Hugging Face's logo

artyomboyko
/
distilhubert-finetuned-gtzan

Audio Classification
Transformers
PyTorch
TensorBoard
hubert
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
1
artyomboyko commited on
Commit
200a2cc
•
1 Parent(s): 27d3926

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +15 -108
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: 0.7554
22
- - Accuracy: 0.83
23
 
24
  ## Model description
25
 
@@ -38,119 +38,26 @@ More information needed
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
- - learning_rate: 5e-06
42
- - train_batch_size: 8
43
- - eval_batch_size: 8
44
  - seed: 42
 
 
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
  - lr_scheduler_type: linear
47
  - lr_scheduler_warmup_ratio: 0.1
48
- - num_epochs: 100
49
 
50
  ### Training results
51
 
52
- | Training Loss | Epoch | Step | Validation Loss | Accuracy |
53
- |:-------------:|:-----:|:-----:|:---------------:|:--------:|
54
- | 2.3168 | 1.0 | 113 | 2.2998 | 0.06 |
55
- | 2.2869 | 2.0 | 226 | 2.2880 | 0.15 |
56
- | 2.2711 | 3.0 | 339 | 2.2649 | 0.32 |
57
- | 2.2407 | 4.0 | 452 | 2.2306 | 0.36 |
58
- | 2.1993 | 5.0 | 565 | 2.1723 | 0.41 |
59
- | 2.1239 | 6.0 | 678 | 2.0768 | 0.54 |
60
- | 2.0001 | 7.0 | 791 | 1.9675 | 0.63 |
61
- | 1.9217 | 8.0 | 904 | 1.8527 | 0.63 |
62
- | 1.7955 | 9.0 | 1017 | 1.7538 | 0.62 |
63
- | 1.8026 | 10.0 | 1130 | 1.6549 | 0.63 |
64
- | 1.6166 | 11.0 | 1243 | 1.5623 | 0.68 |
65
- | 1.5788 | 12.0 | 1356 | 1.4857 | 0.67 |
66
- | 1.4279 | 13.0 | 1469 | 1.4114 | 0.64 |
67
- | 1.3744 | 14.0 | 1582 | 1.3545 | 0.7 |
68
- | 1.1651 | 15.0 | 1695 | 1.2848 | 0.67 |
69
- | 1.1956 | 16.0 | 1808 | 1.2487 | 0.69 |
70
- | 1.1852 | 17.0 | 1921 | 1.1826 | 0.73 |
71
- | 1.1153 | 18.0 | 2034 | 1.1325 | 0.73 |
72
- | 1.0218 | 19.0 | 2147 | 1.1174 | 0.71 |
73
- | 1.0623 | 20.0 | 2260 | 1.0428 | 0.73 |
74
- | 0.9385 | 21.0 | 2373 | 1.0111 | 0.75 |
75
- | 0.9395 | 22.0 | 2486 | 0.9853 | 0.75 |
76
- | 0.8799 | 23.0 | 2599 | 0.9868 | 0.72 |
77
- | 0.74 | 24.0 | 2712 | 0.9136 | 0.76 |
78
- | 0.8411 | 25.0 | 2825 | 0.9025 | 0.74 |
79
- | 0.8784 | 26.0 | 2938 | 0.8757 | 0.76 |
80
- | 0.726 | 27.0 | 3051 | 0.8720 | 0.75 |
81
- | 0.7704 | 28.0 | 3164 | 0.8135 | 0.75 |
82
- | 0.7628 | 29.0 | 3277 | 0.7514 | 0.82 |
83
- | 0.6254 | 30.0 | 3390 | 0.7675 | 0.77 |
84
- | 0.5432 | 31.0 | 3503 | 0.7689 | 0.77 |
85
- | 0.5699 | 32.0 | 3616 | 0.7197 | 0.81 |
86
- | 0.5448 | 33.0 | 3729 | 0.6838 | 0.82 |
87
- | 0.5634 | 34.0 | 3842 | 0.7029 | 0.81 |
88
- | 0.4702 | 35.0 | 3955 | 0.7170 | 0.77 |
89
- | 0.3946 | 36.0 | 4068 | 0.6443 | 0.82 |
90
- | 0.4749 | 37.0 | 4181 | 0.6318 | 0.83 |
91
- | 0.317 | 38.0 | 4294 | 0.6420 | 0.83 |
92
- | 0.3082 | 39.0 | 4407 | 0.6190 | 0.82 |
93
- | 0.2932 | 40.0 | 4520 | 0.6196 | 0.83 |
94
- | 0.2928 | 41.0 | 4633 | 0.6059 | 0.83 |
95
- | 0.2902 | 42.0 | 4746 | 0.6290 | 0.82 |
96
- | 0.3297 | 43.0 | 4859 | 0.6039 | 0.82 |
97
- | 0.2645 | 44.0 | 4972 | 0.5924 | 0.83 |
98
- | 0.2586 | 45.0 | 5085 | 0.6134 | 0.83 |
99
- | 0.2815 | 46.0 | 5198 | 0.6237 | 0.81 |
100
- | 0.3678 | 47.0 | 5311 | 0.6001 | 0.81 |
101
- | 0.2904 | 48.0 | 5424 | 0.5742 | 0.84 |
102
- | 0.1635 | 49.0 | 5537 | 0.6265 | 0.82 |
103
- | 0.1144 | 50.0 | 5650 | 0.5945 | 0.81 |
104
