whisper-base-withnoiseV2

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

• Loss: 0.2076
• Cer: 34.4795

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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.0001	500.0	1000	0.2127	27.9847
0.0001	1000.0	2000	0.2128	34.9570
0.0	1500.0	3000	0.2095	24.1643
0.0	2000.0	4000	0.2076	34.4795

Framework versions

• Transformers 4.35.0.dev0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1