luluw/whisper-medium

Whisper Tiny

This model is a fine-tuned version of openai/whisper-medium on the Personal - Mimic Recording dataset. It achieves the following results on the evaluation set:

• Loss: 0.1404
• Wer: 0.0645

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 75
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3839	0.9932	73	0.1968	0.0975
0.0763	2.0	147	0.1418	0.0879
0.017	2.9932	220	0.1410	0.1200
0.0058	4.0	294	0.1404	0.0645
0.0014	4.9660	365	0.1396	0.0647

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.19.1