snu-nia-12/wav2vec2-large_nia12_phone-ipa_english

wav2vec2-large-nia12_phone-ipa_english

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

• Loss: 0.1531
• Per: 0.0638

Model description

More information needed

Training and evaluation data

Trained & Evaluated on TIMIT dataset

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 64
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Step	Validation Loss	Per
2.0846	500	0.1810	0.0991
0.1857	1000	0.1411	0.0691
0.0948	1500	0.1345	0.0666
0.0646	2000	0.1444	0.0673
0.0502	2500	0.1436	0.0628
0.0381	3000	0.1383	0.0637
0.0309	3500	0.1533	0.0638
0.0261	4000	0.1531	0.0638

Framework versions

• Transformers 4.20.0
• Pytorch 1.12.1+cu113
• Datasets 2.6.2.dev0
• Tokenizers 0.12.1