whisper-small-turkish-0

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2766

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: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 45000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.3721	0.0556	2500	0.3822
0.344	0.1111	5000	0.3584
0.3255	0.1667	7500	0.3362
0.2995	0.2222	10000	0.3249
0.2803	0.2778	12500	0.3112
0.2739	0.3333	15000	0.3050
0.2543	0.3889	17500	0.2985
0.2491	0.4444	20000	0.2923
0.2403	0.5	22500	0.2885
0.2231	0.5556	25000	0.2837
0.2131	0.6111	27500	0.2806
0.199	0.6667	30000	0.2795
0.1911	0.7222	32500	0.2774
0.181	0.7778	35000	0.2766
0.1855	0.8333	37500	0.2761
0.175	0.8889	40000	0.2766
0.1664	1.0394	42500	0.2770
0.1675	1.095	45000	0.2766

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.20.3