SATRNEncoderLayer¶

class mmocr.models.textrecog.SATRNEncoderLayer(d_model=512, d_inner=512, n_head=8, d_k=64, d_v=64, dropout=0.1, qkv_bias=False, init_cfg=None)[source]¶

Implement encoder layer for SATRN, see `SATRN.

<https://arxiv.org/abs/1910.04396>`_.

Parameters

d_model (int) – Dimension \(D_m\) of the input from previous model. Defaults to 512.
d_inner (int) – Hidden dimension of feedforward layers. Defaults to 256.
n_head (int) – Number of parallel attention heads. Defaults to 8.
d_k (int) – Dimension of the key vector. Defaults to 64.
d_v (int) – Dimension of the value vector. Defaults to 64.
dropout (float) – Dropout rate. Defaults to 0.1.
qkv_bias (bool) – Whether to use bias. Defaults to False.
init_cfg (dict or list[dict], optional) – Initialization configs. Defaults to None.

Return type

None

forward(x, h, w, mask=None)[source]¶

Forward propagation of encoder.

Parameters

x (Tensor) – Feature tensor of shape \((N, h*w, D_m)\).
h (int) – Height of the original feature.
w (int) – Width of the original feature.
mask (Tensor, optional) – Mask used for masked multi-head attention. Defaults to None.

Returns

A tensor of shape \((N, h*w, D_m)\).

Return type

Tensor