qq_37418807的博客

RGCNConvfrom torch_geometric.nn import RGCNConv,GraphConvimport torch.nn as nnimport torchnum_features = 64hidden_size = 16conv1 = RGCNConv(in_channels = num_features,out_channels = hidden_size,num_relations = 3,num_bases = 10)print('RGCNConv模型里的参

输入:features:[num_seqs,batch_size,2*D_e]lengths:[batch_size]edge_ind:batch_size个[(a,b)…]输出:scores:[batch_size,max_nodes,num_seqs(node_number)]代码demo:import torch.nn as nnimport torch.nn.functional as Ffrom torch.autograd import Variableimport sys

输入features:[batch_size*len(j),D_e]edge_index:[2,num_edges]edge_norm:[num_edges]edge_type:[num_edges]seq_lengths:[batch_size]umask:[batch_size,num_seqs]输出:log_prob:[batch_size*len(j),num_classes]代码import torchfrom torch_geometric.nn import RGCNCo

代码断点:features, edge_index, edge_norm, edge_type, edge_index_lengths = batch_graphify(      emotions,       qmask,       seq_lengths,       self.window_past,        self.window_future,  

from model import MaskedEdgeAttention,GraphNetwork,batch_graphifyfrom train_DailyDialog import process_data_loader,get_DailyDialogue_loadersimport torch.nn as nnimport datetime as dtimport numpy as npimport torchimport argparseseed = 100path = './s