2、PyTorch张量的运算API(上)

1. 教学视频

2、PyTorch张量的运算API(上)

  • 因比较忙,暂时就做个过场吧。

2. Python代码

  • Python
python 复制代码
#!/usr/bin/env python
# -*- coding:utf-8 -*-
# @FileName  :torch_learn2.py
# @Time      :2024/11/16 19:53
# @Author    :Jason Zhang

import torch

torch.manual_seed(12124)

if __name__ == "__main__":
    run_code = 0
    a = torch.rand([3, 2])
    a_chunk1, a_chunk2 = torch.chunk(a, chunks=2)
    a_chunk11, a_chunk21 = torch.chunk(a, chunks=2, dim=1)
    print(f"a=\n{a}")
    print(f"a_chunk1=\n{a_chunk1}")
    print(f"a_chunk2=\n{a_chunk2}")
    print(f"a_chunk11=\n{a_chunk11}")
    print(f"a_chunk21=\n{a_chunk21}")
    t = torch.tensor([[1, 2], [3, 4]])
    t_gather = torch.gather(t, 1, torch.tensor([[0, 1], [1, 0]]))
    print(f"t=\n{t}")
    print(f"t_gather=\n{t_gather}")
    reshape_12 = torch.arange(12).reshape((3, 4))
    print(f"reshape_12=\n{reshape_12}")
    reshape_11 = reshape_12.reshape((-1, 1))
    print(f"reshape_11=\n{reshape_11}")
    src = torch.arange(1, 11).reshape((2, 5))
    index = torch.tensor([[0, 1, 2, 0]])
    y = torch.zeros(3, 5, dtype=src.dtype).scatter_(0, index, src)
    print(f"src=\n{src}")
    print(f"y=\n{y}")
    stack_a = torch.rand((3, 4))
    stack_b = torch.rand((3, 4))
    stack_ab = torch.stack((stack_a, stack_b))
    print(f"stack_a=\n{stack_a}")
    print(f"stack_b=\n{stack_b}")
    print(f"stack_ab=\n{stack_ab},shape={stack_ab.shape}")
    squeeze_1 = torch.rand((2, 1, 3))
    squeeze_2 = torch.squeeze(squeeze_1)
    print(f"squeeze_1.shape={squeeze_1.shape}")
    print(f"squeeze_2.shape={squeeze_2.shape}")
  • 结果:
python 复制代码
a=
tensor([[0.5555, 0.0484],
        [0.3199, 0.2577],
        [0.8874, 0.6888]])
a_chunk1=
tensor([[0.5555, 0.0484],
        [0.3199, 0.2577]])
a_chunk2=
tensor([[0.8874, 0.6888]])
a_chunk11=
tensor([[0.5555],
        [0.3199],
        [0.8874]])
a_chunk21=
tensor([[0.0484],
        [0.2577],
        [0.6888]])
t=
tensor([[1, 2],
        [3, 4]])
t_gather=
tensor([[1, 2],
        [4, 3]])
reshape_12=
tensor([[ 0,  1,  2,  3],
        [ 4,  5,  6,  7],
        [ 8,  9, 10, 11]])
reshape_11=
tensor([[ 0],
        [ 1],
        [ 2],
        [ 3],
        [ 4],
        [ 5],
        [ 6],
        [ 7],
        [ 8],
        [ 9],
        [10],
        [11]])
src=
tensor([[ 1,  2,  3,  4,  5],
        [ 6,  7,  8,  9, 10]])
y=
tensor([[1, 0, 0, 4, 0],
        [0, 2, 0, 0, 0],
        [0, 0, 3, 0, 0]])
stack_a=
tensor([[0.2410, 0.9222, 0.5832, 0.3587],
        [0.9344, 0.3320, 0.3852, 0.3239],
        [0.7664, 0.9575, 0.2645, 0.5601]])
stack_b=
tensor([[0.4304, 0.7509, 0.3536, 0.7229],
        [0.9026, 0.0793, 0.3076, 0.3272],
        [0.4434, 0.2406, 0.7080, 0.9304]])
stack_ab=
tensor([[[0.2410, 0.9222, 0.5832, 0.3587],
         [0.9344, 0.3320, 0.3852, 0.3239],
         [0.7664, 0.9575, 0.2645, 0.5601]],

        [[0.4304, 0.7509, 0.3536, 0.7229],
         [0.9026, 0.0793, 0.3076, 0.3272],
         [0.4434, 0.2406, 0.7080, 0.9304]]]),shape=torch.Size([2, 3, 4])
squeeze_1.shape=torch.Size([2, 1, 3])
squeeze_2.shape=torch.Size([2, 3])

Process finished with exit code 0
相关推荐
山烛13 分钟前
KNN 算法中的各种距离:从原理到应用
人工智能·python·算法·机器学习·knn·k近邻算法·距离公式
guozhetao26 分钟前
【ST表、倍增】P7167 [eJOI 2020] Fountain (Day1)
java·c++·python·算法·leetcode·深度优先·图论
墨染点香32 分钟前
第七章 Pytorch构建模型详解【构建CIFAR10模型结构】
人工智能·pytorch·python
阿什么名字不会重复呢1 小时前
在线工具+网页平台来学习和操作Python与Excel相关技能
python·数据分析
Vertira1 小时前
python 阿里云 安装 dashscope的简介、安装
开发语言·python
gc_22992 小时前
学习Python中Selenium模块的基本用法(1:简介)
python·selenium
先做个垃圾出来………3 小时前
2116. 判断一个括号字符串是否有效
python
兮℡檬,3 小时前
房价预测|Pytorch
人工智能·pytorch·python
im_AMBER7 小时前
学习日志19 python
python·学习
mortimer10 小时前
安装NVIDIA Parakeet时,我遇到的两个Pip“小插曲”
python·github