File size: 2,563 Bytes
1458ece
2a26cc1
1458ece
 
 
 
 
a0d0d95
1458ece
 
 
 
 
 
 
 
 
a0d0d95
 
1458ece
 
 
 
 
 
a0d0d95
1458ece
 
 
 
 
 
 
a0d0d95
1458ece
a0d0d95
 
1458ece
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
217edc5
 
1458ece
2a26cc1
 
1458ece
 
 
2a26cc1
1458ece
 
 
 
 
 
2a26cc1
 
 
 
 
 
 
 
 
 
1458ece
 
 
 
2a26cc1
1458ece
 
 
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
---

library_name: transformers
license: apache-2.0
base_model: openai/whisper-large-v3
tags:
- generated_from_trainer
datasets:
- fsicoli/common_voice_18_0
metrics:
- wer
model-index:
- name: whisper-large-v3-pt-3000h-3
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: fsicoli/common_voice_18_0 pt
      type: fsicoli/common_voice_18_0
      config: pt
      split: None
      args: pt
    metrics:
    - name: Wer
      type: wer
      value: 0.10736707238949392
---


<!-- 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-large-v3-pt-3000h-3

This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the fsicoli/common_voice_18_0 pt dataset.

It achieves the following results on the evaluation set:

- Loss: 0.1501

- Wer: 0.1074



## 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
- gradient_accumulation_steps: 4
- total_train_batch_size: 32

- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08

- lr_scheduler_type: linear

- lr_scheduler_warmup_steps: 1000
- num_epochs: 10.0

- mixed_precision_training: Native AMP



### Training results



| Training Loss | Epoch  | Step | Validation Loss | Wer    |

|:-------------:|:------:|:----:|:---------------:|:------:|

| 0.1388        | 0.9996 | 691  | 0.1501          | 0.1074 |

| 0.108         | 1.9993 | 1382 | 0.1619          | 0.1153 |

| 0.091         | 2.9989 | 2073 | 0.1697          | 0.1124 |

| 0.0461        | 4.0    | 2765 | 0.1764          | 0.1120 |

| 0.0264        | 4.9996 | 3456 | 0.2024          | 0.1133 |

| 0.0203        | 5.9993 | 4147 | 0.2200          | 0.1099 |

| 0.0129        | 6.9989 | 4838 | 0.2277          | 0.1114 |

| 0.0091        | 8.0    | 5530 | 0.2552          | 0.1067 |

| 0.0063        | 8.9996 | 6221 | 0.2565          | 0.1054 |

| 0.0019        | 9.9964 | 6910 | 0.2671          | 0.1042 |





### Framework versions



- Transformers 4.45.0.dev0

- Pytorch 2.4.0+cu124

- Datasets 2.18.1.dev0

- Tokenizers 0.19.1