File size: 1,674 Bytes
179ff28
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: w-train-medium-Punjabi-4-0-2000-1e-05-8-8-500
  results: []
---

<!-- 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. -->

# w-train-medium-Punjabi-4-0-2000-1e-05-8-8-500

This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.3046
- Bleu: 29.3147

## 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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 200
- training_steps: 2000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Bleu    |
|:-------------:|:-----:|:----:|:---------------:|:-------:|
| 0.0564        | 3.18  | 500  | 1.0654          | 26.3435 |
| 0.0064        | 6.37  | 1000 | 1.1987          | 29.7507 |
| 0.0009        | 9.55  | 1500 | 1.2901          | 29.5279 |
| 0.0006        | 12.74 | 2000 | 1.3046          | 29.3147 |


### Framework versions

- Transformers 4.35.0
- Pytorch 2.0.0+cu117
- Datasets 2.14.6
- Tokenizers 0.14.1