Epoch
One complete pass of the model over the entire training dataset. Models usually train for many epochs, seeing the data again and again to keep improving.
Think of It Like This
One epoch is a single full read-through of the textbook — you normally need several before it sticks.
If your dataset has 10,000 examples, one epoch means the model has seen all 10,000 once. Real training runs for many epochs so the model can keep refining its weights — but piling on too many epochs against the same data is exactly how overfitting creeps in, so the count is something you tune rather than crank.