Hugging Face's logo

Samuael
/
tigrinya-asr-characters

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2
Generated from Trainer
Eval Results
Model card Files Community
Samuael commited on
Commit
8a636cd
·
verified ·
1 Parent(s): ad2f8fd

Samuael/tigrinya-asr-characters

Browse files
Files changed (3) hide show
  1. README.md +29 -20
  2. model.safetensors +1 -1
  3. training_args.bin +1 -1
README.md CHANGED
@@ -23,7 +23,7 @@ model-index:
23
  metrics:
24
  - name: Wer
25
  type: wer
26
- value: 1.0
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 [Samuael/asr-amharic-phoneme-based-37-6](https://huggingface.co/Samuael/asr-amharic-phoneme-based-37-6) on the alffa_amharic dataset.
35
  It achieves the following results on the evaluation set:
36
- - Loss: 5.7515
37
- - Wer: 1.0
38
- - Phoneme Cer: 1.0
39
 
40
  ## Model description
41
 
@@ -54,31 +54,40 @@ More information needed
54
  ### Training hyperparameters
55
 
56
  The following hyperparameters were used during training:
57
- - learning_rate: 3e-05
58
  - train_batch_size: 32
59
  - eval_batch_size: 8
60
  - seed: 42
61
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
62
  - lr_scheduler_type: linear
63
  - lr_scheduler_warmup_steps: 100
64
- - num_epochs: 3
65
 
66
  ### Training results
67
 
68
- | Training Loss | Epoch | Step | Validation Loss | Wer | Phoneme Cer |
69
- |:-------------:|:------:|:----:|:---------------:|:---:|:-----------:|
70
- | 6.9297 | 0.2312 | 200 | 8.0879 | 1.0 | 1.0 |
71
- | 6.3204 | 0.4624 | 400 | 7.6363 | 1.0 | 1.0 |
72
- | 6.1425 | 0.6936 | 600 | 7.2266 | 1.0 | 1.0 |
73
- | 5.936 | 0.9249 | 800 | 6.8506 | 1.0 | 1.0 |
74
- | 5.5689 | 1.1561 | 1000 | 6.7441 | 1.0 | 1.0 |
75
- | 5.5054 | 1.3873 | 1200 | 6.4326 | 1.0 | 1.0 |
76
- | 5.5138 | 1.6185 | 1400 | 6.3467 | 1.0 | 1.0 |
77
- | 5.4229 | 1.8497 | 1600 | 6.1142 | 1.0 | 1.0 |
78
- | 5.1684 | 2.0809 | 1800 | 6.0052 | 1.0 | 1.0 |
79
- | 5.1601 | 2.3121 | 2000 | 5.9630 | 1.0 | 1.0 |
80
- | 5.1866 | 2.5434 | 2200 | 5.9521 | 1.0 | 1.0 |
81
- | 5.0475 | 2.7746 | 2400 | 5.7515 | 1.0 | 1.0 |
 
 
 
 
 
 
 
 
 
82
 
83
 
84
  ### Framework versions
 
23
  metrics:
24
  - name: Wer
25
  type: wer
26
+ value: 1.0118854673149649
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 [Samuael/asr-amharic-phoneme-based-37-6](https://huggingface.co/Samuael/asr-amharic-phoneme-based-37-6) on the alffa_amharic dataset.
35
  It achieves the following results on the evaluation set:
36
+ - Loss: 4.2289
37
+ - Wer: 1.0119
38
+ - Phoneme Cer: 0.7449
39
 
40
  ## Model description
41
 
 
54
  ### Training hyperparameters
55
 
56
  The following hyperparameters were used during training:
57
+ - learning_rate: 0.0001
58
  - train_batch_size: 32
59
  - eval_batch_size: 8
60
  - seed: 42
61
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
62
  - lr_scheduler_type: linear
63
  - lr_scheduler_warmup_steps: 100
64
+ - num_epochs: 5
65
 
66
  ### Training results
67
 
68
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Phoneme Cer |
69
+ |:-------------:|:------:|:----:|:---------------:|:------:|:-----------:|
70
+ | 5.0828 | 0.2312 | 200 | 5.5389 | 1.0 | 1.0 |
71
+ | 4.9657 | 0.4624 | 400 | 5.2934 | 1.0 | 1.0 |
72
+ | 4.9164 | 0.6936 | 600 | 5.1868 | 1.0 | 1.0 |
73
+ | 4.8373 | 0.9249 | 800 | 5.0902 | 1.0 | 1.0 |
74
+ | 4.6232 | 1.1561 | 1000 | 5.0358 | 1.0 | 1.0 |
75
+ | 4.6574 | 1.3873 | 1200 | 4.9342 | 1.0 | 1.0 |
76
+ | 4.5529 | 1.6185 | 1400 | 4.8864 | 1.0 | 1.0 |
77
+ | 4.5468 | 1.8497 | 1600 | 4.8392 | 1.0 | 1.0 |
78
+ | 4.4804 | 2.0809 | 1800 | 4.8041 | 1.0 | 1.0 |
79
+ | 4.4761 | 2.3121 | 2000 | 4.7861 | 1.0 | 1.0 |
80
+ | 4.4033 | 2.5434 | 2200 | 4.7633 | 1.0 | 1.0 |
81
+ | 4.4367 | 2.7746 | 2400 | 4.7418 | 1.0 | 0.9940 |
82
+ | 4.4293 | 3.0058 | 2600 | 4.6574 | 1.0 | 0.9874 |
83
+ | 4.4015 | 3.2370 | 2800 | 4.6215 | 1.0 | 0.8930 |
84
+ | 4.323 | 3.4682 | 3000 | 4.5763 | 1.0005 | 0.8455 |
85
+ | 4.3458 | 3.6994 | 3200 | 4.5150 | 1.0005 | 0.8335 |
86
+ | 4.2924 | 3.9306 | 3400 | 4.4576 | 1.0 | 0.8174 |
87
+ | 4.2297 | 4.1618 | 3600 | 4.3967 | 1.0016 | 0.7924 |
88
+ | 4.1311 | 4.3931 | 3800 | 4.3129 | 1.0011 | 0.7879 |
89
+ | 4.1137 | 4.6243 | 4000 | 4.2518 | 1.0043 | 0.7616 |
90
+ | 4.1176 | 4.8555 | 4200 | 4.2289 | 1.0119 | 0.7449 |
91
 
92
 
93
  ### Framework versions
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:54baa787d28e04f42e68feb1c637a4765c247e76398b466bd95a8d1d9aaa581d
3
  size 1262914496
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:448dff353a39a6884a41d5ba9ffdd6ef365280ceff175d7d1d2a063b8b6122dd
3
  size 1262914496
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:00231a882416722da02890ddc208e2472490472986facbbec4a7d9c693df679c
3
  size 5240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2cc413ae8a801f2a5fe12b76c8092fd657e93aa30801902edc81c42a472e8872
3
  size 5240