Hugging Face's logo

evanarlian
/
wav2vec2-xls-r-113m-id

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
1
evanarlian commited on
Commit
00a9b3b
•
1 Parent(s): 8541d74

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +36 -31
README.md CHANGED
@@ -17,7 +17,7 @@ model-index:
17
  metrics:
18
  - name: Wer
19
  type: wer
20
- value: 0.39516649755557604
21
  ---
22
 
23
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -27,8 +27,8 @@ should probably proofread and complete it, then remove this comment. -->
27
 
28
  This model is a fine-tuned version of [evanarlian/distil-wav2vec2-xls-r-113m-id](https://huggingface.co/evanarlian/distil-wav2vec2-xls-r-113m-id) on the evanarlian/common_voice_11_0_id_filtered dataset.
29
  It achieves the following results on the evaluation set:
30
- - Loss: 0.3280
31
- - Wer: 0.3952
32
 
33
  ## Model description
34
 
@@ -56,40 +56,45 @@ The following hyperparameters were used during training:
56
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
57
  - lr_scheduler_type: linear
58
  - lr_scheduler_warmup_ratio: 0.3
59
- - num_epochs: 25.0
60
  - mixed_precision_training: Native AMP
61
 
62
  ### Training results
63
 
64
  | Training Loss | Epoch | Step | Validation Loss | Wer |
65
  |:-------------:|:-----:|:-----:|:---------------:|:------:|
66
- | 3.2512 | 0.92 | 1000 | 2.9098 | 1.0000 |
67
- | 2.163 | 1.84 | 2000 | 1.4810 | 0.9941 |
68
- | 1.2472 | 2.75 | 3000 | 0.9604 | 0.9196 |
69
- | 1.0166 | 3.67 | 4000 | 0.8240 | 0.8498 |
70
- | 0.8765 | 4.59 | 5000 | 0.6873 | 0.7741 |
71
- | 0.7712 | 5.51 | 6000 | 0.6083 | 0.7111 |
72
- | 0.6892 | 6.43 | 7000 | 0.5546 | 0.6592 |
73
- | 0.6314 | 7.35 | 8000 | 0.5022 | 0.6108 |
74
- | 0.5779 | 8.26 | 9000 | 0.4850 | 0.5825 |
75
- | 0.5245 | 9.18 | 10000 | 0.4665 | 0.5538 |
76
- | 0.4858 | 10.1 | 11000 | 0.4282 | 0.5279 |
77
- | 0.4616 | 11.02 | 12000 | 0.4053 | 0.5082 |
78
- | 0.421 | 11.94 | 13000 | 0.3809 | 0.4935 |
79
- | 0.4064 | 12.86 | 14000 | 0.3706 | 0.4781 |
80
- | 0.3758 | 13.77 | 15000 | 0.3743 | 0.4672 |
81
- | 0.3598 | 14.69 | 16000 | 0.3571 | 0.4521 |
82
- | 0.3441 | 15.61 | 17000 | 0.3455 | 0.4368 |
83
- | 0.3279 | 16.53 | 18000 | 0.3398 | 0.4386 |
84
- | 0.3086 | 17.45 | 19000 | 0.3512 | 0.4284 |
85
- | 0.3013 | 18.37 | 20000 | 0.3321 | 0.4233 |
86
- | 0.2963 | 19.28 | 21000 | 0.3391 | 0.4178 |
87
- | 0.2831 | 20.2 | 22000 | 0.3438 | 0.4114 |
88
- | 0.2801 | 21.12 | 23000 | 0.3336 | 0.4056 |
89
- | 0.2623 | 22.04 | 24000 | 0.3317 | 0.4012 |
90
- | 0.263 | 22.96 | 25000 | 0.3280 | 0.4005 |
91
- | 0.2529 | 23.88 | 26000 | 0.3268 | 0.3951 |
92
- | 0.2492 | 24.79 | 27000 | 0.3280 | 0.3952 |
 
 
 
 
 
93
 
94
 
95
  ### Framework versions
 
17
  metrics:
18
  - name: Wer
19
  type: wer
20
+ value: 0.4274974633336408
21
  ---
22
 
