arindam2007b/whisper-small-sw

Whisper Small sw - Arindam Bose

This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4499
• Wer: 28.8790

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: 16
• 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: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4791	0.4342	1000	0.5847	37.9256
0.396	0.8684	2000	0.4834	31.9206
0.2344	1.3026	3000	0.4631	31.2366
0.2237	1.7369	4000	0.4402	29.6515
0.1096	2.1711	5000	0.4468	29.0480
0.1219	2.6053	6000	0.4431	28.6169
0.0743	3.0395	7000	0.4428	28.3318
0.0691	3.4737	8000	0.4499	28.8790

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1