GRUCell

class brainpy.dyn.GRUCell(num_in, num_out, Wi_initializer=Orthogonal(scale=1.0, axis=-1, rng=[2233284200 4000014544]), Wh_initializer=Orthogonal(scale=1.0, axis=-1, rng=[2233284200 4000014544]), b_initializer=ZeroInit, state_initializer=ZeroInit, activation='tanh', mode=None, train_state=False, name=None)

Gated Recurrent Unit.

The implementation is based on (Chung, et al., 2014) [1] with biases.

Given \(x_t\) and the previous state \(h_{t-1}\) the core computes

\[\begin{split}\begin{array}{ll} z_t &= \sigma(W_{iz} x_t + W_{hz} h_{t-1} + b_z) \\ r_t &= \sigma(W_{ir} x_t + W_{hr} h_{t-1} + b_r) \\ a_t &= \tanh(W_{ia} x_t + W_{ha} (r_t \bigodot h_{t-1}) + b_a) \\ h_t &= (1 - z_t) \bigodot h_{t-1} + z_t \bigodot a_t \end{array}\end{split}\]

where \(z_t\) and \(r_t\) are reset and update gates.

The output is equal to the new hidden state, \(h_t\).

Warning: Backwards compatibility of GRU weights is currently unsupported.

Parameters:

• num_in (int) – The dimension of the input vector

• num_out (int) – The number of hidden unit in the node.

• state_initializer (callable, Initializer, bm.ndarray, jax.numpy.ndarray) – The state initializer.

• Wi_initializer (callable, Initializer, bm.ndarray, jax.numpy.ndarray) – The input weight initializer.

• Wh_initializer (callable, Initializer, bm.ndarray, jax.numpy.ndarray) – The hidden weight initializer.

• b_initializer (optional, callable, Initializer, bm.ndarray, jax.numpy.ndarray) – The bias weight initializer.

• activation (str, callable) – The activation function. It can be a string or a callable function. See brainpy.math.activations for more details.

References

[1]

Chung, J., Gulcehre, C., Cho, K. and Bengio, Y., 2014. Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555.

update(x)

The function to specify the updating rule.