jialicheng/unlearn_speech_commands_wav2vec2-large_bad_teaching_8_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: 54.2699
• Accuracy: 0.6209
• Test Accuracy: 0.6209
• Df Accuracy: 0.1354
• Unlearn Overall Accuracy: 0.7428
• Unlearn Time: 791.2602

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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Overall Accuracy	Unlearn Overall Accuracy	Time
No log	1.0	252	14.0799	0.2200	0.7175	0.7175	-1
No log	2.0	504	54.2699	0.1354	0.7428	0.7428	-1
No log	3.0	756	94.2331	0.1354	0.7428	0.7428	-1
0.0	4.0	1008	125.8463	0.1354	0.7428	0.7428	-1
0.0	5.0	1260	136.3317	0.1354	0.7428	0.7428	-1

Framework versions

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