wav2vec2-lv-60-espeak-cv-ft-WCTC-test-phocab-ds-f7

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: 16.4494
• Per: 0.0770

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
1310.7713	0.7194	400	238.3591	1.0
479.3892	1.4388	800	227.4793	1.0
457.2696	2.1583	1200	209.4901	1.0
357.6957	2.8777	1600	93.8761	0.5478
201.3866	3.5971	2000	37.6993	0.1410
148.4454	4.3165	2400	27.1875	0.1032
126.0141	5.0360	2800	21.9473	0.1002
114.2221	5.7554	3200	19.6805	0.0909
106.8823	6.4748	3600	17.2865	0.0871
103.0399	7.1942	4000	16.6731	0.0770
102.9444	7.9137	4400	16.4494	0.0770

Framework versions

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