如何从 PyTorch split() 获取张量

发布于 2025-01-11 17:33:35 字数 368 浏览 0 评论 0原文

PyTorch 的 split 函数返回张量元组。但我需要批量矩阵乘以结果。有没有一种简单的方法来分割张量并返回张量？这就是我尝试过的：

m = [[2, 3, 5, 7],
     [11, 13, 17, 19],
     [23, 29, 31, 37],
     [41, 43, 47, 53]]
m_split = torch.tensor(m).split(2, dim=1)
torch.tensor([[[2, 3, 5, 7]]]).matmul(m_split)

这给了我一个错误，因为 m_split 是张量的元组而不是张量。我可以进行 view 或 reshape 调用吗？

原文

PyTorch's split function returns back a tuple of tensors. But I need to batch matrix multiply the result. Is there an easy way to split a tensor and get back a tensor? This is what I tried:

m = [[2, 3, 5, 7],
     [11, 13, 17, 19],
     [23, 29, 31, 37],
     [41, 43, 47, 53]]
m_split = torch.tensor(m).split(2, dim=1)
torch.tensor([[[2, 3, 5, 7]]]).matmul(m_split)

This gives me an error because m_split is a tuple of tensors rather than being a tensor. Is there a view or reshape call I can make instead?

分享到QQ

分享到微博

如果你对这篇内容有疑问，欢迎到本站社区发帖提问参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。

发布评论

需要登录才能够评论，你可以免费注册一个本站的账号。

做个ˇ局外人 2025-01-18 17:33:35

我认为你可以这样做

m = [[2, 3, 5, 7],
     [11, 13, 17, 19],
     [23, 29, 31, 37],
     [41, 43, 47, 53]]
m_split = torch.tensor(m).tensor_split(2, dim=1)
m_split=torch.stack(list(m_split), dim=0)
torch.tensor([[[2, 3, 5, 7]]]).matmul(m_split)

i think you can do as following

m = [[2, 3, 5, 7],
     [11, 13, 17, 19],
     [23, 29, 31, 37],
     [41, 43, 47, 53]]
m_split = torch.tensor(m).tensor_split(2, dim=1)
m_split=torch.stack(list(m_split), dim=0)
torch.tensor([[[2, 3, 5, 7]]]).matmul(m_split)

回复收藏 0 原文

~没有更多了~