jialicheng/unlearn_speech_commands_hubert-base_bad_teaching_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: 140133.8906
• Accuracy: 0.6178
• Test Accuracy: 0.6178
• Df Accuracy: 0.1365
• Unlearn Overall Accuracy: 0.7406
• Unlearn Time: 9044.6562

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
0.0	0.5	1468	1.6647	0.6496	0.6108	0.6108	-1
0.0	1.0	2936	41.8264	0.1409	0.7418	0.7418	-1
0.0	1.5	4404	186.5033	0.1400	0.7404	0.7404	-1
0.0	2.0	5872	662.5567	0.1397	0.7399	0.7399	-1
0.0	2.5	7340	1073.7369	0.1390	0.7379	0.7379	-1
0.0	3.0	8808	4373.4556	0.1385	0.7390	0.7390	-1
0.0	3.5	10276	2357.4761	0.1390	0.7390	0.7390	-1
0.0	4.0	11744	6139.2163	0.1395	0.7401	0.7401	-1
0.0	4.5	13212	8287.3916	0.1387	0.7377	0.7377	-1
0.0	5.0	14680	12035.0781	0.1390	0.7380	0.7380	-1
0.0	5.5	16148	8776.9424	0.1375	0.7377	0.7377	-1
0.0	6.0	17616	19031.5059	0.1395	0.7397	0.7397	-1
0.0	6.5	19084	18370.6270	0.1390	0.7382	0.7382	-1
0.0	7.0	20552	19890.7129	0.1387	0.7388	0.7388	-1
0.0	7.49	22020	30039.3711	0.1392	0.7399	0.7399	-1
0.0	7.99	23488	24278.0586	0.1378	0.7388	0.7388	-1
0.0	8.49	24956	30144.4492	0.1375	0.7394	0.7394	-1
0.0	8.99	26424	36426.7383	0.1382	0.7402	0.7402	-1
0.0	9.49	27892	38128.0078	0.1382	0.7391	0.7391	-1
0.0	9.99	29360	57813.1562	0.1375	0.7407	0.7407	-1
0.0	10.49	30828	46250.6289	0.1368	0.7382	0.7382	-1
0.0	10.99	32296	95817.5156	0.1387	0.7402	0.7402	-1
0.0	11.49	33764	90252.1719	0.1382	0.7404	0.7404	-1
0.0	11.99	35232	94591.4141	0.1385	0.7402	0.7402	-1
0.0	12.49	36700	85062.7344	0.1370	0.7387	0.7387	-1
0.0	12.99	38168	79767.2422	0.1363	0.7392	0.7392	-1
0.0	13.49	39636	125117.4609	0.1378	0.7395	0.7395	-1
0.0	13.99	41104	107255.1875	0.1375	0.7398	0.7398	-1
0.0	14.49	42572	93985.9609	0.1368	0.7400	0.7400	-1
0.0	14.99	44040	140844.4219	0.1373	0.7407	0.7407	-1
0.0	15.49	45508	129205.8516	0.1380	0.7408	0.7408	-1
0.0	15.99	46976	110236.0	0.1365	0.7392	0.7392	-1
0.0	16.49	48444	134991.9688	0.1368	0.7401	0.7401	-1
0.0	16.99	49912	148763.4219	0.1368	0.7410	0.7410	-1
0.0	17.49	51380	128247.8828	0.1370	0.7405	0.7405	-1
0.0	17.99	52848	138579.3125	0.1368	0.7407	0.7407	-1
0.0	18.49	54316	139503.6406	0.1368	0.7409	0.7409	-1
0.0	18.99	55784	140640.0	0.1365	0.7404	0.7404	-1
0.0	19.49	57252	135256.6875	0.1365	0.7404	0.7404	-1
0.0	19.99	58720	140127.0938	0.1365	0.7406	0.7406	-1

Framework versions

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