README.md

metadata

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: 0.85

distilhubert-finetuned-gtzan

This model is a fine-tuned version of ntu-spml/distilhubert on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 0.5569
• Accuracy: 0.85

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: 8e-05
• train_batch_size: 6
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 192
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 60

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.85	4	2.2836	0.14
2.2984	1.92	9	2.2574	0.18
2.2856	2.99	14	2.2060	0.32
2.2478	3.84	18	2.1331	0.37
2.1775	4.91	23	1.9859	0.47
2.0557	5.97	28	1.8086	0.52
1.8764	6.83	32	1.6783	0.53
1.7133	7.89	37	1.5235	0.54
1.5661	8.96	42	1.4048	0.58
1.4544	9.81	46	1.3279	0.6
1.3365	10.88	51	1.2591	0.67
1.2228	11.95	56	1.1587	0.7
1.1298	12.8	60	1.1476	0.68
1.0601	13.87	65	1.0066	0.77
0.9886	14.93	70	0.9855	0.76
0.923	16.0	75	0.9767	0.73
0.923	16.85	79	0.8896	0.79
0.8539	17.92	84	0.8421	0.78
0.788	18.99	89	0.8270	0.8
0.7253	19.84	93	0.7764	0.82
0.6523	20.91	98	0.6998	0.85
0.5853	21.97	103	0.6891	0.87
0.5372	22.83	107	0.7106	0.8
0.4815	23.89	112	0.6542	0.82
0.4461	24.96	117	0.6136	0.87
0.3841	25.81	121	0.6338	0.81
0.3505	26.88	126	0.6082	0.87
0.3143	27.95	131	0.5776	0.88
0.2913	28.8	135	0.5833	0.86
0.2519	29.87	140	0.5543	0.89
0.2234	30.93	145	0.5606	0.84
0.1994	32.0	150	0.5726	0.86
0.1994	32.85	154	0.5391	0.86
0.1789	33.92	159	0.5908	0.83
0.1615	34.99	164	0.5498	0.85
0.1444	35.84	168	0.5389	0.85
0.1303	36.91	173	0.5829	0.84
0.1192	37.97	178	0.5278	0.87
0.1074	38.83	182	0.6011	0.83
0.1001	39.89	187	0.5260	0.87
0.0935	40.96	192	0.5778	0.84
0.0885	41.81	196	0.5563	0.86
0.0827	42.88	201	0.5556	0.86
0.0785	43.95	206	0.5807	0.84
0.0767	44.8	210	0.5649	0.85
0.0722	45.87	215	0.5551	0.85
0.0718	46.93	220	0.5432	0.86
0.0701	48.0	225	0.5720	0.85
0.0701	48.85	229	0.5695	0.85
0.068	49.92	234	0.5642	0.85
0.0673	50.99	239	0.5571	0.85
0.0672	51.2	240	0.5569	0.85

Framework versions

• Transformers 4.32.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.14.4
• Tokenizers 0.13.3