mkenfenheuer/whisper-small-de-int8-ct2

This model is a converted version of mkenfenheuer/whisper-small-de converted to ctranslate2 with int8 quantization.

Whisper Small DE - Maximilian Kenfenheuer

This model is a fine-tuned version of openai/whisper-small on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.1231
• Wer: 7.5440

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: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1911	0.25	1000	0.1567	9.4088
0.1959	0.5	2000	0.1382	8.5261
0.1843	0.75	3000	0.1285	7.8513
0.2134	1.0	4000	0.1231	7.5440

Framework versions

• Transformers 4.48.0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0