IoanaLiviaPopescu/ IoanaLiviaPopescu/real-data-synth-data-1600-1-St-Emil-whisper-small

This model is a fine-tuned version of openai/whisper-small on the IoanaLivia/RealVoiceSynthVoice-1600-1-St-Emil dataset. It achieves the following results on the evaluation set:

• Loss: 0.3718
• Wer: 17.0385

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: 32
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0	0	0.6024	27.8812
0.282	1.0	51	0.3977	18.3109
0.1077	2.0	102	0.3658	17.3151
0.0561	3.0	153	0.3718	17.0385
0.0328	4.0	204	0.3881	17.3889
0.023	5.0	255	0.4000	17.7208

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1