hyojin99/whisper_hyojin4

ft_model

This model is a fine-tuned version of openai/whisper-base on the EBRC dataset. It achieves the following results on the evaluation set:

• Loss: 0.3928
• Cer: 16.6510

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: 5e-05
• train_batch_size: 64
• 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: 50
• training_steps: 3125
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.4643	0.83	625	0.4565	20.5573
0.2593	1.67	1250	0.4113	17.9447
0.1449	2.5	1875	0.3965	17.4427
0.0774	3.33	2500	0.3921	16.7845
0.0433	4.17	3125	0.3928	16.6510

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2