File size: 1,464 Bytes
1afdbfe
 
 
 
 
 
 
 
bdd5733
1afdbfe
bdd5733
9599246
1afdbfe
 
 
 
 
bdd5733
1afdbfe
bdd5733
1afdbfe
9599246
 
 
 
 
 
 
 
 
 
1afdbfe
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language:
- hi
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
datasets:
- konnakol
model-index:
- name: Whisper Hi - Gopika
  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. -->

# Whisper Hi - Gopika

This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the Konnakol dataset.
It achieves the following results on the evaluation set:
- eval_loss: 0.2581
- eval_wer: 0.0278
- eval_cer: 0.0297
- eval_half_cycle_error: 0.0278
- eval_full_cycle_error: 0.0532
- eval_runtime: 10.6512
- eval_samples_per_second: 2.159
- eval_steps_per_second: 0.188
- epoch: 125.0
- step: 750

## 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: 32
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 1000
- mixed_precision_training: Native AMP

### Framework versions

- Transformers 4.41.1
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1