IoanaLivia/real_data_6_h_synth_standard_A_horoscope-whisper-small

This model is a fine-tuned version of openai/whisper-small on the IoanaLivia/real_data_6_h_synth_standard_A_horoscope dataset. It achieves the following results on the evaluation set:

• Loss: 0.3101
• Wer: 17.1910

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
0.1573	1.0	151	0.2676	18.0162
0.0525	2.0	302	0.2681	17.3801
0.0226	3.0	453	0.2895	17.5176
0.0102	4.0	604	0.3101	17.1910
0.0053	5.0	755	0.3305	18.7382
0.0031	6.0	906	0.3422	19.6837
0.0022	7.0	1057	0.3547	18.6866
0.0015	8.0	1208	0.3584	18.2225
0.0011	9.0	1359	0.3690	18.5319
0.001	10.0	1510	0.3741	18.3772
0.0009	11.0	1661	0.3776	18.4459
0.0008	12.0	1812	0.3807	18.5319
0.0007	13.0	1963	0.3828	18.4631
0.0007	14.0	2114	0.3840	18.4459
0.0007	15.0	2265	0.3846	18.4459

Framework versions

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