meiiny00/whisper-medium-checkpoint-8000

Whisper Medium New Train

This model is a fine-tuned version of openai/whisper-medium on the Common Voice 18.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0204
• Wer: 2.2783

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: 32
• eval_batch_size: 32
• 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: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2733	0.4077	1000	0.2585	32.5924
0.1527	0.8153	2000	0.1246	16.7238
0.0655	1.2230	3000	0.0776	10.5668
0.0455	1.6307	4000	0.0514	6.7675
0.0162	2.0383	5000	0.0353	4.4772
0.0129	2.4460	6000	0.0274	3.4364
0.0117	2.8536	7000	0.0220	2.5110
0.0044	3.2613	8000	0.0204	2.2783

Framework versions

• Transformers 4.44.2
• Pytorch 2.3.1
• Datasets 3.0.0
• Tokenizers 0.19.1