wav2vec2-btb-ccv-ft-btb-cy

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

• Loss: 0.4314
• Wer: 0.3285

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: 16
• 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: 300
• training_steps: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.1139	200	0.7148	0.4603
No log	0.2278	400	0.6810	0.4795
1.4735	0.3417	600	0.6035	0.4486
1.4735	0.4556	800	0.6223	0.5105
0.7681	0.5695	1000	0.5656	0.4336
0.7681	0.6834	1200	0.5275	0.4008
0.7681	0.7973	1400	0.5284	0.4028
0.7159	0.9112	1600	0.4990	0.3914
0.7159	1.0251	1800	0.4855	0.3730
0.6203	1.1390	2000	0.4740	0.3623
0.6203	1.2528	2200	0.4589	0.3536
0.6203	1.3667	2400	0.4539	0.3407
0.5447	1.4806	2600	0.4410	0.3357
0.5447	1.5945	2800	0.4347	0.3293
0.5392	1.7084	3000	0.4314	0.3285

Framework versions

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