FelixK7/whisper-medium-lv-ver3

Whisper medium LV - Felikss Kleins

This model is a fine-tuned version of FelixK7/whisper-medium-lv-ver2 on the Common Voice 16.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2882
• Wer: 19.3925

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: 2e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_hf with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	99.0002	200	0.1666	11.4486
0.0028	199.0002	400	0.2083	13.5514
0.0007	299.0002	600	0.2815	20.7944
0.0008	399.0002	800	0.2882	19.3925

Framework versions

• Transformers 4.46.0.dev0
• Pytorch 2.0.1
• Datasets 3.0.1
• Tokenizers 0.20.1