wav2vec2-lv-60-espeak-cv-ft-WCTC-test-phocab-ds-f0

This model is a fine-tuned version of facebook/wav2vec2-lv-60-espeak-cv-ft on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 52.1965
• Per: 0.3111

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Per
1364.575	0.7194	400	249.5449	1.0
483.1877	1.4388	800	234.9056	1.0
456.6452	2.1583	1200	221.9890	1.0
388.4952	2.8777	1600	140.9373	0.7747
261.8777	3.5971	2000	92.5198	0.5364
186.9818	4.3165	2400	65.6089	0.3680
147.0983	5.0360	2800	58.5163	0.3392
127.9282	5.7554	3200	54.0905	0.3232
117.0765	6.4748	3600	53.4431	0.3156
113.0818	7.1942	4000	52.8302	0.3179
111.6685	7.9137	4400	52.1965	0.3111

Framework versions

• Transformers 4.57.6
• Pytorch 2.9.1+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2