GRU Cell: Reset & Update Gates

Input Sequence

Pattern

Steps / second 3

Gate Bias

Reset gate bias r 0.0

Update gate bias z 0.0

Live Readouts

Step t

Input xₜ

0.00

Mean r (reset)

0.00

Mean z (update)

0.00

‖h‖ (memory norm)

0.00

‖h − h̃‖ (update size)

0.00

How it works

A Gated Recurrent Unit updates its hidden state h_t (memory) from the previous hidden state h_t-1 and the current input x_t using exactly two learned gates — no separate cell state like an LSTM needs:

r_t = σ(W_r·[h_(t-1), x_t] + b_r)         reset gate
z_t = σ(W_z·[h_(t-1), x_t] + b_z)         update gate
h̃_t = tanh(W_h·[r_t⊙h_(t-1), x_t] + b_h)  candidate state
h_t = (1 − z_t)⊙h_(t-1) + z_t⊙h̃_t         new hidden state

Reset gate r — near 0, it wipes h_t-1 out of the candidate computation, so h̃_t is built almost fresh from x_t alone; near 1, the full memory feeds the candidate.
Update gate z — near 0, h_t barely changes (the cell "holds" its memory across the step); near 1, h_t is replaced almost entirely by the new candidate h̃_t. This convex blend is also why h stays bounded — it can never explode the way a plain RNN's hidden state can.
Bias sliders — shift every gate's pre-activation before the sigmoid, the same way a trained bias vector would after learning to always-forget or always-remember in a particular regime. Push the reset bias very negative to see the candidate ignore history; push the update bias very negative to freeze the hidden state entirely.

8 hidden units are shown as five rows of bars: input x_t, reset gate r_t, update gate z_t, candidate h̃_t, and the persistent hidden state h_t. Bar height and color both track each unit's value. The weight matrices are fixed at load (Xavier-scaled random), so what you're watching is a genuine — if untrained — GRU cell reacting to the sequence in real time.

Input x_t Reset r_t Update z_t Candidate h̃_t Hidden h_t

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.