jialicheng/unlearn_speech_commands_hubert-base_random_label_8_42

superb_ks_42

This model is a fine-tuned version of facebook/hubert-base-ls960 on the superb dataset. It achieves the following results on the evaluation set:

• Loss: 0.1851
• Accuracy: 0.9828
• Test Accuracy: 0.9828
• Df Accuracy: 0.9760
• Unlearn Overall Accuracy: 0.5034
• Unlearn Time: 3255.7985

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: 32
• 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
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Overall Accuracy	Unlearn Overall Accuracy	Time
0.5192	1.0	1597	0.1967	0.9804	0.5013	0.5013	-1
0.345	2.0	3194	0.1964	0.9770	0.5016	0.5016	-1
0.3488	3.0	4791	0.1939	0.9782	0.5010	0.5010	-1
0.3461	4.0	6388	0.1945	0.9772	0.5005	0.5005	-1
0.3365	5.0	7985	0.1853	0.9787	0.5013	0.5013	-1
0.3339	6.0	9582	0.1863	0.9792	0.5011	0.5011	-1
0.3202	7.0	11179	0.1719	0.9782	0.5015	0.5015	-1
0.3232	8.0	12776	0.1803	0.9790	0.5011	0.5011	-1
0.3127	9.0	14373	0.1715	0.9785	0.5016	0.5016	-1
0.3094	10.0	15970	0.1700	0.9799	0.5018	0.5018	-1
0.3055	11.0	17567	0.1977	0.9799	0.5008	0.5008	-1
0.303	12.0	19164	0.1674	0.9814	0.5011	0.5011	-1
0.29	13.0	20761	0.1750	0.9812	0.5005	0.5005	-1
0.2962	14.0	22358	0.1851	0.9760	0.5034	0.5034	-1
0.2914	15.0	23955	0.1698	0.9816	0.5009	0.5009	-1
0.2938	16.0	25552	0.1737	0.9804	0.5016	0.5016	-1
0.287	17.0	27149	0.1829	0.9799	0.5014	0.5014	-1
0.2829	18.0	28746	0.1694	0.9799	0.5019	0.5019	-1
0.2788	19.0	30343	0.1769	0.9799	0.5018	0.5018	-1
0.276	20.0	31940	0.1784	0.9794	0.5020	0.5020	-1

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.2+cu118
• Datasets 2.18.0
• Tokenizers 0.15.2