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: 0.84

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: 0.9085
Accuracy: 0.84

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: 5e-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
training_steps: 3000

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
2.0825	0.88	100	0.47	1.8392
1.4043	1.77	200	0.67	1.2675
1.0686	2.65	300	0.71	1.0186
0.8037	3.54	400	0.74	0.9198
0.6215	4.42	500	0.78	0.7636
0.5106	5.31	600	0.76	0.7937
0.3844	6.19	700	0.78	0.6909
0.3043	7.08	800	0.77	0.7279
0.2453	7.96	900	0.82	0.6447
0.211	8.85	1000	0.84	0.6404
0.2268	9.73	1100	0.77	0.7198
0.1565	10.62	1200	0.83	0.6704
0.0694	11.5	1300	0.83	0.8017
0.0568	12.39	1400	0.8	0.7841
0.0441	13.27	1500	0.81	0.7757
0.0302	14.16	1600	0.84	0.7819
0.0116	15.04	1700	0.83	0.7949
0.0289	15.93	1800	0.85	0.8057
0.0115	16.81	1900	0.83	0.8271
0.0081	17.7	2000	0.86	0.8005
0.0124	18.58	2100	0.8	0.8927
0.0219	19.47	2200	0.85	0.8126
0.0161	20.35	2300	0.85	0.8464
0.0157	21.24	2400	0.86	0.8459
0.0039	22.12	2500	0.8	1.0282
0.0145	23.01	2600	0.9218	0.84
0.0149	23.89	2700	0.9085	0.84

Framework versions

Transformers 4.32.0
Pytorch 1.12.1+cu113
Datasets 2.14.4
Tokenizers 0.13.3