jialicheng/unlearn_speech_commands_hubert-base_bad_teaching_2_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: 274600.625
• Accuracy: 0.6168
• Test Accuracy: 0.6168
• Df Accuracy: 0.1195
• Unlearn Overall Accuracy: 0.7487
• Unlearn Time: 9214.8598

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	1564	2.3320	0.2419	0.7980	0.7980	-1
0.0	1.0	3128	43.3185	0.2380	0.6914	0.6914	-1
0.0	1.5	4692	204.8400	0.2351	0.6923	0.6923	-1
0.0	2.0	6256	655.8871	0.2282	0.6939	0.6939	-1
0.0	2.5	7820	1597.6744	0.2233	0.6948	0.6948	-1
0.0	3.0	9384	3150.5979	0.2214	0.6971	0.6971	-1
0.0	3.5	10948	5529.0425	0.2047	0.7029	0.7029	-1
0.0	4.0	12512	8752.1719	0.2008	0.7015	0.7015	-1
0.0	4.5	14076	12488.5586	0.2047	0.7038	0.7038	-1
0.0	5.0	15640	17659.9004	0.2018	0.7077	0.7077	-1
0.0	5.5	17204	22695.4512	0.1832	0.7125	0.7125	-1
0.0	6.0	18768	29303.1426	0.1714	0.7121	0.7121	-1
0.0	6.5	20332	35666.2539	0.1822	0.7165	0.7165	-1
0.0	7.0	21896	43349.8242	0.1714	0.7207	0.7207	-1
0.0	7.5	23460	51642.0039	0.1655	0.7233	0.7233	-1
0.0	7.99	25024	61990.25	0.1694	0.7236	0.7236	-1
0.0	8.49	26588	71746.25	0.1616	0.7268	0.7268	-1
0.0	8.99	28152	83252.8438	0.1587	0.7286	0.7286	-1
0.0	9.49	29716	96622.7344	0.1567	0.7312	0.7312	-1
0.0	9.99	31280	106065.3594	0.1499	0.7300	0.7300	-1
0.0	10.49	32844	116312.9688	0.1538	0.7311	0.7311	-1
0.0	10.99	34408	133755.5156	0.1528	0.7325	0.7325	-1
0.0	11.49	35972	145134.125	0.1499	0.7345	0.7345	-1
0.0	11.99	37536	171217.9219	0.1430	0.7326	0.7326	-1
0.0	12.49	39100	165460.2031	0.1401	0.7375	0.7375	-1
0.0	12.99	40664	186145.6719	0.1430	0.7372	0.7372	-1
0.0	13.49	42228	190899.4375	0.1430	0.7378	0.7378	-1
0.0	13.99	43792	197004.9219	0.1381	0.7397	0.7397	-1
0.0	14.49	45356	209167.8906	0.1342	0.7398	0.7398	-1
0.0	14.99	46920	220491.7812	0.1342	0.7406	0.7406	-1
0.0	15.49	48484	239906.6094	0.1342	0.7417	0.7417	-1
0.0	15.99	50048	242919.1875	0.1342	0.7425	0.7425	-1
0.0	16.49	51612	249936.1719	0.1303	0.7431	0.7431	-1
0.0	16.99	53176	259517.2344	0.1312	0.7441	0.7441	-1
0.0	17.49	54740	261543.9531	0.1244	0.7469	0.7469	-1
0.0	17.99	56304	267596.125	0.1244	0.7465	0.7465	-1
0.0	18.49	57868	271231.7188	0.1234	0.7465	0.7465	-1
0.0	18.99	59432	272808.6562	0.1214	0.7479	0.7479	-1
0.0	19.49	60996	273415.9688	0.1205	0.7479	0.7479	-1
0.0	19.99	62560	274608.5938	0.1195	0.7487	0.7487	-1

Framework versions

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