How to set batch size and steps per epoch when using TFRecord files?
I have a training data that consists of 6011 images. Now , I have converted this data to tfrecord files , where each file consists of 128 records and there are 47 files ( note that the last file consists of 123 records only ) . So my question is how do I correctly set the batch size and steps per epoch for training the model ? Do I set the batch size to 128 ( since each file consists of 128 records ) and the steps per epoch to 47 ( since there are 47 files ) ?
Topic epochs training deep-learning
Category Data Science