cv-14-rakugo

This model is a fine-tuned version of pinot/wav2vec2-xls-r-300m-ja-cv-14_4 on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 5.1127
• Wer: 0.6460

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: 0.1
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.59	1	12.0211	0.3851
No log	1.78	3	12.0211	0.3851
No log	2.96	5	5.0175	0.4161
No log	3.56	6	2.7911	0.5093
No log	4.74	8	2.4989	0.5093
No log	5.93	10	4.0524	0.4907
No log	6.52	11	3.6691	0.4720
No log	7.7	13	4.6285	0.5280
No log	8.89	15	4.5016	0.5652
No log	9.48	16	4.4013	0.5776
No log	10.67	18	4.3718	0.6025
No log	11.85	20	5.1127	0.6460

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0