Sercan/whisper-small-cv-tr

Whisper Small Turkish

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_11_0 tr dataset. It achieves the following results on the evaluation set:

• Loss: 0.2860
• Wer: 16.3181
• Cer: 4.1450

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: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.1563	1.0	2500	0.2524	19.8570	5.1738
0.032	2.01	5000	0.2567	18.5627	4.7793
0.013	3.01	7500	0.2637	17.7723	4.6664
0.0057	4.02	10000	0.2703	17.0596	4.3662
0.0012	5.02	12500	0.2696	17.8322	5.2286
0.003	6.03	15000	0.2800	16.7200	4.2972
0.0003	7.03	17500	0.2834	16.4091	4.2018
0.0002	8.04	20000	0.2860	16.3181	4.1450

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.9.1.dev0
• Tokenizers 0.13.2