Hugging Face's logo

peterdamn
/
distilhubert-finetuned-gtzan

Audio Classification
Transformers
PyTorch
hubert
Generated from Trainer
Model card Files Community
1
peterdamn commited on
Commit
67a4912
·
1 Parent(s): 280d82e

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +27 -37
README.md CHANGED
@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
18
 
19
  This model is a fine-tuned version of [ntu-spml/distilhubert](https://huggingface.co/ntu-spml/distilhubert) on the GTZAN dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.9809
22
- - Accuracy: 0.86
23
 
24
  ## Model description
25
 
@@ -38,52 +38,42 @@ More information needed
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
- - learning_rate: 5e-05
42
- - train_batch_size: 1
43
- - eval_batch_size: 1
44
  - seed: 42
45
- - gradient_accumulation_steps: 4
46
  - total_train_batch_size: 4
47
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
  - lr_scheduler_type: linear
49
  - lr_scheduler_warmup_ratio: 0.1
50
- - num_epochs: 30
51
  - mixed_precision_training: Native AMP
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Accuracy |
56
  |:-------------:|:-----:|:----:|:---------------:|:--------:|
57
- | 1.9576 | 1.0 | 224 | 2.0177 | 0.51 |
58
- | 1.4587 | 2.0 | 449 | 1.3450 | 0.6 |
59
- | 0.8217 | 3.0 | 674 | 0.9782 | 0.72 |
60
- | 0.7722 | 4.0 | 899 | 0.6004 | 0.82 |
61
- | 0.3364 | 5.0 | 1123 | 0.5571 | 0.81 |
62
- | 0.3117 | 6.0 | 1348 | 0.6176 | 0.84 |
63
- | 0.129 | 7.0 | 1573 | 0.6289 | 0.84 |
64
- | 0.0062 | 8.0 | 1798 | 0.8338 | 0.83 |
65
- | 0.0158 | 9.0 | 2022 | 0.6889 | 0.86 |
66
- | 0.0034 | 10.0 | 2247 | 0.9704 | 0.87 |
67
- | 0.0016 | 11.0 | 2472 | 0.8214 | 0.86 |
68
- | 0.0011 | 12.0 | 2697 | 0.6679 | 0.9 |
69
- | 0.0008 | 13.0 | 2921 | 0.7900 | 0.87 |
70
- | 0.0006 | 14.0 | 3146 | 0.8782 | 0.87 |
71
- | 0.0006 | 15.0 | 3371 | 0.8960 | 0.85 |
72
- | 0.0004 | 16.0 | 3596 | 0.9461 | 0.85 |
73
- | 0.0004 | 17.0 | 3820 | 0.9201 | 0.85 |
74
- | 0.0003 | 18.0 | 4045 | 0.9782 | 0.84 |
75
- | 0.0004 | 19.0 | 4270 | 0.9895 | 0.86 |
76
- | 0.0003 | 20.0 | 4495 | 0.9889 | 0.84 |
77
- | 0.0003 | 21.0 | 4719 | 0.8960 | 0.86 |
78
- | 0.0003 | 22.0 | 4944 | 1.0281 | 0.85 |
79
- | 0.0002 | 23.0 | 5169 | 0.8940 | 0.86 |
80
- | 0.0002 | 24.0 | 5394 | 0.8929 | 0.86 |
81
- | 0.0002 | 25.0 | 5618 | 0.9515 | 0.86 |
82
- | 0.0002 | 26.0 | 5843 | 0.9025 | 0.85 |
83
- | 0.0002 | 27.0 | 6068 | 0.9667 | 0.86 |
84
- | 0.0002 | 28.0 | 6293 | 0.9715 | 0.86 |
85
- | 0.0002 | 29.0 | 6517 | 0.9674 | 0.86 |
86
- | 0.0002 | 29.9 | 6720 | 0.9809 | 0.86 |
87
 
88
 
89
  ### Framework versions
 
18
 
19
  This model is a fine-tuned version of [ntu-spml/distilhubert](https://huggingface.co/ntu-spml/distilhubert) on the GTZAN dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 0.9739
22
+ - Accuracy: 0.76
23
 
24
  ## Model description
25
 
 
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
+ - learning_rate: 5e-06
42
+ - train_batch_size: 2
43
+ - eval_batch_size: 2
44
  - seed: 42
45
+ - gradient_accumulation_steps: 2
46
  - total_train_batch_size: 4
47
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
  - lr_scheduler_type: linear
49
  - lr_scheduler_warmup_ratio: 0.1
50
+ - num_epochs: 20
51
  - mixed_precision_training: Native AMP
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Accuracy |
56
  |:-------------:|:-----:|:----:|:---------------:|:--------:|
57
+ | 2.2896 | 1.0 | 225 | 2.2805 | 0.25 |
58
+ | 2.1881 | 2.0 | 450 | 2.1587 | 0.49 |
59
+ | 1.9686 | 3.0 | 675 | 1.9439 | 0.54 |
60
+ | 1.8591 | 4.0 | 900 | 1.7870 | 0.58 |
61
+ | 1.7089 | 5.0 | 1125 | 1.6137 | 0.63 |
62
+ | 1.6092 | 6.0 | 1350 | 1.5151 | 0.67 |
63
+ | 1.5813 | 7.0 | 1575 | 1.4201 | 0.68 |
64
+ | 1.3893 | 8.0 | 1800 | 1.3214 | 0.68 |
65
+ | 1.3429 | 9.0 | 2025 | 1.2664 | 0.68 |
66
+ | 1.3234 | 10.0 | 2250 | 1.2207 | 0.67 |
67
+ | 1.4794 | 11.0 | 2475 | 1.1531 | 0.73 |
68
+ | 1.3373 | 12.0 | 2700 | 1.1131 | 0.71 |
69
+ | 1.0693 | 13.0 | 2925 | 1.0817 | 0.73 |
70
+ | 1.14 | 14.0 | 3150 | 1.0618 | 0.72 |
71
+ | 0.8953 | 15.0 | 3375 | 1.0229 | 0.73 |
72
+ | 1.2187 | 16.0 | 3600 | 1.0202 | 0.74 |
73
+ | 1.1009 | 17.0 | 3825 | 0.9971 | 0.76 |
74
+ | 0.9364 | 18.0 | 4050 | 0.9846 | 0.75 |
75
+ | 1.132 | 19.0 | 4275 | 0.9772 | 0.76 |
76
+ | 0.8295 | 20.0 | 4500 | 0.9739 | 0.76 |
 
 
 
 
 
 
 
 
 
 
77
 
78
 
79
  ### Framework versions