This model is a fine-tuned version of openai/whisper-small on the IoanaLivia/real_data_1_h_synth_standard_B_horoscope dataset. It achieves the following results on the evaluation set:

Loss: 0.2705
Wer: 18.4459

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: 8
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
num_epochs: 15
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0	0	0.4170	30.1014
0.3332	1.0	42	0.2841	19.8556
0.1356	2.0	84	0.2705	18.4459
0.0678	3.0	126	0.2814	18.9789
0.0331	4.0	168	0.3037	20.0791
0.0162	5.0	210	0.3249	19.4774
0.0098	6.0	252	0.3424	19.2711
0.0065	7.0	294	0.3458	18.8413
0.005	8.0	336	0.3524	19.0992
0.0039	9.0	378	0.3605	19.0992
0.0034	10.0	420	0.3654	18.8929
0.003	11.0	462	0.3680	18.9445
0.0028	12.0	504	0.3708	18.8929
0.0026	13.0	546	0.3729	19.0132
0.0025	14.0	588	0.3737	19.0304
0.0025	14.6506	615	0.3738	19.0132

Framework versions

Transformers 4.51.3
Pytorch 2.6.0+cu124
Datasets 3.6.0
Tokenizers 0.21.1