Hugging Face's logo

kiranpantha
/
10epochs-w2v-bert-2.0-nepali-unlabeled-1

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Nepali
wav2vec2-bert
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
kiranpantha commited on
Commit
65cc01e
·
verified ·
1 Parent(s): fdc60a6

End of training

Browse files
Files changed (3) hide show
  1. README.md +15 -48
  2. model.safetensors +1 -1
  3. runs/Oct26_21-08-08_ml/events.out.tfevents.1729956266.ml.4821.1 +2 -2
README.md CHANGED
@@ -23,7 +23,7 @@ model-index:
23
  metrics:
24
  - name: Wer
25
  type: wer
26
- value: 0.44157399486740806
27
  ---
28
 
29
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -33,9 +33,9 @@ should probably proofread and complete it, then remove this comment. -->
33
 
34
  This model is a fine-tuned version of [kiranpantha/w2v-bert-2.0-nepali](https://huggingface.co/kiranpantha/w2v-bert-2.0-nepali) on the kiranpantha/OpenSLR54-Balanced-Nepali dataset.
35
  It achieves the following results on the evaluation set:
36
- - Cer: 0.1081
37
- - Loss: 0.5478
38
- - Wer: 0.4416
39
 
40
  ## Model description
41
 
@@ -61,54 +61,21 @@ The following hyperparameters were used during training:
61
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
62
  - lr_scheduler_type: linear
63
  - lr_scheduler_warmup_steps: 500
64
- - num_epochs: 10
65
  - mixed_precision_training: Native AMP
66
 
67
  ### Training results
68
 
69
- | Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
70
- |:-------------:|:------:|:-----:|:------:|:---------------:|:------:|
71
- | 0.6781 | 0.24 | 300 | 0.0709 | 0.3132 | 0.3307 |
72
- | 0.6893 | 0.48 | 600 | 0.0904 | 0.3884 | 0.3814 |
73
- | 0.7145 | 0.72 | 900 | 0.1008 | 0.4009 | 0.4229 |
74
- | 0.6766 | 0.96 | 1200 | 0.1132 | 0.4541 | 0.4710 |
75
- | 0.6203 | 1.2 | 1500 | 0.1019 | 0.4530 | 0.4311 |
76
- | 0.6012 | 1.44 | 1800 | 0.0996 | 0.4123 | 0.4209 |
77
- | 0.5652 | 1.6800 | 2100 | 0.1058 | 0.4564 | 0.4520 |
78
- | 0.5543 | 1.92 | 2400 | 0.1038 | 0.4196 | 0.4301 |
79
- | 0.542 | 2.16 | 2700 | 0.1046 | 0.4174 | 0.4296 |
80
- | 0.508 | 2.4 | 3000 | 0.1107 | 0.4492 | 0.4515 |
81
- | 0.5139 | 2.64 | 3300 | 0.1065 | 0.4508 | 0.4542 |
82
- | 0.5375 | 2.88 | 3600 | 0.0984 | 0.4197 | 0.4188 |
83
- | 0.4918 | 3.12 | 3900 | 0.1043 | 0.4454 | 0.4284 |
84
- | 0.4756 | 3.36 | 4200 | 0.1030 | 0.4294 | 0.4234 |
85
- | 0.4519 | 3.6 | 4500 | 0.1069 | 0.4535 | 0.4388 |
86
- | 0.4276 | 3.84 | 4800 | 0.1018 | 0.4424 | 0.4246 |
87
- | 0.4392 | 4.08 | 5100 | 0.1050 | 0.4747 | 0.4317 |
88
- | 0.3968 | 4.32 | 5400 | 0.1006 | 0.4702 | 0.4113 |
89
- | 0.3926 | 4.5600 | 5700 | 0.1038 | 0.4667 | 0.4233 |
90
- | 0.3985 | 4.8 | 6000 | 0.1049 | 0.4451 | 0.4344 |
91
- | 0.399 | 5.04 | 6300 | 0.1095 | 0.4678 | 0.4517 |
92
- | 0.3322 | 5.28 | 6600 | 0.1102 | 0.4642 | 0.4320 |
93
- | 0.3851 | 5.52 | 6900 | 0.1112 | 0.4587 | 0.4465 |
94
- | 0.4644 | 5.76 | 7200 | 0.1369 | 0.5375 | 0.5227 |
95
- | 0.3065 | 6.0 | 7500 | 0.1014 | 0.5160 | 0.4193 |
96
- | 0.296 | 6.24 | 7800 | 0.1113 | 0.5292 | 0.4448 |
97
- | 0.2849 | 6.48 | 8100 | 0.1070 | 0.4961 | 0.4359 |
98
- | 0.3039 | 6.72 | 8400 | 0.1013 | 0.4727 | 0.4246 |
99
- | 0.2873 | 6.96 | 8700 | 0.1032 | 0.4992 | 0.4200 |
100
- | 0.2359 | 7.2 | 9000 | 0.1027 | 0.5055 | 0.4207 |
101
- | 0.2271 | 7.44 | 9300 | 0.1034 | 0.5132 | 0.4204 |
102
- | 0.224 | 7.68 | 9600 | 0.1040 | 0.5238 | 0.4171 |
103
- | 0.2344 | 7.92 | 9900 | 0.1029 | 0.5154 | 0.4248 |
104
- | 0.1865 | 8.16 | 10200 | 0.1037 | 0.5587 | 0.4317 |
105
- | 0.1653 | 8.4 | 10500 | 0.1029 | 0.5661 | 0.4231 |
106
- | 0.1819 | 8.64 | 10800 | 0.1063 | 0.5822 | 0.4375 |
107
- | 0.1717 | 8.88 | 11100 | 0.1029 | 0.5710 | 0.4228 |
108
- | 0.159 | 9.12 | 11400 | 0.1043 | 0.5892 | 0.4323 |
109
- | 0.1444 | 9.36 | 11700 | 0.1047 | 0.5768 | 0.4346 |
110
- | 0.1449 | 9.6 | 12000 | 0.1059 | 0.5714 | 0.4371 |
111
- | 0.1491 | 9.84 | 12300 | 0.1081 | 0.5478 | 0.4416 |
112
 
