whisper_large_v2_fixed_timestamps

This model is a fine-tuned version of openai/whisper-large-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6408
• Cer: 14.2277
• Wer: 24.0848

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: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.8984	1.0	6265	0.6727	15.8499	27.0820
0.6156	2.0	12530	0.6507	15.4948	26.1944
0.5175	3.0	18795	0.6408	14.2277	24.0848

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.2+cu118
• Datasets 2.19.0
• Tokenizers 0.19.1