def add_attention(self, input_name, attention_num, from_seq_length, to_seq_length, output_name): layer = caffe_net.LayerParameter(name=output_name, type='Attention', bottom=[input_name], top=[output_name]) layer.attention_param(attention_num, from_seq_length, to_seq_length) self.caffe_model.add_layer(layer) self.data_dict[output_name] = Operators.attention(self.data_dict[input_name], attention_num, from_seq_length, to_seq_length, output_name) return output_name
def add_attention(self, input_name, output_name): layer = caffe_net.LayerParameter(name=output_name, type='Attention', bottom=[input_name], top=[output_name]) layer.attention_param(self.num_attentions) self.caffe_model.add_layer(layer) self.data_dict[output_name] = Operators.attention(self.data_dict[input_name], self.num_attentions, output_name)