Hugging Face's logo

Shehryar718
/
URDU-ASR

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
Shehryar718 commited on
Commit
359b218
•
1 Parent(s): 8efc8ef

End of training

Browse files
Files changed (3) hide show
  1. README.md +16 -20
  2. model.safetensors +1 -1
  3. runs/Nov03_12-17-42_276e28c95be2/events.out.tfevents.1699014042.276e28c95be2.686.0 +2 -2
README.md CHANGED
@@ -1,6 +1,4 @@
1
  ---
2
- license: apache-2.0
3
- base_model: Talha/URDU-ASR
4
  tags:
5
  - generated_from_trainer
6
  datasets:
@@ -22,7 +20,7 @@ model-index:
22
  metrics:
23
  - name: Wer
24
  type: wer
25
- value: 1.0023598591821734
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -30,11 +28,11 @@ should probably proofread and complete it, then remove this comment. -->
30
 
31
  # URDU-ASR
32
 
33
- This model is a fine-tuned version of [Talha/URDU-ASR](https://huggingface.co/Talha/URDU-ASR) on the common_voice_13_0 dataset.
34
  It achieves the following results on the evaluation set:
35
- - Loss: 3.1901
36
- - Wer: 1.0024
37
- - Cer: 0.9455
38
 
39
  ## Model description
40
 
@@ -53,34 +51,32 @@ More information needed
53
  ### Training hyperparameters
54
 
55
  The following hyperparameters were used during training:
56
- - learning_rate: 7.5e-05
57
  - train_batch_size: 8
58
  - eval_batch_size: 8
59
  - seed: 42
60
- - gradient_accumulation_steps: 16
61
- - total_train_batch_size: 128
62
- - optimizer: Adam with betas=(0.85,0.99) and epsilon=1e-08
63
  - lr_scheduler_type: linear
64
  - lr_scheduler_warmup_ratio: 0.1
65
  - num_epochs: 5
 
66
 
67
  ### Training results
68
 
69
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
70
  |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
71
- | 19.3758 | 0.59 | 25 | 8.7836 | 1.0 | 0.9999 |
72
- | 6.0744 | 1.17 | 50 | 4.7540 | 1.0 | 0.9999 |
73
- | 4.446 | 1.76 | 75 | 4.0785 | 1.0 | 0.9999 |
74
- | 3.7656 | 2.34 | 100 | 3.5164 | 1.0024 | 0.9457 |
75
- | 3.4626 | 2.93 | 125 | 3.3191 | 1.0024 | 0.9454 |
76
- | 3.2974 | 3.51 | 150 | 3.2566 | 1.0024 | 0.9449 |
77
- | 3.2203 | 4.1 | 175 | 3.2009 | 1.0024 | 0.9456 |
78
- | 3.1955 | 4.69 | 200 | 3.1901 | 1.0024 | 0.9455 |
79
 
80
 
81
  ### Framework versions
82
 
83
- - Transformers 4.34.1
84
  - Pytorch 2.1.0+cu118
85
  - Datasets 2.14.6
86
  - Tokenizers 0.14.1
 
1
  ---
 
 
2
  tags:
3
  - generated_from_trainer
4
  datasets:
 
20
  metrics:
21
  - name: Wer
22
  type: wer
23
+ value: 0.4850090912607838
24
  ---
25
 
26
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
28
 
29
  # URDU-ASR
30
 
31
+ This model was trained from scratch on the common_voice_13_0 dataset.
32
  It achieves the following results on the evaluation set:
33
+ - Loss: 0.6352
34
+ - Wer: 0.4850
35
+ - Cer: 0.2045
36
 
37
  ## Model description
38
 
 
51
  ### Training hyperparameters
52
 
53
  The following hyperparameters were used during training:
54
+ - learning_rate: 0.0003
55
  - train_batch_size: 8
56
  - eval_batch_size: 8
57
  - seed: 42
58
+ - gradient_accumulation_steps: 2
59
+ - total_train_batch_size: 16
60
+ - optimizer: Adam with betas=(0.9,0.99) and epsilon=1e-08
61
  - lr_scheduler_type: linear
62
  - lr_scheduler_warmup_ratio: 0.1
63
  - num_epochs: 5
64
+ - mixed_precision_training: Native AMP
65
 
66
  ### Training results
67
 
68
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
69
  |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
70
+ | 2.2192 | 1.0 | 341 | 0.6603 | 0.5302 | 0.2229 |
71
+ | 0.3189 | 2.0 | 683 | 0.6316 | 0.5287 | 0.2295 |
72
+ | 0.2507 | 3.0 | 1024 | 0.6513 | 0.5032 | 0.2141 |
73
+ | 0.2076 | 4.0 | 1366 | 0.6459 | 0.5038 | 0.2131 |
74
+ | 0.1711 | 4.99 | 1705 | 0.6352 | 0.4850 | 0.2045 |
 
 
 
75
 
76
 
77
  ### Framework versions
78
 
79
+ - Transformers 4.35.0
80
  - Pytorch 2.1.0+cu118
81
  - Datasets 2.14.6
82
  - Tokenizers 0.14.1
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:58a3a727789dee1f218dba9aa876ff83646ad3730fa42d2f6695be2c603bff7b
3
  size 1262168280
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8db45df545b408bd954a248c32f0778e9add4c32f7e5fbc5cc960fb3aa449221
3
  size 1262168280
runs/Nov03_12-17-42_276e28c95be2/events.out.tfevents.1699014042.276e28c95be2.686.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fe9a8ac3904872a89ca3ff1636f35232f7e8e61cfa300176009bbaa69af51397
3
- size 8510
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0aa5793cb2f2baf87fa43e17e878c213b60fe7cbbb7cbd87c0bedd2cde54022
3
+ size 8864