jstoone/whisper-small-da-cv11

Whisper Small da - Common Voice+FLEURS

This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0, FLEURS dataset. It achieves the following results on the evaluation set:

• Loss: 0.9242
• Wer: 138.8532

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-07
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.22	15.0	1000	1.2238	197.5434
1.004	30.0	2000	1.0386	197.9221
0.9104	45.01	3000	0.9704	156.6544
0.8455	60.01	4000	0.9352	142.3619
0.8237	75.01	5000	0.9242	138.8532

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2