sgangireddy commited on
Commit
12ee278
1 Parent(s): 3da2876

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +81 -0
README.md ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - generated_from_trainer
5
+ datasets:
6
+ - multilingual_librispeech
7
+ metrics:
8
+ - wer
9
+ model-index:
10
+ - name: openai/whisper-large-v2
11
+ results:
12
+ - task:
13
+ name: Automatic Speech Recognition
14
+ type: automatic-speech-recognition
15
+ dataset:
16
+ name: multilingual_librispeech
17
+ type: multilingual_librispeech
18
+ config: french
19
+ split: test
20
+ args: french
21
+ metrics:
22
+ - name: Wer
23
+ type: wer
24
+ value: 4.561620226935377
25
+ ---
26
+
27
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
28
+ should probably proofread and complete it, then remove this comment. -->
29
+
30
+ # openai/whisper-large-v2
31
+
32
+ This model is a fine-tuned version of [openai/whisper-large-v2](https://huggingface.co/openai/whisper-large-v2) on the multilingual_librispeech dataset.
33
+ It achieves the following results on the evaluation set:
34
+ - Loss: 0.0903
35
+ - Wer: 4.5616
36
+
37
+ ## Model description
38
+
39
+ More information needed
40
+
41
+ ## Intended uses & limitations
42
+
43
+ More information needed
44
+
45
+ ## Training and evaluation data
46
+
47
+ More information needed
48
+
49
+ ## Training procedure
50
+
51
+ ### Training hyperparameters
52
+
53
+ The following hyperparameters were used during training:
54
+ - learning_rate: 1e-05
55
+ - train_batch_size: 8
56
+ - eval_batch_size: 8
57
+ - seed: 42
58
+ - gradient_accumulation_steps: 8
59
+ - total_train_batch_size: 64
60
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
61
+ - lr_scheduler_type: linear
62
+ - lr_scheduler_warmup_steps: 500
63
+ - training_steps: 4000
64
+ - mixed_precision_training: Native AMP
65
+
66
+ ### Training results
67
+
68
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
69
+ |:-------------:|:-----:|:----:|:---------------:|:------:|
70
+ | 0.1303 | 0.25 | 1000 | 0.1219 | 6.3618 |
71
+ | 0.0751 | 0.5 | 2000 | 0.1033 | 5.3905 |
72
+ | 0.0613 | 0.75 | 3000 | 0.0970 | 4.9193 |
73
+ | 0.0796 | 1.0 | 4000 | 0.0903 | 4.5616 |
74
+
75
+
76
+ ### Framework versions
77
+
78
+ - Transformers 4.26.0.dev0
79
+ - Pytorch 1.13.0+cu117
80
+ - Datasets 2.7.1.dev0
81
+ - Tokenizers 0.13.2