ckandemir commited on
Commit
b534a34
1 Parent(s): 0429cf7

End of training

Browse files
README.md CHANGED
@@ -2,41 +2,31 @@
2
  language:
3
  - tr
4
  license: apache-2.0
 
5
  tags:
6
  - hf-asr-leaderboard
7
  - generated_from_trainer
8
  datasets:
9
  - mozilla-foundation/common_voice_11_0
10
- metrics:
11
- - wer
12
- base_model: openai/whisper-small
13
  model-index:
14
- - name: Whisper Small Tr - Canberk Kandemir
15
- results:
16
- - task:
17
- type: automatic-speech-recognition
18
- name: Automatic Speech Recognition
19
- dataset:
20
- name: Common Voice 11.0
21
- type: mozilla-foundation/common_voice_11_0
22
- config: tr
23
- split: None
24
- args: 'config: tr, split: test'
25
- metrics:
26
- - type: wer
27
- value: 43.06339873086104
28
- name: Wer
29
  ---
30
 
31
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
32
  should probably proofread and complete it, then remove this comment. -->
33
 
34
- # Whisper Small Tr
35
 
36
  This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 11.0 dataset.
37
  It achieves the following results on the evaluation set:
38
- - Loss: 0.5432
39
- - Wer: 43.0634
 
 
 
 
 
40
 
41
  ## Model description
42
 
@@ -55,37 +45,21 @@ More information needed
55
  ### Training hyperparameters
56
 
57
  The following hyperparameters were used during training:
58
- - learning_rate: 7e-05
59
- - train_batch_size: 16
60
- - eval_batch_size: 8
61
  - seed: 42
62
  - gradient_accumulation_steps: 2
63
- - total_train_batch_size: 32
64
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
65
  - lr_scheduler_type: linear
66
- - lr_scheduler_warmup_steps: 1000
67
- - training_steps: 5000
68
  - mixed_precision_training: Native AMP
69
 
70
- ### Training results
71
-
72
- | Training Loss | Epoch | Step | Validation Loss | Wer |
73
- |:-------------:|:-----:|:----:|:---------------:|:-------:|
74
- | 0.4399 | 0.44 | 500 | 0.6307 | 61.0351 |
75
- | 0.4322 | 0.89 | 1000 | 0.6820 | 58.6909 |
76
- | 0.2857 | 1.33 | 1500 | 0.6496 | 54.3867 |
77
- | 0.2839 | 1.77 | 2000 | 0.6088 | 49.6497 |
78
- | 0.1467 | 2.21 | 2500 | 0.5813 | 47.3346 |
79
- | 0.1268 | 2.66 | 3000 | 0.5647 | 46.1315 |
80
- | 0.0711 | 3.1 | 3500 | 0.5532 | 44.8196 |
81
- | 0.0658 | 3.54 | 4000 | 0.5444 | 43.4670 |
82
- | 0.0601 | 3.99 | 4500 | 0.5372 | 43.4146 |
83
- | 0.0304 | 4.43 | 5000 | 0.5432 | 43.0634 |
84
-
85
-
86
  ### Framework versions
87
 
88
- - Transformers 4.37.2
89
  - Pytorch 2.1.0+cu121
90
- - Datasets 2.17.0
91
  - Tokenizers 0.15.2
 
2
  language:
3
  - tr
4
  license: apache-2.0
5
+ base_model: openai/whisper-small
6
  tags:
7
  - hf-asr-leaderboard
8
  - generated_from_trainer
9
  datasets:
10
  - mozilla-foundation/common_voice_11_0
 
 
 
11
  model-index:
12
+ - name: Whisper Small Tr - CK
13
+ results: []
 
 
 
 
 
 
 
 
 
 
 
 
 
14
  ---
15
 
16
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
  should probably proofread and complete it, then remove this comment. -->
18
 
19
+ # Whisper Small Tr - CK
20
 
21
  This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 11.0 dataset.
22
  It achieves the following results on the evaluation set:
23
+ - eval_loss: 0.5128
24
+ - eval_wer: 44.7692
25
+ - eval_runtime: 2371.4188
26
+ - eval_samples_per_second: 4.277
27
+ - eval_steps_per_second: 0.475
28
+ - epoch: 2.99
29
+ - step: 3000
30
 
31
  ## Model description
32
 
 
45
  ### Training hyperparameters
46
 
47
  The following hyperparameters were used during training:
48
+ - learning_rate: 1e-05
49
+ - train_batch_size: 18
50
+ - eval_batch_size: 9
51
  - seed: 42
52
  - gradient_accumulation_steps: 2
53
+ - total_train_batch_size: 36
54
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
55
  - lr_scheduler_type: linear
56
+ - lr_scheduler_warmup_steps: 500
57
+ - training_steps: 4000
58
  - mixed_precision_training: Native AMP
59
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
60
  ### Framework versions
61
 
62
+ - Transformers 4.38.2
63
  - Pytorch 2.1.0+cu121
64
+ - Datasets 2.17.1
65
  - Tokenizers 0.15.2
generation_config.json CHANGED
@@ -144,6 +144,7 @@
144
  "<|yo|>": 50325,
145
  "<|zh|>": 50260
146
  },
 
147
  "max_initial_timestamp_index": 50,
148
  "max_length": 448,
149
  "no_timestamps_token_id": 50363,
@@ -240,9 +241,10 @@
240
  50361,
241
  50362
242
  ],
 
243
  "task_to_id": {
244
  "transcribe": 50359,
245
  "translate": 50358
246
  },
247
- "transformers_version": "4.37.2"
248
  }
 
144
  "<|yo|>": 50325,
145
  "<|zh|>": 50260
146
  },
147
+ "language": "<|tr|>",
148
  "max_initial_timestamp_index": 50,
149
  "max_length": 448,
150
  "no_timestamps_token_id": 50363,
 
241
  50361,
242
  50362
243
  ],
244
+ "task": "transcribe",
245
  "task_to_id": {
246
  "transcribe": 50359,
247
  "translate": 50358
248
  },
249
+ "transformers_version": "4.38.2"
250
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e08dc91232c35ead15815381bd2444bce156cc02a3ed5d9e4a4f64f28571f0aa
3
  size 151061672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aecc1eeb4ab8d17ec95542ccafa36fb70f6e5f4115c2ff74e43749a1c4657746
3
  size 151061672
runs/Mar01_06-55-37_b1a84b09baef/events.out.tfevents.1709276146.b1a84b09baef.61975.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d85d3d36cf3da50dc30599bf72d73a39b85feb857fb9c07e118494fe58472eda
3
- size 32598
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba9600fd71ef0685c19499103a69005f37605d2e6c864d6fecb4b066dd931317
3
+ size 36818