Check_Model_1

This model is a fine-tuned version of facebook/wav2vec2-large on the common_voice dataset. It achieves the following results on the evaluation set:

• Loss: 0.5522
• Wer: 0.3748
• Cer: 0.1158

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
2.1839	3.23	400	0.8796	0.7306	0.2332
0.6388	6.45	800	0.8702	0.6410	0.2200
0.4695	9.68	1200	0.7064	0.5360	0.1632
0.3659	12.9	1600	0.5814	0.5211	0.1662
0.285	16.13	2000	0.6394	0.5041	0.1663
0.2254	19.35	2400	0.5889	0.4428	0.1405
0.1801	22.58	2800	0.5712	0.4013	0.1182
0.1392	25.81	3200	0.5914	0.3934	0.1177
0.1051	29.03	3600	0.5522	0.3748	0.1158

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu117
• Datasets 1.18.3
• Tokenizers 0.13.3