- | 0.1677 | 51.0 | 5763 | 0.5986 | 0.83 |
105
- | 0.1879 | 52.0 | 5876 | 0.6099 | 0.83 |
106
- | 0.1977 | 53.0 | 5989 | 0.5745 | 0.84 |
107
- | 0.1255 | 54.0 | 6102 | 0.5959 | 0.82 |
108
- | 0.0789 | 55.0 | 6215 | 0.6409 | 0.83 |
109
- | 0.0634 | 56.0 | 6328 | 0.5985 | 0.82 |
110
- | 0.1688 | 57.0 | 6441 | 0.5848 | 0.84 |
111
- | 0.1464 | 58.0 | 6554 | 0.6173 | 0.83 |
112
- | 0.1089 | 59.0 | 6667 | 0.6245 | 0.83 |
113
- | 0.0963 | 60.0 | 6780 | 0.6343 | 0.82 |
114
- | 0.0548 | 61.0 | 6893 | 0.6277 | 0.83 |
115
- | 0.1293 | 62.0 | 7006 | 0.6128 | 0.83 |
116
- | 0.0406 | 63.0 | 7119 | 0.6339 | 0.83 |
117
- | 0.0532 | 64.0 | 7232 | 0.6480 | 0.83 |
118
- | 0.214 | 65.0 | 7345 | 0.6661 | 0.81 |
119
- | 0.1246 | 66.0 | 7458 | 0.6637 | 0.83 |
120
- | 0.036 | 67.0 | 7571 | 0.6527 | 0.85 |
121
- | 0.1168 | 68.0 | 7684 | 0.6517 | 0.84 |
122
- | 0.0322 | 69.0 | 7797 | 0.6714 | 0.83 |
123
- | 0.0362 | 70.0 | 7910 | 0.6912 | 0.81 |
124
- | 0.1088 | 71.0 | 8023 | 0.6830 | 0.85 |
125
- | 0.0258 | 72.0 | 8136 | 0.7039 | 0.82 |
126
- | 0.0776 | 73.0 | 8249 | 0.6931 | 0.83 |
127
- | 0.0684 | 74.0 | 8362 | 0.6688 | 0.82 |
128
- | 0.0169 | 75.0 | 8475 | 0.6966 | 0.83 |
129
- | 0.1039 | 76.0 | 8588 | 0.6914 | 0.83 |
130
- | 0.0361 | 77.0 | 8701 | 0.6978 | 0.84 |
131
- | 0.0143 | 78.0 | 8814 | 0.7023 | 0.84 |
132
- | 0.0161 | 79.0 | 8927 | 0.7156 | 0.83 |
133
- | 0.0207 | 80.0 | 9040 | 0.7264 | 0.82 |
134
- | 0.0129 | 81.0 | 9153 | 0.7155 | 0.82 |
135
- | 0.0161 | 82.0 | 9266 | 0.7418 | 0.81 |
136
- | 0.0284 | 83.0 | 9379 | 0.7270 | 0.82 |
137
- | 0.0166 | 84.0 | 9492 | 0.7460 | 0.83 |
138
- | 0.0144 | 85.0 | 9605 | 0.7430 | 0.83 |
139
- | 0.0121 | 86.0 | 9718 | 0.7459 | 0.83 |
140
- | 0.0116 | 87.0 | 9831 | 0.7579 | 0.83 |
141
- | 0.0106 | 88.0 | 9944 | 0.7485 | 0.83 |
142
- | 0.0104 | 89.0 | 10057 | 0.7586 | 0.81 |
143
- | 0.0121 | 90.0 | 10170 | 0.7579 | 0.84 |
144
- | 0.01 | 91.0 | 10283 | 0.7474 | 0.83 |
145
- | 0.0099 | 92.0 | 10396 | 0.7528 | 0.83 |
146
- | 0.0117 | 93.0 | 10509 | 0.7603 | 0.82 |
147
- | 0.0174 | 94.0 | 10622 | 0.7646 | 0.83 |
148
- | 0.0103 | 95.0 | 10735 | 0.7557 | 0.83 |
149
- | 0.0102 | 96.0 | 10848 | 0.7548 | 0.83 |
150
- | 0.0101 | 97.0 | 10961 | 0.7519 | 0.83 |
151
- | 0.0096 | 98.0 | 11074 | 0.7557 | 0.83 |
152
- | 0.0098 | 99.0 | 11187 | 0.7556 | 0.83 |
153
- | 0.0098 | 100.0 | 11300 | 0.7554 | 0.83 |
154
 
155
 
156
  ### 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: 0.4738
22
+ - Accuracy: 0.88
23
 
24
  ## Model description
25
 
 
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
+ - learning_rate: 4e-06
42
+ - train_batch_size: 4
43
+ - eval_batch_size: 4
44
  - seed: 42
45
+ - gradient_accumulation_steps: 2
46
+ - total_train_batch_size: 8
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: 5
51
 
52
  ### Training results
53
 
54
+ | Training Loss | Epoch | Step | Validation Loss | Accuracy |
55
+ |:-------------:|:-----:|:----:|:---------------:|:--------:|
56
+ | 0.2468 | 1.0 | 112 | 0.4818 | 0.86 |
57
+ | 0.1157 | 2.0 | 225 | 0.4982 | 0.85 |
58
+ | 0.1855 | 3.0 | 337 | 0.4842 | 0.86 |
59
+ | 0.169 | 4.0 | 450 | 0.4762 | 0.88 |
60
+ | 0.1003 | 4.98 | 560 | 0.4738 | 0.88 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
61
 
62
 
63
  ### Framework versions