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: 0.815

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: 1.2091
• Accuracy: 0.815

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: 12
• eval_batch_size: 12
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2632	1.0	67	2.2116	0.335
1.8978	2.0	134	1.8129	0.5
1.5811	3.0	201	1.4946	0.66
1.1795	4.0	268	1.2851	0.65
1.0256	5.0	335	1.1538	0.66
0.9168	6.0	402	1.0270	0.69
0.9383	7.0	469	0.9349	0.73
0.5988	8.0	536	0.8443	0.795
0.4844	9.0	603	0.8053	0.775
0.422	10.0	670	0.7710	0.785
0.2138	11.0	737	0.7353	0.8
0.1834	12.0	804	0.8303	0.78
0.1789	13.0	871	0.7801	0.805
0.1649	14.0	938	0.8433	0.775
0.0259	15.0	1005	0.7846	0.8
0.0825	16.0	1072	0.9268	0.795
0.0091	17.0	1139	1.0432	0.795
0.0053	18.0	1206	0.9703	0.8
0.0038	19.0	1273	0.9689	0.82
0.0246	20.0	1340	1.0611	0.81
0.0023	21.0	1407	1.0502	0.82
0.0023	22.0	1474	1.0703	0.815
0.0016	23.0	1541	1.0911	0.825
0.0015	24.0	1608	1.1375	0.795
0.0013	25.0	1675	1.1529	0.815
0.0172	26.0	1742	1.1258	0.815
0.0011	27.0	1809	1.1206	0.82
0.001	28.0	1876	1.1492	0.82
0.0009	29.0	1943	1.1490	0.815
0.0008	30.0	2010	1.1527	0.815
0.0008	31.0	2077	1.2008	0.815
0.0638	32.0	2144	1.1685	0.815
0.0007	33.0	2211	1.1749	0.815
0.0858	34.0	2278	1.1683	0.815
0.0006	35.0	2345	1.1772	0.815
0.0007	36.0	2412	1.1801	0.815
0.0006	37.0	2479	1.1956	0.815
0.0006	38.0	2546	1.1937	0.815
0.0055	39.0	2613	1.2110	0.82
0.0006	40.0	2680	1.2023	0.815
0.0006	41.0	2747	1.2093	0.815
0.001	42.0	2814	1.2075	0.815
0.0006	43.0	2881	1.2079	0.815
0.0662	44.0	2948	1.2054	0.815
0.0006	45.0	3015	1.2066	0.815
0.0006	46.0	3082	1.2089	0.815
0.0006	47.0	3149	1.2093	0.815
0.0005	48.0	3216	1.2096	0.815
0.0005	49.0	3283	1.2094	0.815
0.0006	50.0	3350	1.2091	0.815

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0