File size: 2,431 Bytes
1ebfc03
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3511028
1ebfc03
 
 
 
 
 
 
 
 
3511028
 
1ebfc03
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3511028
1ebfc03
 
 
 
 
 
3511028
1ebfc03
 
 
 
 
 
3511028
 
 
 
 
 
 
 
 
 
1ebfc03
 
 
 
 
 
 
 
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
---

library_name: transformers
license: apache-2.0
base_model: ntu-spml/distilhubert
tags:
- generated_from_trainer
datasets:
- marsyas/gtzan
metrics:
- accuracy
model-index:
- name: distilhubert-finetuned-gtzan
  results:
  - task:
      name: Audio Classification
      type: audio-classification
    dataset:
      name: GTZAN
      type: marsyas/gtzan
      config: all
      split: train
      args: all
    metrics:
    - name: Accuracy
      type: accuracy
      value: 0.7692307692307693
---


<!-- 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. -->

# distilhubert-finetuned-gtzan

This model is a fine-tuned version of [ntu-spml/distilhubert](https://huggingface.co/ntu-spml/distilhubert) on the GTZAN dataset.
It achieves the following results on the evaluation set:
- Loss: 1.2256
- Accuracy: 0.7692

## 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: 9e-05

- train_batch_size: 10

- eval_batch_size: 10

- seed: 42

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

- lr_scheduler_type: linear

- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 18

- mixed_precision_training: Native AMP



### Training results



| Training Loss | Epoch | Step | Validation Loss | Accuracy |

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

| 1.9893        | 1.0   | 70   | 1.9671          | 0.4615   |

| 1.1923        | 2.0   | 140  | 1.3634          | 0.5256   |

| 1.1937        | 3.0   | 210  | 1.0865          | 0.6154   |

| 0.5684        | 4.0   | 280  | 0.9352          | 0.6795   |

| 0.4571        | 5.0   | 350  | 0.7889          | 0.7564   |

| 0.1854        | 6.0   | 420  | 0.8209          | 0.7308   |

| 0.0688        | 7.0   | 490  | 0.9835          | 0.7692   |

| 0.087         | 8.0   | 560  | 1.1710          | 0.7179   |

| 0.0109        | 9.0   | 630  | 1.0900          | 0.7692   |

| 0.0049        | 10.0  | 700  | 1.2256          | 0.7692   |





### Framework versions



- Transformers 4.45.1

- Pytorch 2.4.1+cu121

- Datasets 3.0.1

- Tokenizers 0.20.0