请我喝杯咖啡☕
*备忘录：
我的帖子解释了 flatten() 和 ravel()。我的帖子解释了 unflatten()。
flatten() 可以通过从零个或多个元素的 0d 或多个 d 张量中选择维度来移除零个或多个维度，得到零个或多个元素的 1d 或多个 d 张量，如下所示：
*备忘录：
初始化的第一个参数是 start_dim(optional-default:1-type:int)。初始化的第二个参数是 end_dim（可选-默认：-1-类型：int）。第一个参数是输入（必需类型：int、float、complex 或 bool 的张量）。flatten() 可以将 0d 张量更改为 1d 张量。flatten() 对于一维张量没有任何作用。flatten() 和 flatten() 的区别是：flatten() 的 start_dim 默认值为 1，而 flatten() 的 start_dim 默认值为 0。基本上，flatten() 用于定义模型，而 flatten() 不用于定义模型。
import torchfrom torch import nnflatten = nn.Flatten()flatten# Flatten(start_dim=1, end_dim=-1)flatten.start_dim# 1flatten.end_dim# -1my_tensor = torch.tensor(7)flatten = nn.Flatten(start_dim=0, end_dim=0)flatten = nn.Flatten(start_dim=0, end_dim=-1)flatten = nn.Flatten(start_dim=-1, end_dim=0)flatten = nn.Flatten(start_dim=-1, end_dim=-1)flatten(input=my_tensor)# tensor([7])my_tensor = torch.tensor([7, 1, -8, 3, -6, 0])flatten = nn.Flatten(start_dim=0, end_dim=0)flatten = nn.Flatten(start_dim=0, end_dim=-1)flatten = nn.Flatten(start_dim=-1, end_dim=0)flatten = nn.Flatten(start_dim=-1, end_dim=-1)flatten(input=my_tensor)# tensor([7, 1, -8, 3, -6, 0])my_tensor = torch.tensor([[7, 1, -8], [3, -6, 0]])flatten = nn.Flatten(start_dim=0, end_dim=1)flatten = nn.Flatten(start_dim=0, end_dim=-1)flatten = nn.Flatten(start_dim=-2, end_dim=1)flatten = nn.Flatten(start_dim=-2, end_dim=-1)flatten(input=my_tensor)# tensor([7, 1, -8, 3, -6, 0])flatten = nn.Flatten()flatten = nn.Flatten(start_dim=0, end_dim=0)flatten = nn.Flatten(start_dim=-1, end_dim=-1)flatten = nn.Flatten(start_dim=0, end_dim=-2)flatten = nn.Flatten(start_dim=1, end_dim=1)flatten = nn.Flatten(start_dim=1, end_dim=-1)flatten = nn.Flatten(start_dim=-1, end_dim=1)flatten = nn.Flatten(start_dim=-1, end_dim=-1)flatten = nn.Flatten(start_dim=-2, end_dim=0)flatten = nn.Flatten(start_dim=-2, end_dim=-2)flatten(input=my_tensor)# tensor([[7, 1, -8], [3, -6, 0]])my_tensor = torch.tensor([[[7], [1], [-8]], [[3], [-6], [0]]])flatten = nn.Flatten(start_dim=0, end_dim=2)flatten = nn.Flatten(start_dim=0, end_dim=-1)flatten = nn.Flatten(start_dim=-3, end_dim=2)flatten = nn.Flatten(start_dim=-3, end_dim=-1)flatten(input=my_tensor)# tensor([7, 1, -8, 3, -6, 0])flatten = nn.Flatten(start_dim=0, end_dim=0)flatten = nn.Flatten(start_dim=0, end_dim=-3)flatten = nn.Flatten(start_dim=1, end_dim=1)flatten = nn.Flatten(start_dim=1, end_dim=-2)flatten = nn.Flatten(start_dim=2, end_dim=2)flatten = nn.Flatten(start_dim=2, end_dim=-1)flatten = nn.Flatten(start_dim=-1, end_dim=2)flatten = nn.Flatten(start_dim=-1, end_dim=-1)flatten = nn.Flatten(start_dim=-2, end_dim=1)flatten = nn.Flatten(start_dim=-2, end_dim=-2)flatten = nn.Flatten(start_dim=-3, end_dim=0)flatten = nn.Flatten(start_dim=-3, end_dim=-3)flatten(input=my_tensor)# tensor([[[7], [1], [-8]], [[3], [-6], [0]]])flatten = nn.Flatten(start_dim=0, end_dim=1)flatten = nn.Flatten(start_dim=0, end_dim=-2)flatten = nn.Flatten(start_dim=-3, end_dim=1)flatten = nn.Flatten(start_dim=-3, end_dim=-2)flatten(input=my_tensor)# tensor([[7], [1], [-8], [3], [-6], [0]])flatten = nn.Flatten()flatten = nn.Flatten(start_dim=1, end_dim=2)flatten = nn.Flatten(start_dim=1, end_dim=-1)flatten = nn.Flatten(start_dim=-2, end_dim=2)flatten = nn.Flatten(start_dim=-2, end_dim=-1)flatten(input=my_tensor)# tensor([[7, 1, -8], [3, -6, 0]])my_tensor = torch.tensor([[[7.], [1.], [-8.]], [[3.], [-6.], [0.]]])flatten = nn.Flatten()flatten(input=my_tensor)# tensor([[7., 1., -8.], [3., -6., 0.]])my_tensor = torch.tensor([[[7.+0.j], [1.+0.j], [-8.+0.j]], [[3.+0.j], [-6.+0.j], [0.+0.j]]])flatten = nn.Flatten()flatten(input=my_tensor)# tensor([[7.+0.j, 1.+0.j, -8.+0.j],# [3.+0.j, -6.+0.j, 0.+0.j]])my_tensor = torch.tensor([[[True], [False], [True]], [[False], [True], [False]]])flatten = nn.Flatten()flatten(input=my_tensor)# tensor([[True, False, True],# [False, True, False]])
以上就是在 PyTorch 中展平的详细内容，更多请关注创想鸟其它相关文章！
版权声明：本文内容由互联网用户自发贡献，该文观点仅代表作者本人。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， 请发送邮件至 chuangxiangniao@163.com 举报，一经查实，本站将立刻删除。
发布者：程序猿，转转请注明出处：https://www.chuangxiangniao.com/p/1351550.html
微信扫一扫 
支付宝扫一扫 
