 0.8293 | 28.4203 | 0.1096 |
88
+ | 0.3984 | 2.6603 | 11000 | 0.4314 | 0.18 | 0.0766 | 0.699 | 0.2025 | 0.0705 | 0.2025 | 0.8292 | 28.3997 | 0.1096 |
89
+ | 0.3924 | 2.7812 | 11500 | 0.4310 | 0.1802 | 0.0766 | 0.699 | 0.2019 | 0.0696 | 0.2018 | 0.8289 | 28.4055 | 0.1085 |
90
+ | 0.3975 | 2.9022 | 12000 | 0.4305 | 0.1795 | 0.0761 | 0.6996 | 0.2025 | 0.0699 | 0.2023 | 0.8296 | 28.4033 | 0.1096 |
91
+
92
+
93
+ ### Framework versions
94
+
95
+ - Transformers 4.49.0
96
+ - Pytorch 2.6.0+cu118
97
+ - Datasets 3.3.2
98
+ - Tokenizers 0.21.0
generation_config.json CHANGED
@@ -1,5 +1,4 @@
1
  {
2
- "_from_model_config": true,
3
  "bos_token_id": 0,
4
  "decoder_start_token_id": 2,
5
  "early_stopping": true,
 
1
  {
 
2
  "bos_token_id": 0,
3
  "decoder_start_token_id": 2,
4
  "early_stopping": true,