Hugging Face's logo

ogbi
/
results

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
ogbi commited on
Commit
dbce1ac
•
1 Parent(s): 6ff62cc

End of training

Browse files
Files changed (3) hide show
  1. README.md +19 -43
  2. model.safetensors +1 -1
  3. runs/Jun11_22-09-57_04c0d8a803ab/events.out.tfevents.1718143907.04c0d8a803ab.2735.0 +2 -2
README.md CHANGED
@@ -22,7 +22,7 @@ model-index:
22
  metrics:
23
  - name: Wer
24
  type: wer
25
- value: 0.47755303404045385
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -32,9 +32,9 @@ should probably proofread and complete it, then remove this comment. -->
32
 
33
  This model is a fine-tuned version of [facebook/mms-1b](https://huggingface.co/facebook/mms-1b) on the audiofolder dataset.
34
  It achieves the following results on the evaluation set:
35
- - Loss: 2.0362
36
- - Wer: 0.4776
37
- - Cer: 0.2418
38
 
39
  ## Model description
40
 
@@ -57,54 +57,30 @@ The following hyperparameters were used during training:
57
  - train_batch_size: 4
58
  - eval_batch_size: 8
59
  - seed: 42
60
- - gradient_accumulation_steps: 2
61
- - total_train_batch_size: 8
62
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
63
  - lr_scheduler_type: linear
64
  - lr_scheduler_warmup_steps: 500
65
- - num_epochs: 30
66
  - mixed_precision_training: Native AMP
67
 
68
  ### Training results
69
 
70
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
71
  |:-------------:|:-------:|:----:|:---------------:|:------:|:------:|
72
- | 3.0435 | 0.8247 | 40 | 2.9730 | 1.0 | 1.0 |
73
- | 1.7257 | 1.6495 | 80 | 1.5913 | 0.8155 | 0.3725 |
74
- | 1.2559 | 2.4742 | 120 | 1.3284 | 0.6922 | 0.3104 |
75
- | 1.0506 | 3.2990 | 160 | 1.2850 | 0.6374 | 0.2965 |
76
- | 0.9286 | 4.1237 | 200 | 1.2555 | 0.5530 | 0.2589 |
77
- | 0.9036 | 4.9485 | 240 | 1.2799 | 0.5969 | 0.2754 |
78
- | 0.8737 | 5.7732 | 280 | 1.3403 | 0.5826 | 0.2751 |
79
- | 0.7498 | 6.5979 | 320 | 1.2827 | 0.5590 | 0.2747 |
80
- | 0.772 | 7.4227 | 360 | 1.3825 | 0.5792 | 0.2854 |
81
- | 0.6421 | 8.2474 | 400 | 1.5132 | 0.6167 | 0.2903 |
82
- | 0.6352 | 9.0722 | 440 | 1.4360 | 0.5654 | 0.2712 |
83
- | 0.6812 | 9.8969 | 480 | 1.3643 | 0.5816 | 0.2905 |
84
- | 0.6641 | 10.7216 | 520 | 1.5933 | 0.6014 | 0.2834 |
85
- | 0.5466 | 11.5464 | 560 | 1.4724 | 0.5565 | 0.2766 |
86
- | 0.5291 | 12.3711 | 600 | 1.5755 | 0.5545 | 0.2685 |
87
- | 0.41 | 13.1959 | 640 | 1.5586 | 0.5560 | 0.2744 |
88
- | 0.3679 | 14.0206 | 680 | 1.5788 | 0.5644 | 0.2746 |
89
- | 0.2847 | 14.8454 | 720 | 1.6370 | 0.5407 | 0.2655 |
90
- | 0.314 | 15.6701 | 760 | 1.7064 | 0.5456 | 0.2693 |
91
- | 0.231 | 16.4948 | 800 | 1.5744 | 0.5466 | 0.2705 |
92
- | 0.2123 | 17.3196 | 840 | 1.7739 | 0.5358 | 0.2661 |
93
- | 0.1241 | 18.1443 | 880 | 1.8826 | 0.5116 | 0.2536 |
94
- | 0.1361 | 18.9691 | 920 | 1.8170 | 0.5264 | 0.2639 |
95
- | 0.1363 | 19.7938 | 960 | 1.7761 | 0.5131 | 0.2550 |
96
- | 0.1061 | 20.6186 | 1000 | 1.9435 | 0.5032 | 0.2531 |
97
- | 0.077 | 21.4433 | 1040 | 1.9086 | 0.5067 | 0.2554 |
98
- | 0.086 | 22.2680 | 1080 | 1.8842 | 0.5150 | 0.2560 |
99
- | 0.0791 | 23.0928 | 1120 | 1.9835 | 0.5027 | 0.2504 |
100
- | 0.0925 | 23.9175 | 1160 | 1.9887 | 0.4983 | 0.2493 |
101
- | 0.1021 | 24.7423 | 1200 | 1.9644 | 0.5002 | 0.2465 |
102
- | 0.0623 | 25.5670 | 1240 | 2.0038 | 0.4889 | 0.2456 |
103
- | 0.0537 | 26.3918 | 1280 | 1.9788 | 0.4879 | 0.2455 |
104
- | 0.0428 | 27.2165 | 1320 | 1.9889 | 0.4780 | 0.2433 |
105
- | 0.1203 | 28.0412 | 1360 | 2.0245 | 0.4805 | 0.2459 |
106
- | 0.0365 | 28.8660 | 1400 | 2.0261 | 0.4810 | 0.2427 |
107
- | 0.0247 | 29.6907 | 1440 | 2.0362 | 0.4776 | 0.2418 |
108
 
