wav2vec2-lv-60-espeak-cv-ft-WCTC-test-phocab-ds-f5

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: 14.8731
• Per: 0.0661

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
1313.4922	0.7194	400	239.3583	1.0
475.4594	1.4388	800	224.8250	1.0
455.1527	2.1583	1200	207.4702	1.0
358.7267	2.8777	1600	76.0772	0.3829
190.6968	3.5971	2000	36.9121	0.1572
146.9133	4.3165	2400	25.7155	0.1154
125.3709	5.0360	2800	21.4662	0.0953
114.2245	5.7554	3200	18.3575	0.0778
106.5464	6.4748	3600	16.5186	0.0711
102.6576	7.1942	4000	15.5472	0.0677
101.9647	7.9137	4400	14.8731	0.0661

Framework versions

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