chinese-wav2vec2-large-CV16

This model is a fine-tuned version of TencentGameMate/chinese-wav2vec2-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 27.1177
• Wer: 1.0

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.PAGED_LION_8BIT and the args are: No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.08333333333333333
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
122.8458	0.5042	60	105.1826	1.0
77.2555	1.0084	120	55.8250	1.0
45.8774	1.5126	180	31.9373	1.0
36.0461	2.0168	240	27.7970	1.0
36.2704	2.5210	300	27.1177	1.0

Framework versions

• Transformers 4.48.2
• Pytorch 2.2.1+cu121
• Datasets 2.17.1
• Tokenizers 0.21.0