whisper-large-v3-turbo-Telugu-Version1

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8897
• Wer: 103.8462

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0234	142.8571	2000	0.4991	98.3516
0.0024	285.7143	4000	0.6494	95.6044
0.0008	428.5714	6000	0.7260	95.0549
0.0004	571.4286	8000	0.7513	94.5055
0.0003	714.2857	10000	0.7775	95.0549
0.0002	857.1429	12000	0.8183	109.3407
0.0002	1000.0	14000	0.8304	92.3077
0.0001	1142.8571	16000	0.8528	96.1538
0.0001	1285.7143	18000	0.8839	100.0
0.0001	1428.5714	20000	0.8897	103.8462

Framework versions

• PEFT 0.14.0
• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.20.1