Hugging Face's logo

AliPouryousefi86
/
hezar-FineTuned-OCR-V2

Model card Files Community
AliPouryousefi86 commited on
Commit
3c55978
·
verified ·
1 Parent(s): a075291

Upload folder using huggingface_hub

Browse files
Files changed (6) hide show
  1. model.pt +3 -0
  2. model_config.yaml +83 -0
  3. preprocessor/image_processor_config.yaml +12 -0
  4. train/dataset_config.yaml +84 -0
  5. train/optimizer.pt +3 -0
  6. train/train_config.yaml +29 -0
model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b2f9e277bc285a99519e7a62963c9d08761eb587c8d10a4ad19ac2db18c60b7
3
+ size 36848443
model_config.yaml ADDED
@@ -0,0 +1,83 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ name: crnn_image2text
2
+ config_type: model
3
+ id2label:
4
+ 0: ''
5
+ 1: آ
6
+ 2: ا
7
+ 3: ب
8
+ 4: پ
9
+ 5: ت
10
+ 6: ث
11
+ 7: ج
12
+ 8: چ
13
+ 9: ح
14
+ 10: خ
15
+ 11: د
16
+ 12: ذ
17
+ 13: ر
18
+ 14: ز
19
+ 15: ژ
20
+ 16: س
21
+ 17: ش
22
+ 18: ص
23
+ 19: ض
24
+ 20: ط
25
+ 21: ظ
26
+ 22: ع
27
+ 23: غ
28
+ 24: ف
29
+ 25: ق
30
+ 26: ک
31
+ 27: گ
32
+ 28: ل
33
+ 29: م
34
+ 30: ن
35
+ 31: و
36
+ 32: ه
37
+ 33: ی
38
+ 34: ' '
39
+ 35: '۱'
40
+ 36: '۲'
41
+ 37: '۳'
42
+ 38: '۴'
43
+ 39: '۵'
44
+ 40: '۶'
45
+ 41: '۷'
46
+ 42: '۸'
47
+ 43: '۹'
48
+ 44: '۰'
49
+ 45: ء
50
+ 46: ؤ
51
+ 47: ئ
52
+ 48: أ
53
+ 49: ّ
54
+ 50: /
55
+ 51: (
56
+ 52: )
57
+ 53: +
58
+ 54: '-'
59
+ 55: ':'
60
+ 56: ،
61
+ 57: '!'
62
+ 58: .
63
+ 59: ؛
64
+ 60: '='
65
+ 61: '%'
66
+ 62: ؟
67
+ 63: '1'
68
+ 64: '2'
69
+ 65: '3'
70
+ 66: '4'
71
+ 67: '5'
72
+ 68: '6'
73
+ 69: '7'
74
+ 70: '8'
75
+ 71: '9'
76
+ 72: '0'
77
+ blank_id: 0
78
+ n_channels: 1
79
+ image_height: 32
80
+ image_width: 128
81
+ map2seq_in_dim: 1024
82
+ map2seq_out_dim: 64
83
+ rnn_dim: 256
preprocessor/image_processor_config.yaml ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ name: image_processor
2
+ config_type: preprocessor
3
+ mean:
4
+ - 0.6595
5
+ std:
6
+ - 0.1501
7
+ rescale: 0.00392156862745098
8
+ size:
9
+ - 384
10
+ - 32
11
+ mirror: true
12
+ gray_scale: true
train/dataset_config.yaml ADDED
@@ -0,0 +1,84 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ name: ocr
2
+ config_type: dataset
3
+ path: train.csv
4
+ task: image2text
5
+ hf_load_kwargs: {}
6
+ text_split_type: char_split
7
+ id2label:
8
+ 0: ''
9
+ 1: آ
10
+ 2: ا
11
+ 3: ب
12
+ 4: پ
13
+ 5: ت
14
+ 6: ث
15
+ 7: ج
16
+ 8: چ
17
+ 9: ح
18
+ 10: خ
19
+ 11: د
20
+ 12: ذ
21
+ 13: ر
22
+ 14: ز
23
+ 15: ژ
24
+ 16: س
25
+ 17: ش
26
+ 18: ص
27
+ 19: ض
28
+ 20: ط
29
+ 21: ظ
30
+ 22: ع
31
+ 23: غ
32
+ 24: ف
33
+ 25: ق
34
+ 26: ک
35
+ 27: گ
36
+ 28: ل
37
+ 29: م
38
+ 30: ن
39
+ 31: و
40
+ 32: ه
41
+ 33: ی
42
+ 34: ' '
43
+ 35: '۱'
44
+ 36: '۲'
45
+ 37: '۳'
46
+ 38: '۴'
47
+ 39: '۵'
48
+ 40: '۶'
49
+ 41: '۷'
50
+ 42: '۸'
51
+ 43: '۹'
52
+ 44: '۰'
53
+ 45: ء
54
+ 46: ؤ
55
+ 47: ئ
56
+ 48: أ
57
+ 49: ّ
58
+ 50: /
59
+ 51: (
60
+ 52: )
61
+ 53: +
62
+ 54: '-'
63
+ 55: ':'
64
+ 56: ،
65
+ 57: '!'
66
+ 58: .
67
+ 59: ؛
68
+ 60: '='
69
+ 61: '%'
70
+ 62: ؟
71
+ 63: '1'
72
+ 64: '2'
73
+ 65: '3'
74
+ 66: '4'
75
+ 67: '5'
76
+ 68: '6'
77
+ 69: '7'
78
+ 70: '8'
79
+ 71: '9'
80
+ 72: '0'
81
+ text_column: text
82
+ images_paths_column: image_path
83
+ max_length: 88
84
+ reverse_digits: true
train/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:233e5051bf6f63d42e0cbde791deece5bfa7d355ea795af5d5d9d41f1746eca8
3
+ size 73640378
train/train_config.yaml ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ name: trainer
2
+ config_type: trainer
3
+ output_dir: crnn-ocr-fa
4
+ task: image2text
5
+ device: cuda
6
+ num_epochs: 8
7
+ init_weights_from: hezarai/crnn-fa-printed-96-long
8
+ resume_from_checkpoint: true
9
+ num_dataloader_workers: 0
10
+ dataloader_shuffle: true
11
+ seed: 42
12
+ learning_rate: 2.0e-05
13
+ weight_decay: 0.0
14
+ batch_size: 16
15
+ gradient_accumulation_steps: 1
16
+ distributed: false
17
+ use_cpu: false
18
+ do_evaluate: true
19
+ evaluate_with_generate: true
20
+ metrics:
21
+ - cer
22
+ - wer
23
+ metric_for_best_model: evaluation.cer
24
+ save_enabled: true
25
+ save_freq: deprecated
26
+ save_steps: 7125
27
+ log_steps: 100
28
+ checkpoints_dir: checkpoints
29
+ logs_dir: logs