Hello, can someone explain what an epoch is in ML/DL pls, I see that term way too often but didn't really understand what it is, how it works or why do we need it.
#What are epochs in machine learning/deep learning?
karmic cobalt
onyx vessel
Hello, can someone explain what an epoch is in ML/DL pls, I see that term way to...
I noticed you posted few questions in this forum
I would recommend to learn first about ML ( traditional ML )
also check sentdex neural network series
and also answer to your question is,
this is sequential flow while training
-> forward pass
-> calculate loss
-> gradients calculations
this is called as 1 epochs
no to converge a model on a dataaset we need multiple iterations/epochs so that we are ensuring model is learning
karmic cobalt
also check `sentdex` neural network series
will have a look, ty !
random abyss
Hello, can someone explain what an epoch is in ML/DL pls, I see that term way to...
Hello! An epoch simply means 1 pass over the dataset. Its a term used in machine learning during training where the model iterates upon the entire training dataset once. 1 epoch means one pass over the dataset. Likewise, n epochs means training through the dataset for the nth time. A batch is a round of forward pass and backward pass for whatever batch size you might have. If you have a batch size of 64, that means each batch is done in tandem with 64 examples. In summary, epochs refer to N iterations of the full dataset. Suppose you have a dataset of 30 examples. Then, one epoch would mean the model iterating on all 30 examples of the dataset. A batch is one act of "iteration" that the model does during training based on the batch size. With a dataset of 30 examples, a batch refers to the iteration while the batch size refers to how many examples you process in one batch.