23
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
27
 
28
  This model is a fine-tuned version of [evanarlian/distil-wav2vec2-xls-r-113m-id](https://huggingface.co/evanarlian/distil-wav2vec2-xls-r-113m-id) on the evanarlian/common_voice_11_0_id_filtered dataset.
29
  It achieves the following results on the evaluation set:
30
+ - Loss: 0.4804
31
+ - Wer: 0.4275
32
 
33
  ## Model description
34
 
 
56
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
57
  - lr_scheduler_type: linear
58
  - lr_scheduler_warmup_ratio: 0.3
59
+ - num_epochs: 30.0
60
  - mixed_precision_training: Native AMP
61
 
62
  ### Training results
63
 
64
  | Training Loss | Epoch | Step | Validation Loss | Wer |
65
  |:-------------:|:-----:|:-----:|:---------------:|:------:|
66
+ | 3.2694 | 0.92 | 1000 | 2.9168 | 1.0000 |
67
+ | 2.2449 | 1.84 | 2000 | 1.5711 | 0.9901 |
68
+ | 1.2118 | 2.75 | 3000 | 1.0133 | 0.9261 |
69
+ | 0.971 | 3.67 | 4000 | 0.8860 | 0.8743 |
70
+ | 0.8472 | 4.59 | 5000 | 0.7562 | 0.8180 |
71
+ | 0.7436 | 5.51 | 6000 | 0.6800 | 0.7505 |
72
+ | 0.6603 | 6.43 | 7000 | 0.6275 | 0.7023 |
73
+ | 0.5961 | 7.35 | 8000 | 0.5913 | 0.6589 |
74
+ | 0.5458 | 8.26 | 9000 | 0.5605 | 0.6358 |
75
+ | 0.5113 | 9.18 | 10000 | 0.5346 | 0.6039 |
76
+ | 0.463 | 10.1 | 11000 | 0.5052 | 0.5689 |
77
+ | 0.4326 | 11.02 | 12000 | 0.4880 | 0.5497 |
78
+ | 0.3981 | 11.94 | 13000 | 0.4778 | 0.5357 |
79
+ | 0.3602 | 12.86 | 14000 | 0.4656 | 0.5198 |
80
+ | 0.3501 | 13.77 | 15000 | 0.4510 | 0.5085 |
81
+ | 0.3199 | 14.69 | 16000 | 0.4617 | 0.5010 |
82
+ | 0.3058 | 15.61 | 17000 | 0.4385 | 0.4880 |
83
+ | 0.2844 | 16.53 | 18000 | 0.4638 | 0.4930 |
84
+ | 0.2729 | 17.45 | 19000 | 0.4594 | 0.4783 |
85
+ | 0.2648 | 18.37 | 20000 | 0.4521 | 0.4703 |
86
+ | 0.2515 | 19.28 | 21000 | 0.4727 | 0.4627 |
87
+ | 0.2428 | 20.2 | 22000 | 0.4566 | 0.4587 |
88
+ | 0.2343 | 21.12 | 23000 | 0.4554 | 0.4545 |
89
+ | 0.2228 | 22.04 | 24000 | 0.4670 | 0.4506 |
90
+ | 0.2135 | 22.96 | 25000 | 0.4458 | 0.4446 |
91
+ | 0.2067 | 23.88 | 26000 | 0.4571 | 0.4402 |
92
+ | 0.2065 | 24.79 | 27000 | 0.4680 | 0.4359 |
93
+ | 0.1968 | 25.71 | 28000 | 0.4702 | 0.4346 |
94
+ | 0.1914 | 26.63 | 29000 | 0.4687 | 0.4320 |
95
+ | 0.182 | 27.55 | 30000 | 0.4807 | 0.4332 |
96
+ | 0.1771 | 28.47 | 31000 | 0.4824 | 0.4308 |
97
+ | 0.1728 | 29.38 | 32000 | 0.4804 | 0.4275 |
98
 
99
 
100
  ### Framework versions