GRU vs LSTM

Medium

neural-networks Google Amazon Microsoft

How does a Gated Recurrent Unit (GRU) differ structurally from an LSTM, and what practical trade-off does this create?

Multiple Choice

GRU uses two gates (reset and update) instead of three, and has no separate cell state, making it faster to train with comparable performance GRU adds a fourth gate for attention, making it more expressive but slower than LSTM GRU removes all gating and relies solely on residual connections for long-term memory GRU uses convolutional gates instead of fully connected gates for spatial efficiency

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

GRU vs LSTM

Multiple Choice

Related problems

Deep-dive guides