whisper_large_v2_fixed_timestamps_phase2

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: 0.7693
• Cer: 13.9523
• Wer: 23.5161

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Cer	Validation Loss	Wer
0.8984	1.0	6265	15.8499	0.6727	27.0820
0.6156	2.0	12530	15.4948	0.6507	26.1944
0.5175	3.0	18795	14.2277	0.6408	24.0848
0.5372	4.0	25060	0.6782	15.4893	26.5959
0.4756	5.0	31325	0.6849	14.3892	24.5451
0.4197	6.0	37590	0.7009	14.5883	24.5874
0.3731	7.0	43855	0.7143	14.4249	24.3789
0.3318	8.0	50120	0.7251	14.1660	23.9351
0.2972	9.0	56385	0.7580	13.9645	23.5739
0.2714	10.0	62650	0.7693	13.9523	23.5161

Framework versions

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