whisper-large-v3-Hindi-Version1

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

• Loss: 0.1571
• Wer: 18.1667

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-06
• 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: 1000
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1799	6.7797	2000	0.1806	21.3881
0.1631	13.5593	4000	0.1678	20.0703
0.1436	20.3390	6000	0.1622	19.4748
0.145	27.1186	8000	0.1593	18.8403
0.1316	33.8983	10000	0.1578	18.5670
0.1293	40.6780	12000	0.1574	18.5182
0.1281	47.4576	14000	0.1570	18.4010
0.1258	54.2373	16000	0.1569	18.0594
0.1192	61.0169	18000	0.1571	18.4108
0.128	67.7966	20000	0.1571	18.1667

Framework versions

• PEFT 0.12.1.dev0
• Transformers 4.45.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1