File size: 2,914 Bytes
e9a1552
c8dc207
 
e9a1552
 
 
 
f13b86d
 
e9a1552
 
f13b86d
 
 
 
 
 
 
 
 
 
 
 
 
7a39fe1
e9a1552
 
 
 
 
 
 
c8dc207
f13b86d
7a39fe1
 
e9a1552
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7a39fe1
c8dc207
 
e9a1552
 
 
 
7a39fe1
e9a1552
 
f13b86d
 
 
 
7a39fe1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f13b86d
 
e9a1552
 
 
c8dc207
 
e9a1552
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
91
92
93
94
95
96
97
---
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: 0.82
---

<!-- 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: 0.6623
- Accuracy: 0.82

## 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: 3e-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_ratio: 0.1
- num_epochs: 20
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 2.2457        | 1.0   | 113  | 2.1827          | 0.33     |
| 1.8385        | 2.0   | 226  | 1.6935          | 0.61     |
| 1.46          | 3.0   | 339  | 1.4282          | 0.63     |
| 1.1508        | 4.0   | 452  | 1.1055          | 0.7      |
| 0.9972        | 5.0   | 565  | 0.8945          | 0.74     |
| 0.7826        | 6.0   | 678  | 0.7784          | 0.77     |
| 0.6802        | 7.0   | 791  | 0.7184          | 0.8      |
| 0.4635        | 8.0   | 904  | 0.7725          | 0.76     |
| 0.3746        | 9.0   | 1017 | 0.5875          | 0.84     |
| 0.264         | 10.0  | 1130 | 0.7612          | 0.75     |
| 0.1995        | 11.0  | 1243 | 0.6099          | 0.81     |
| 0.135         | 12.0  | 1356 | 0.6306          | 0.81     |
| 0.0974        | 13.0  | 1469 | 0.5947          | 0.83     |
| 0.0563        | 14.0  | 1582 | 0.7485          | 0.8      |
| 0.0443        | 15.0  | 1695 | 0.6977          | 0.79     |
| 0.0565        | 16.0  | 1808 | 0.6331          | 0.83     |
| 0.0295        | 17.0  | 1921 | 0.6538          | 0.82     |
| 0.0178        | 18.0  | 2034 | 0.6977          | 0.82     |
| 0.0191        | 19.0  | 2147 | 0.6453          | 0.83     |
| 0.0147        | 20.0  | 2260 | 0.6623          | 0.82     |


### Framework versions

- Transformers 4.41.2
- Pytorch 2.1.2
- Datasets 2.19.2
- Tokenizers 0.19.1