1f8dd5463cd199ffa64ad98fe3274726ba7301bf,examples/attention_tagger.py,,main,#,138

Before Change

        prefix = HashEmbed(width // 2, 100, column=2)
        suffix = HashEmbed(width // 2, 100, column=3)

        model = (
            with_flatten(
                (lower_case | shape | prefix | suffix)
                >> Maxout(width, pieces=3), pad=0)
            >> PositionEncode(1000, width)
            >> flatten_add_lengths
            >> Residual(MultiHeadedAttention(nM=width, nH=1))
            >> unflatten
            >> with_flatten(Softmax(nr_tag))
        )

After Change

        prefix = HashEmbed(width // 2, 100, column=2)
        suffix = HashEmbed(width // 2, 100, column=3)

        model = (
            with_flatten(
                (lower_case | shape | prefix | suffix)
                >> Maxout(width, pieces=3))
            >> PositionEncode(1000, width)
            >> flatten_add_lengths
            >> Residual(
                get_qkv_self_attention(Affine(width*3, width), nM=width, nH=4)
                >> MultiHeadedAttention(nM=width, nH=4)
                >> with_getitem(0, Affine(width, width)))
            >> unflatten
            >> with_flatten(Softmax(nr_tag))
        )

Instances