med_v1_M04_1e-05

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

• Loss: 3.4848
• Wer: 1.0
• Cer: 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: 1e-05
• train_batch_size: 20
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 18400
• num_epochs: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
38.6628	200.0	36800	3.4943	1.0	1.0
3.1727	400.0	73600	3.4369	1.0	1.0
3.0757	600.0	110400	3.4349	1.0	1.0
3.0212	800.0	147200	3.4523	1.0	1.0
2.9854	1000.0	184000	3.4652	1.0	1.0
2.9605	1200.0	220800	3.4733	1.0	1.0
2.943	1400.0	257600	3.4786	1.0	1.0
2.9311	1600.0	294400	3.4810	1.0	1.0
2.9235	1800.0	331200	3.4837	1.0	1.0
2.92	2000.0	368000	3.4848	1.0	1.0

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.2+cu102
• Datasets 2.3.2
• Tokenizers 0.12.1