File size: 2,552 Bytes
3cc38a5
 
 
e294886
3cc38a5
4c20bb1
a88066b
 
3cc38a5
e294886
3cc38a5
 
072353a
3cc38a5
e294886
3cc38a5
 
 
072353a
3cc38a5
07c1722
e294886
a88066b
3cc38a5
072353a
f348567
072353a
3cc38a5
 
 
 
 
e294886
3cc38a5
e294886
3cc38a5
674d21a
 
3cc38a5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b914315
3cc38a5
 
 
 
 
 
 
674d21a
 
3cc38a5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
674d21a
 
3cc38a5
 
 
 
 
 
 
 
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
---
license: apache-2.0
tags:
- whisper-event
- generated_from_trainer
- hf-asr-leaderboard
- pashto
- ps
datasets:
- google/fleurs
metrics:
- wer
base_model: openai/whisper-medium
model-index:
- name: Whisper Medium Pashto
  results:
  - task:
      type: automatic-speech-recognition
      name: Automatic Speech Recognition
    dataset:
      name: google/fleurs
      type: google/fleurs
      args: 'config: ps_af, split: test'
    metrics:
    - type: wer
      value: 49.4855
      name: Wer
---

<!-- 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 Medium Pashto

This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the google/fleurs ps_af dataset.
It achieves the following results on the evaluation set:
- Loss: 1.4807
- Wer: 50.5448

## 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: 3e-07
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 10
- training_steps: 1200
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Wer     |
|:-------------:|:------:|:----:|:---------------:|:-------:|
| 0.0334        | 14.29  | 100  | 1.0348          | 50.0908 |
| 0.0021        | 28.57  | 200  | 1.1971          | 49.4855 |
| 0.0007        | 42.86  | 300  | 1.2651          | 49.7352 |
| 0.0006        | 57.14  | 400  | 1.3084          | 49.9697 |
| 0.0005        | 71.43  | 500  | 1.3479          | 50.0605 |
| 0.0004        | 85.71  | 600  | 1.3835          | 50.3027 |
| 0.0004        | 100.0  | 700  | 1.4139          | 50.4540 |
| 0.0004        | 114.29 | 800  | 1.4382          | 50.4616 |
| 0.0004        | 128.57 | 900  | 1.4545          | 50.5297 |
| 0.0003        | 142.86 | 1000 | 1.4603          | 50.5675 |
| 0.0003        | 157.14 | 1100 | 1.4750          | 50.5599 |
| 0.0003        | 171.43 | 1200 | 1.4807          | 50.5448 |


### Framework versions

- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2