jialicheng/unlearn_speech_commands_hubert-base_random_label_10_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: 0.1972
• Accuracy: 0.9801
• Test Accuracy: 0.9801
• Df Accuracy: 0.9761
• Unlearn Overall Accuracy: 0.5020
• Unlearn Time: 3290.3072

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
0.5961	1.0	1597	0.2042	0.9769	0.5018	0.5018	-1
0.3942	2.0	3194	0.1972	0.9761	0.5020	0.5020	-1
0.4031	3.0	4791	0.2011	0.9779	0.5017	0.5017	-1
0.3894	4.0	6388	0.2182	0.9793	0.5005	0.5005	-1
0.3859	5.0	7985	0.2111	0.9779	0.5014	0.5014	-1
0.3796	6.0	9582	0.1843	0.9798	0.5011	0.5011	-1
0.3758	7.0	11179	0.1780	0.9793	0.5014	0.5014	-1
0.3677	8.0	12776	0.1899	0.9806	0.5015	0.5015	-1
0.3617	9.0	14373	0.1821	0.9808	0.5003	0.5003	-1
0.357	10.0	15970	0.1885	0.9812	0.5009	0.5009	-1
0.3557	11.0	17567	0.1725	0.9793	0.5020	0.5020	-1
0.3524	12.0	19164	0.1845	0.9806	0.5011	0.5011	-1
0.3396	13.0	20761	0.1820	0.9818	0.5008	0.5008	-1
0.3484	14.0	22358	0.1846	0.9806	0.5020	0.5020	-1
0.3389	15.0	23955	0.1860	0.9810	0.5020	0.5020	-1
0.3407	16.0	25552	0.1832	0.9820	0.5011	0.5011	-1
0.3347	17.0	27149	0.1963	0.9816	0.5013	0.5013	-1
0.3321	18.0	28746	0.1940	0.9820	0.5005	0.5005	-1
0.327	19.0	30343	0.1962	0.9816	0.5010	0.5010	-1
0.3279	20.0	31940	0.1949	0.9820	0.5008	0.5008	-1

Framework versions

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