train05

This model is a fine-tuned version of openai/whisper-base on the ksponspeech_05 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4059
• Cer: 13.3884

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: 128
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.434	0.13	100	0.4239	14.0769
0.4208	0.25	200	0.4212	14.0593
0.4245	0.38	300	0.4196	13.9936
0.4173	0.51	400	0.4169	14.0972
0.4245	0.63	500	0.4147	14.0310
0.4103	0.76	600	0.4102	14.1257
0.4097	0.88	700	0.4059	13.3884

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.15.0
• Tokenizers 0.15.0