Prediction-MindSpore/stock-forcast-models/models/decoder.py

import mindspore.nn as nn
import mindspore.ops.operations as ops

class DecoderLayer(nn.Cell):
    def __init__(self, self_attention, cross_attention, d_model, d_ff=None,
                 dropout=0.1, activation="relu"):
        super(DecoderLayer, self).__init__()
        d_ff = d_ff or 4 * d_model
        self.self_attention = self_attention
        self.cross_attention = cross_attention
        self.conv1 = nn.Conv1d(d_model, d_ff, kernel_size=1)
        self.conv2 = nn.Conv1d(d_ff, d_model, kernel_size=1)
        self.norm1 = nn.LayerNorm(d_model)
        self.norm2 = nn.LayerNorm(d_model)
        self.norm3 = nn.LayerNorm(d_model)
        self.dropout = nn.Dropout(dropout)
        self.activation = ops.ReLU() if activation == "relu" else ops.GELU()

    def construct(self, x, cross, x_mask=None, cross_mask=None):
        attn_output, _ = self.self_attention(x, x, x, x_mask)
        x = x + self.dropout(attn_output)
        x = self.norm1(x)

        attn_output, _ = self.cross_attention(x, cross, cross, cross_mask)
        x = x + self.dropout(attn_output)
        x = self.norm2(x)

        y = x
        y = y.transpose(1, -1)
        y = self.dropout(self.activation(self.conv1(y)))
        y = self.conv2(y).transpose(1, -1)
        y = self.dropout(y)

        x = x + y
        x = self.norm3(x)

        return x

class Decoder(nn.Cell):
    def __init__(self, layers, norm_layer=None):
        super(Decoder, self).__init__()
        self.layers = nn.LayerList(layers)
        self.norm = norm_layer

    def construct(self, x, cross, x_mask=None, cross_mask=None):
        for layer in self.layers:
            x = layer(x, cross, x_mask=x_mask, cross_mask=cross_mask)

        if self.norm is not None:
            x = self.norm(x)

        return x