Hugging Face's logo

kiranpantha
/
w2v-bert-2.0-nepali-unlabeled-4

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Nepali
wav2vec2-bert
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
w2v-bert-2.0-nepali-unlabeled-4
File size: 2,427 Bytes 
7844992
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
---
library_name: transformers
language:
- ne
license: mit
base_model: facebook/w2v-bert-2.0
tags:
- generated_from_trainer
datasets:
- kiranpantha/OpenSLR54-Balanced-Nepali
metrics:
- wer
model-index:
- name: Wave2Vec2-Bert2.0 - Kiran Pantha
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: kiranpantha/OpenSLR54-Balanced-Nepali
      type: kiranpantha/OpenSLR54-Balanced-Nepali
      args: 'config: ne, split: train,test'
    metrics:
    - name: Wer
      type: wer
      value: 0.45406330196749356
---

<!-- 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. -->

# Wave2Vec2-Bert2.0 - Kiran Pantha

This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co./facebook/w2v-bert-2.0) on the kiranpantha/OpenSLR54-Balanced-Nepali dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5896
- Wer: 0.4541
- Cer: 0.1135

## 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: 5e-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: 500
- num_epochs: 2
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch  | Step | Validation Loss | Wer    | Cer    |
|:-------------:|:------:|:----:|:---------------:|:------:|:------:|
| 0.1466        | 0.24   | 300  | 0.5865          | 0.4662 | 0.1138 |
| 0.2372        | 0.48   | 600  | 0.6488          | 0.5092 | 0.1314 |
| 0.2822        | 0.72   | 900  | 0.5943          | 0.4713 | 0.1179 |
| 0.2517        | 0.96   | 1200 | 0.5902          | 0.4876 | 0.1252 |
| 0.1943        | 1.2    | 1500 | 0.6223          | 0.4814 | 0.1228 |
| 0.1776        | 1.44   | 1800 | 0.5888          | 0.4566 | 0.1155 |
| 0.1567        | 1.6800 | 2100 | 0.6007          | 0.4612 | 0.1154 |
| 0.1495        | 1.92   | 2400 | 0.5896          | 0.4541 | 0.1135 |


### Framework versions

- Transformers 4.44.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.19.1