procit006/whisper-small-nl-aug27

Whisper Small

This model is a fine-tuned version of openai/whisper-small on the Common Voice + STC Aug 27 + Speechgen dataset. It achieves the following results on the evaluation set:

• Loss: 0.0202
• Wer: 1.3962

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 150
• training_steps: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0892	0.2047	500	0.0897	6.6898
0.048	0.4093	1000	0.0480	3.4328
0.0379	0.6140	1500	0.0339	2.2033
0.0299	0.8187	2000	0.0269	2.2453
0.0074	1.0233	2500	0.0216	1.4842
0.0055	1.2280	3000	0.0202	1.3962

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1