w2v2-base-10k-voxpopuli-ft-de_lr1e-4_at0.8_da1

This model is a fine-tuned version of facebook/wav2vec2-base-10k-voxpopuli-ft-de on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.8259
• Wer: 0.1632

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: 32
• 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: 1000
• num_epochs: 60
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8086	5.43	250	0.8872	0.2456
0.1358	10.87	500	1.1126	0.1858
0.0916	16.3	750	1.5579	0.1991
0.0716	21.74	1000	1.2674	0.1833
0.0584	27.17	1250	1.6728	0.1768
0.0464	32.61	1500	1.8259	0.1632

Framework versions

• Transformers 4.35.0
• Pytorch 2.0.0
• Datasets 2.14.6
• Tokenizers 0.14.1