deepdml commited on
Commit
55acf7f
·
1 Parent(s): 8fa81c5

End of training

Browse files
Files changed (1) hide show
  1. README.md +84 -0
README.md ADDED
@@ -0,0 +1,84 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - ig
4
+ license: apache-2.0
5
+ base_model: openai/whisper-tiny
6
+ tags:
7
+ - generated_from_trainer
8
+ datasets:
9
+ - deepdml/igbo-dict-expansion-16khz
10
+ - deepdml/igbo-dict-16khz
11
+ metrics:
12
+ - wer
13
+ model-index:
14
+ - name: Whisper Tiny ig
15
+ results:
16
+ - task:
17
+ name: Automatic Speech Recognition
18
+ type: automatic-speech-recognition
19
+ dataset:
20
+ name: google/fleurs
21
+ type: deepdml/igbo-dict-expansion-16khz
22
+ config: ig_ng
23
+ split: test
24
+ args: ig_ng
25
+ metrics:
26
+ - name: Wer
27
+ type: wer
28
+ value: 89.15504591613625
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 Tiny ig
35
+
36
+ This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the google/fleurs dataset.
37
+ It achieves the following results on the evaluation set:
38
+ - Loss: 2.8423
39
+ - Wer: 89.1550
40
+ - Cer: 51.2543
41
+
42
+ ## Model description
43
+
44
+ More information needed
45
+
46
+ ## Intended uses & limitations
47
+
48
+ More information needed
49
+
50
+ ## Training and evaluation data
51
+
52
+ More information needed
53
+
54
+ ## Training procedure
55
+
56
+ ### Training hyperparameters
57
+
58
+ The following hyperparameters were used during training:
59
+ - learning_rate: 1e-05
60
+ - train_batch_size: 128
61
+ - eval_batch_size: 128
62
+ - seed: 42
63
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
64
+ - lr_scheduler_type: linear
65
+ - lr_scheduler_warmup_ratio: 0.04
66
+ - training_steps: 5000
67
+
68
+ ### Training results
69
+
70
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
71
+ |:-------------:|:-------:|:----:|:---------------:|:-------:|:-------:|
72
+ | 0.2126 | 2.0344 | 1000 | 3.7178 | 86.3321 | 49.8998 |
73
+ | 0.1179 | 4.0688 | 2000 | 2.8735 | 87.9452 | 52.9006 |
74
+ | 0.0729 | 7.0204 | 3000 | 2.6541 | 86.7208 | 50.7179 |
75
+ | 0.0519 | 9.0548 | 4000 | 2.7895 | 89.4806 | 51.4953 |
76
+ | 0.0331 | 12.0064 | 5000 | 2.8423 | 89.1550 | 51.2543 |
77
+
78
+
79
+ ### Framework versions
80
+
81
+ - Transformers 4.42.0.dev0
82
+ - Pytorch 2.3.0+cu121
83
+ - Datasets 2.19.1
84
+ - Tokenizers 0.19.1