jialicheng/unlearn_speech_commands_wav2vec2-large_random_label_10_42

superb_ks_42

This model is a fine-tuned version of facebook/wav2vec2-large on the superb dataset. It achieves the following results on the evaluation set:

• Loss: 0.2619
• Accuracy: 0.9613
• Test Accuracy: 0.9613
• Df Accuracy: 0.9401
• Unlearn Overall Accuracy: 0.5106
• Unlearn Time: 8889.7177

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
0.5797	1.0	1597	0.2006	0.9781	0.5017	0.5017	-1
0.4475	2.0	3194	0.2619	0.9401	0.5106	0.5106	-1
0.4817	3.0	4791	0.2555	0.9667	0.5042	0.5042	-1
0.474	4.0	6388	0.2657	0.9701	0.5008	0.5008	-1
0.4632	5.0	7985	0.2097	0.9706	0.5032	0.5032	-1
0.4548	6.0	9582	0.2234	0.9716	0.5032	0.5032	-1
0.4462	7.0	11179	0.2285	0.9726	0.5013	0.5013	-1
0.4317	8.0	12776	0.2170	0.9769	0.5014	0.5014	-1
0.4215	9.0	14373	0.2663	0.9771	0.4992	0.4992	-1
0.4148	10.0	15970	0.2204	0.9734	0.5025	0.5025	-1
0.4063	11.0	17567	0.2119	0.9767	0.5020	0.5020	-1
0.3976	12.0	19164	0.2212	0.9751	0.5018	0.5018	-1
0.3862	13.0	20761	0.2095	0.9775	0.5020	0.5020	-1
0.3865	14.0	22358	0.2162	0.9781	0.5015	0.5015	-1
0.3717	15.0	23955	0.2064	0.9789	0.5018	0.5018	-1
0.3706	16.0	25552	0.2204	0.9794	0.5007	0.5007	-1
0.36	17.0	27149	0.2032	0.9802	0.5000	0.5000	-1
0.3594	18.0	28746	0.2068	0.9800	0.5013	0.5013	-1
0.3508	19.0	30343	0.2005	0.9794	0.5015	0.5015	-1
0.3481	20.0	31940	0.2063	0.9802	0.5009	0.5009	-1

Framework versions

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