jialicheng/unlearn_speech_commands_hubert-base_bad_teaching_4_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: 145389.0469
• Accuracy: 0.6199
• Test Accuracy: 0.6199
• Df Accuracy: 0.1635
• Unlearn Overall Accuracy: 0.7282
• Unlearn Time: 9164.5264

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	1532	1.9539	0.5996	0.6348	0.6348	-1
0.0	1.0	3064	26.1186	0.1845	0.7230	0.7230	-1
0.0	1.5	4596	89.9674	0.1845	0.7179	0.7179	-1
0.0	2.0	6128	253.3492	0.1826	0.7157	0.7157	-1
0.0	2.5	7660	631.4303	0.1826	0.7167	0.7167	-1
0.0	3.0	9192	1095.6816	0.1816	0.7189	0.7189	-1
0.0	3.5	10724	2312.6934	0.1816	0.7191	0.7191	-1
0.0	4.0	12256	2479.0049	0.1806	0.7182	0.7182	-1
0.0	4.5	13788	3809.8308	0.1772	0.7170	0.7170	-1
0.0	5.0	15320	4455.6138	0.1777	0.7185	0.7185	-1
0.0	5.5	16852	7533.3687	0.1796	0.7201	0.7201	-1
0.0	6.0	18384	8388.7559	0.1777	0.7198	0.7198	-1
0.0	6.5	19916	10954.2783	0.1791	0.7193	0.7193	-1
0.0	7.0	21448	20003.2617	0.1762	0.7215	0.7215	-1
0.0	7.5	22980	16605.4883	0.1752	0.7194	0.7194	-1
0.0	7.99	24512	24020.8633	0.1767	0.7219	0.7219	-1
0.0	8.49	26044	30725.0078	0.1752	0.7223	0.7223	-1
0.0	8.99	27576	29158.8203	0.1728	0.7227	0.7227	-1
0.0	9.49	29108	28054.3887	0.1718	0.7226	0.7226	-1
0.0	9.99	30640	41348.4961	0.1738	0.7229	0.7229	-1
0.0	10.49	32172	47708.1211	0.1723	0.7229	0.7229	-1
0.0	10.99	33704	53620.0547	0.1728	0.7234	0.7234	-1
0.0	11.49	35236	56777.2070	0.1694	0.7246	0.7246	-1
0.0	11.99	36768	61912.2734	0.1689	0.7254	0.7254	-1
0.0	12.49	38300	67168.25	0.1703	0.7246	0.7246	-1
0.0	12.99	39832	74464.25	0.1679	0.7257	0.7257	-1
0.0	13.49	41364	85865.7031	0.1664	0.7259	0.7259	-1
0.0	13.99	42896	79835.9609	0.1659	0.7262	0.7262	-1
0.0	14.49	44428	104636.9844	0.1669	0.7267	0.7267	-1
0.0	14.99	45960	96585.5156	0.1650	0.7272	0.7272	-1
0.0	15.49	47492	112208.3594	0.1650	0.7277	0.7277	-1
0.0	15.99	49024	121757.125	0.1659	0.7270	0.7270	-1
0.0	16.49	50556	121264.0156	0.1635	0.7275	0.7275	-1
0.0	16.99	52088	134592.6719	0.1635	0.7280	0.7280	-1
0.0	17.49	53620	135230.6406	0.1645	0.7278	0.7278	-1
0.0	17.99	55152	143017.4375	0.1625	0.7287	0.7287	-1
0.0	18.49	56684	140805.1094	0.1635	0.7279	0.7279	-1
0.0	18.99	58216	138893.3438	0.1630	0.7282	0.7282	-1
0.0	19.49	59748	144538.75	0.1635	0.7283	0.7283	-1
0.0	19.99	61280	145391.9062	0.1635	0.7282	0.7282	-1

Framework versions

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