wav2vec2-lv-60-espeak-cv-ft-WCTC-test-phocab-ds-f8

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: 20.3507
• Per: 0.0965

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
1314.8512	0.7194	400	243.1245	1.0
479.9098	1.4388	800	230.8923	1.0
443.617	2.1583	1200	181.3888	0.9670
296.7681	2.8777	1600	67.2370	0.3331
183.1587	3.5971	2000	44.3972	0.2027
145.8179	4.3165	2400	31.9055	0.1545
124.7866	5.0360	2800	26.7603	0.1191
112.7309	5.7554	3200	23.0639	0.1030
106.9614	6.4748	3600	21.6685	0.1022
103.4507	7.1942	4000	20.7531	0.0965
99.4513	7.9137	4400	20.3507	0.0965

Framework versions

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