AI engineering glossary · Deep learning core
What is BPTT (backpropagation through time)?
Backpropagation through time (also called BPTT): The way recurrent networks are trained: the network is unrolled across the sequence steps into a long feedforward graph and standard backpropagation is applied through all of them. Long sequences make gradients vanish or explode, so it is often truncated.
Backpropagation through time: an example
Truncating to windows of 100 steps limits how far back gradients travel while keeping cost manageable.
Related terms
More Deep learning core terms
- Backward pass
- Batch normalisation
- Batch renormalisation
- Batch size
- bfloat16
- Bias term
- Bidirectional RNN
- Binary cross-entropy
Where you learn it
Backpropagation through time is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.