admin管理员组

文章数量:1336424

I want to add a pooling layer on my model. I want to preserve the num of node as I m doing node prediction without batching.

import torch
from torch_geometric.nn import global_add_pool
# Assume your input tensor is x with shape (num_nodes, num_node_features)
x = torch.randn(10, 10)  # Example tensor with (num_nodes, num_node_features) = (10, 10)
# Apply global_add_pool
x_pooled = global_add_pool(x, batch=None)

x_pooled (1,10) But I want a pooling method to give me (10,x)

本文标签： pytorchGraph Neural Network Pooling for keeping the same shape of node without batchingStack Overflow

版权声明：本文标题：pytorch - Graph Neural Network Pooling for keeping the same shape of node without batching - Stack Overflow 内容由网友自发贡献，该文观点仅代表作者本人， 转载请联系作者并注明出处：http://www.betaflare.com/web/1742404689a2468579.html， 本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容，一经查实，本站将立刻删除。