wav2vec2-xlsr-53-ft-btb-cy

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the DEWIBRYNJONES/BANC-TRAWSGRIFIADAU-BANGOR-CLEAN - DEFAULT dataset. It achieves the following results on the evaluation set:

• Loss: 0.7008
• Wer: 0.4816

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: 0.0003
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 120
• training_steps: 1200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.4556	200	2.5809	0.9998
No log	0.9112	400	0.8879	0.6325
2.4327	1.3667	600	0.6609	0.4897
2.4327	1.8223	800	0.6689	0.5018
0.8178	2.2779	1000	0.6797	0.4852
0.8178	2.7335	1200	0.7008	0.4816

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1