jialicheng/unlearn_speech_commands_hubert-large_neggrad_10_42

superb_ks_42

This model is a fine-tuned version of facebook/hubert-large-ll60k on the superb dataset. It achieves the following results on the evaluation set:

• Loss: -10.4071
• Accuracy: 0.6209
• Test Accuracy: 0.6209
• Df Accuracy: 0.1405
• Unlearn Overall Accuracy: 0.7402
• Unlearn Time: 1025.3563

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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Overall Accuracy	Unlearn Overall Accuracy	Time
No log	1.0	160	-10.4071	0.1405	0.7402	0.7402	-1
No log	2.0	320	-26.8660	0.1405	0.7402	0.7402	-1
No log	3.0	480	-47.9967	0.1405	0.7402	0.7402	-1
No log	4.0	640	-72.2298	0.1405	0.7402	0.7402	-1
No log	5.0	800	-97.3898	0.1405	0.7402	0.7402	-1
No log	6.0	960	-121.2981	0.1405	0.7402	0.7402	-1
-123.4445	7.0	1120	-142.0651	0.1405	0.7402	0.7402	-1
-123.4445	8.0	1280	-158.1099	0.1405	0.7402	0.7402	-1
-123.4445	9.0	1440	-168.2401	0.1405	0.7402	0.7402	-1
-123.4445	10.0	1600	-171.7223	0.1405	0.7402	0.7402	-1

Framework versions

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