File size: 2,689 Bytes
e8cc43a
 
 
 
 
 
 
 
72d743a
2635931
e8cc43a
 
 
 
 
 
 
 
 
9bed6a4
72d743a
2635931
e8cc43a
 
 
 
fdac13c
e8cc43a
 
 
 
 
 
 
4b7bb56
91f02f4
fdac13c
 
e8cc43a
91f02f4
 
 
e8cc43a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fdac13c
 
e8cc43a
fdac13c
 
e8cc43a
 
fdac13c
e8cc43a
 
 
fdac13c
e8cc43a
 
 
 
fdac13c
 
 
 
 
 
 
 
 
 
 
 
e8cc43a
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
---
language:
- te
license: apache-2.0
tags:
- whisper-event
- generated_from_trainer
datasets:
- openslr
- google/fleurs
metrics:
- wer
model-index:
- name: whisper-small-telugu-large-data
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: google/fleurs
      type: openslr
      config: te_in
      split: None
    metrics:
    - name: Wer
      type: wer
      value: 38.84604916991744
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# whisper-small-telugu-large-data

This [model](steja/whisper-small-telugu-large-data) is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the google/fleurs and openslr dataset in telugu.
It achieves the following results on the evaluation set (google/fleurs, test set):
- Loss: 0.3310
- Wer: 38.8460

[openai/whisper-small](https://huggingface.co/openai/whisper-small) has the following zero shot performance on google/fleurs test set:
- Wer: 117.91

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 4
- eval_batch_size: 16
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer     |
|:-------------:|:-----:|:----:|:---------------:|:-------:|
| 0.128         | 2.27  | 500  | 0.2015          | 45.1692 |
| 0.0462        | 4.55  | 1000 | 0.1877          | 41.1050 |
| 0.0184        | 6.82  | 1500 | 0.2241          | 40.5153 |
| 0.0045        | 9.09  | 2000 | 0.2590          | 39.7260 |
| 0.0019        | 11.36 | 2500 | 0.2824          | 39.0819 |
| 0.0006        | 13.64 | 3000 | 0.3002          | 38.9096 |
| 0.0002        | 15.91 | 3500 | 0.3141          | 38.5920 |
| 0.0001        | 18.18 | 4000 | 0.3232          | 38.7463 |
| 0.0001        | 20.45 | 4500 | 0.3289          | 38.8370 |
| 0.0001        | 22.73 | 5000 | 0.3310          | 38.8460 |


### Framework versions

- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2