109
 
110
  ### Framework versions
 
22
  metrics:
23
  - name: Wer
24
  type: wer
25
+ value: 0.5569807597434633
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
32
 
33
  This model is a fine-tuned version of [facebook/mms-1b](https://huggingface.co/facebook/mms-1b) on the audiofolder dataset.
34
  It achieves the following results on the evaluation set:
35
+ - Loss: 1.5862
36
+ - Wer: 0.5570
37
+ - Cer: 0.2802
38
 
39
  ## Model description
40
 
 
57
  - train_batch_size: 4
58
  - eval_batch_size: 8
59
  - seed: 42
60
+ - gradient_accumulation_steps: 4
61
+ - total_train_batch_size: 16
62
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
63
  - lr_scheduler_type: linear
64
  - lr_scheduler_warmup_steps: 500
65
+ - num_epochs: 20
66
  - mixed_precision_training: Native AMP
67
 
68
  ### Training results
69
 
70
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
71
  |:-------------:|:-------:|:----:|:---------------:|:------:|:------:|
72
+ | 3.6617 | 1.6495 | 40 | 3.4133 | 0.9941 | 0.8573 |
73
+ | 3.0704 | 3.2990 | 80 | 3.0021 | 1.0 | 0.9985 |
74
+ | 2.8663 | 4.9485 | 120 | 2.8530 | 0.9531 | 0.7800 |
75
+ | 1.995 | 6.5979 | 160 | 1.8117 | 0.9260 | 0.5088 |
76
+ | 1.3171 | 8.2474 | 200 | 1.3296 | 0.6744 | 0.2939 |
77
+ | 1.0979 | 9.8969 | 240 | 1.2741 | 0.5787 | 0.2667 |
78
+ | 0.9623 | 11.5464 | 280 | 1.2679 | 0.5570 | 0.2624 |
79
+ | 0.8367 | 13.1959 | 320 | 1.3316 | 0.5910 | 0.2622 |
80
+ | 0.7728 | 14.8454 | 360 | 1.4157 | 0.5575 | 0.2705 |
81
+ | 0.5491 | 16.4948 | 400 | 1.5152 | 0.5668 | 0.2760 |
82
+ | 0.5494 | 18.1443 | 440 | 1.6297 | 0.5417 | 0.2683 |
83
+ | 0.5065 | 19.7938 | 480 | 1.5862 | 0.5570 | 0.2802 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
84
 
85
 
86
  ### Framework versions
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1f8a692629f0c39f50ae245b7a0f4a8f9673019fce533f186fba92e96a5fc753
3
  size 3850275464
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b612b1ebbba413c8d4999af071687e951d8846f67d024ff8b7a8c8a092ba8520
3
  size 3850275464
runs/Jun11_22-09-57_04c0d8a803ab/events.out.tfevents.1718143907.04c0d8a803ab.2735.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3447699d0d79f4de3d76b09d716dca0c97942626ef4a72b3e1c99224cb3f090e
3
- size 12200
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9627068090d5f43c518ede56bce085c56f7167013029525dc1cc297c6b39a6e5
3
+ size 21017