wav2vec2-base-wakeword

This model is a fine-tuned version of facebook/wav2vec2-base on the superb dataset. It achieves the following results on the evaluation set:

• Loss: 0.1988
• Accuracy: 0.8980

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 0
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• 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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5835	0.9832	44	0.4825	0.8305
0.3198	1.9888	89	0.3220	0.8681
0.2074	2.9944	134	0.2879	0.8571
0.164	4.0	179	0.2867	0.8454
0.1524	4.9832	223	0.2757	0.8414
0.1529	5.9888	268	0.3233	0.8273
0.1256	6.9944	313	0.2192	0.8666
0.1169	8.0	358	0.1988	0.8980
0.1128	8.9832	402	0.2188	0.8713
0.1252	9.8324	440	0.2259	0.8689

Framework versions

• Transformers 4.44.2
• Pytorch 2.3.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1