113
 
114
  ### Framework versions
 
23
  metrics:
24
  - name: Wer
25
  type: wer
26
+ value: 0.3611633875106929
27
  ---
28
 
29
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
33
 
34
  This model is a fine-tuned version of [kiranpantha/w2v-bert-2.0-nepali](https://huggingface.co/kiranpantha/w2v-bert-2.0-nepali) on the kiranpantha/OpenSLR54-Balanced-Nepali dataset.
35
  It achieves the following results on the evaluation set:
36
+ - Loss: 0.3414
37
+ - Wer: 0.3612
38
+ - Cer: 0.0805
39
 
40
  ## Model description
41
 
 
61
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
62
  - lr_scheduler_type: linear
63
  - lr_scheduler_warmup_steps: 500
64
+ - num_epochs: 2
65
  - mixed_precision_training: Native AMP
66
 
67
  ### Training results
68
 
69
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
70
+ |:-------------:|:------:|:----:|:---------------:|:------:|:------:|
71
+ | 0.4176 | 0.24 | 300 | 0.3260 | 0.3485 | 0.0772 |
72
+ | 0.4128 | 0.48 | 600 | 0.3514 | 0.3620 | 0.0810 |
73
+ | 0.4161 | 0.72 | 900 | 0.3460 | 0.3618 | 0.0810 |
74
+ | 0.3578 | 0.96 | 1200 | 0.3366 | 0.3528 | 0.0804 |
75
+ | 0.359 | 1.2 | 1500 | 0.3595 | 0.3577 | 0.0787 |
76
+ | 0.3371 | 1.44 | 1800 | 0.3446 | 0.3634 | 0.0808 |
77
+ | 0.3309 | 1.6800 | 2100 | 0.3399 | 0.3677 | 0.0818 |
78
+ | 0.3441 | 1.92 | 2400 | 0.3414 | 0.3612 | 0.0805 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
79
 
80
 
81
  ### Framework versions
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:35b4c6328be70f2552c2b5afa66c115af519463d47cdae142fe6d23fab680bc7
3
  size 2423081060
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:76d49df1a774d4abdb525f44a8af8069a27d8af517e38280ad6e6c35ee0e1476
3
  size 2423081060
runs/Oct26_21-08-08_ml/events.out.tfevents.1729956266.ml.4821.1 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4281eb0ff6a4b5a8c1c21d06f2e6da83092e86ac41da05b40f0483d491a88fd4
3
- size 10913
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a9de3faa3973b8d18cda4a1fad35d7ae1445b17e19d32249bd7c98333bd2dbc
3
+ size 11267