wav2vec2-lv-60-espeak-cv-ft-WCTC-test-phocab-ds-f2

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: 18.0492
• Per: 0.0906

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
1310.3143	0.7194	400	229.5593	1.0
478.9703	1.4388	800	218.7479	1.0
455.1237	2.1583	1200	198.7823	1.0
353.171	2.8777	1600	88.7909	0.4674
210.2026	3.5971	2000	44.8467	0.2317
154.7437	4.3165	2400	30.6975	0.1587
129.805	5.0360	2800	24.1387	0.1305
115.7423	5.7554	3200	21.0714	0.1110
109.4557	6.4748	3600	19.6631	0.0974
102.8998	7.1942	4000	18.8954	0.0906
101.03	7.9137	4400	18.0492	0.0906

Framework versions

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