wav2vec2-lv-60-espeak-cv-ft-WCTC-test-phocab-ds-f1

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: 19.3612
• Per: 0.0815

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
1315.395	0.7194	400	244.5888	1.0
481.8904	1.4388	800	231.1810	1.0
457.0693	2.1583	1200	216.3756	1.0
364.8352	2.8777	1600	94.4524	0.4705
209.3284	3.5971	2000	46.0901	0.1903
158.0425	4.3165	2400	32.7669	0.1340
133.3767	5.0360	2800	25.7148	0.1003
118.7471	5.7554	3200	22.5087	0.0890
111.7019	6.4748	3600	20.6984	0.0797
106.347	7.1942	4000	19.6716	0.0778
105.7139	7.9137	4400	19.3612	0.0815

Framework versions

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