wav2vec2-large-xls-r-300m-j-phoneme-common-13

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:

• Loss: inf
• Wer: 0.0133

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: 0.0003
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1714	6.95	2000	0.0803	0.0920
0.0994	12.9	4000	inf	0.0542
0.0331	19.35	6000	inf	0.0238
0.0205	25.8	8000	inf	0.0196
0.0157	32.25	10000	inf	0.0146
0.0142	38.69	12000	inf	0.0133
0.0143	45.14	14000	inf	0.0133

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3