Bagus/ast-finetuned-audioset-10-10-0.4593-finetuned-gtzan-2

ast-finetuned-audioset-10-10-0.4593-finetuned-gtzan

This model is a fine-tuned version of MIT/ast-finetuned-audioset-10-10-0.4593 on the GTZAN dataset. It achieves the following results on the evaluation set:

• Loss: 2.3496
• 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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.9076	1.0	225	0.8234	0.75
0.3522	2.0	450	0.4291	0.9
0.4656	3.0	675	0.4656	0.83
0.2739	4.0	900	0.6314	0.9
0.575	5.0	1125	0.7786	0.85
0.4433	6.0	1350	1.1706	0.88
1.4075	7.0	1575	2.5171	0.83
0.1059	8.0	1800	1.5907	0.84
0.1521	9.0	2025	3.7424	0.72
0.5736	10.0	2250	2.0911	0.82
0.7552	11.0	2475	3.2042	0.81
0.166	12.0	2700	1.8762	0.86
0.0	13.0	2925	1.0614	0.91
1.2229	14.0	3150	3.0105	0.78
1.0135	15.0	3375	2.2024	0.85
0.098	16.0	3600	1.6070	0.87
0.0	17.0	3825	2.5323	0.82
0.0	18.0	4050	2.2202	0.86
0.0	19.0	4275	2.2681	0.85
0.0	20.0	4500	2.2394	0.86
1.8867	21.0	4725	4.2168	0.74
0.6094	22.0	4950	4.7781	0.72
0.3684	23.0	5175	2.6412	0.81
0.0	24.0	5400	2.8745	0.82
0.0	25.0	5625	2.9487	0.79
0.0	26.0	5850	2.5325	0.82
0.1597	27.0	6075	2.0327	0.85
0.0	28.0	6300	3.0062	0.84
0.0	29.0	6525	2.3104	0.8
0.0	30.0	6750	2.9985	0.83
0.0	31.0	6975	2.9385	0.82
0.0	32.0	7200	2.1102	0.87
0.0	33.0	7425	2.0060	0.86
1.1173	34.0	7650	1.9131	0.87
0.0	35.0	7875	2.4819	0.84
0.0	36.0	8100	2.0951	0.87
0.0	37.0	8325	1.9796	0.85
0.0	38.0	8550	2.0940	0.85
0.9059	39.0	8775	2.0714	0.85
0.0	40.0	9000	4.0729	0.75
0.0	41.0	9225	2.8627	0.83
0.0	42.0	9450	4.0389	0.76
0.0	43.0	9675	2.3248	0.85
0.6586	44.0	9900	4.9549	0.75
0.0	45.0	10125	3.3910	0.81
0.0	46.0	10350	3.9627	0.77
0.0	47.0	10575	3.4481	0.83
0.0	48.0	10800	2.7042	0.85
0.0	49.0	11025	2.8337	0.85
0.0	50.0	11250	2.4333	0.85
0.0	51.0	11475	2.6346	0.84
0.0	52.0	11700	1.8957	0.88
1.7162	53.0	11925	2.7006	0.85
0.0	54.0	12150	2.2261	0.86
0.0	55.0	12375	1.5562	0.89
0.0	56.0	12600	1.4557	0.91
0.0	57.0	12825	1.6862	0.89
0.0	58.0	13050	1.6635	0.9
0.0	59.0	13275	2.5130	0.85
0.0	60.0	13500	2.1794	0.84
0.0	61.0	13725	3.1630	0.82
0.0	62.0	13950	2.4938	0.84
0.0	63.0	14175	2.9464	0.82
0.0	64.0	14400	3.0567	0.81
0.0	65.0	14625	3.0951	0.82
0.0	66.0	14850	2.8673	0.82
0.0	67.0	15075	2.9092	0.82
0.0	68.0	15300	2.2521	0.85
0.0	69.0	15525	2.5049	0.82
0.0	70.0	15750	2.4376	0.84
0.0	71.0	15975	2.6660	0.82
0.0	72.0	16200	2.5182	0.86
0.0	73.0	16425	2.3814	0.85
0.0	74.0	16650	2.3093	0.85
0.0	75.0	16875	2.3014	0.85
0.0	76.0	17100	2.3845	0.86
0.0	77.0	17325	2.2978	0.85
0.0	78.0	17550	2.4215	0.85
0.7047	79.0	17775	2.3462	0.84
0.0	80.0	18000	2.3230	0.85
0.0	81.0	18225	2.3648	0.85
0.0	82.0	18450	2.2962	0.85
0.0	83.0	18675	2.4231	0.84
0.0	84.0	18900	2.2588	0.86
0.0	85.0	19125	2.4144	0.84
0.0	86.0	19350	2.4220	0.84
0.0	87.0	19575	2.3860	0.84
0.0	88.0	19800	2.3356	0.84
0.0	89.0	20025	2.3223	0.85
0.0	90.0	20250	2.3554	0.83
0.0	91.0	20475	2.3344	0.84
0.6906	92.0	20700	2.3568	0.85
0.0	93.0	20925	2.3905	0.84
0.0	94.0	21150	2.3920	0.85
0.0	95.0	21375	2.3935	0.85
0.0	96.0	21600	2.3392	0.85
0.0	97.0	21825	2.3437	0.85
0.0	98.0	22050	2.3434	0.85
0.0	99.0	22275	2.3503	0.85
0.0	100.0	22500	2.3496	0.85

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0
• Datasets 2.15.0
• Tokenizers 0.15.0