02pytorch基础语法

机器学习/02pytorch基础语法/创建一个浮点数张量.py

@@ -0,0 +1,5 @@
+import torch
+
+# 在100~1000中随机整数
+t4 = torch.FloatTensor(2, 3, 4)
+print(t4)

机器学习/02pytorch基础语法/创建一个空张量.py

@@ -0,0 +1,4 @@
+import torch
+
+t4 = torch.empty((2, 2, 3))
+print(t4)

机器学习/02pytorch基础语法/创建值1填充的张量.py

@@ -0,0 +1,4 @@
+import torch
+
+t4 = torch.ones((2, 2, 3))
+print(t4)

机器学习/02pytorch基础语法/创建随机张量.py

@@ -0,0 +1,5 @@
+import torch
+
+# 在100~1000中随机整数
+t4 = torch.randint(low=100, high=1000, size=(2, 3, 4))
+print(t4)

机器学习/02pytorch基础语法/判断用的是cpu运算还是gpu运算.py

@@ -0,0 +1,4 @@
+import torch
+
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+print(device)

机器学习/02pytorch基础语法/张量.py

@@ -0,0 +1,13 @@
+import torch
+import numpy as np
+
+list_1 = [
+    [1, 2, 3, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+]
+print(list_1)
+print(np.array(list_1))
+
+# 张量
+print(torch.tensor(list_1))

机器学习/02pytorch基础语法/张量和nump数组之间转换.py

@@ -0,0 +1,8 @@
+import torch
+
+
+t5 = torch.randint(low=100, high=1000, size=(2, 3, 4))
+print(t5.numpy())
+
+t6 = torch.tensor([[[[[[[[[[[100]]]]]]]]]]])
+print(t6.item())

机器学习/02pytorch基础语法/张量数据合并拉升.py

@@ -0,0 +1,17 @@
+import torch
+
+
+list_8 = [[[792, 436, 928, 303],
+           [809, 170, 778, 652],
+           [967, 520, 419, 184]],
+
+          [[402, 754, 327, 979],
+           [713, 926, 879, 934],
+           [540, 953, 209, 369]]]
+t8 = torch.tensor(list_8)
+
+# 合并成单列单行
+print(t8.view(-1))
+
+# 拉升成4行6列
+print(t8.view(4, 6))

机器学习/02pytorch基础语法/张量求最大值.py

@@ -0,0 +1,17 @@
+import torch
+
+list_7 = [
+    [792, 436, 928, 303],
+    [809, 170, 778, 652],
+    [967, 520, 419, 184]
+]
+t7 = torch.tensor(list_7)
+
+# 求最大值
+print(t7.max())
+
+# 列求最大值
+print(t7.max(dim=0))
+
+# 行求最大值
+print(t7.max(dim=1))

机器学习/02pytorch基础语法/张量的切片操作.py

@@ -0,0 +1,14 @@
+import torch
+
+list_10 = [[[792, 436, 928, 303],
+            [809, 170, 778, 652],
+            [967, 520, 419, 184]],
+           [[402, 754, 327, 979],
+            [713, 926, 879, 934],
+            [540, 953, 209, 369]]]
+t10 = torch.tensor(list_10)
+
+# 切第一个维度为1，其他维度不切
+print(t10[1:, :, :])
+
+print(t10[1:, 2:, 3].item())

机器学习/02pytorch基础语法/张量的计算.py

@@ -0,0 +1,11 @@
+import torch
+
+list_10 = [[[792, 436, 928, 303],
+            [809, 170, 778, 652],
+            [967, 520, 419, 184]],
+           [[402, 754, 327, 979],
+            [713, 926, 879, 934],
+            [540, 953, 209, 369]]]
+t10 = torch.tensor(list_10)
+
+print(t10*100)

机器学习/02pytorch基础语法/张量的轴交换.py

@@ -0,0 +1,17 @@
+import torch
+
+list_9 = [[402, 754, 327, 979],
+          [713, 926, 879, 934],
+          [540, 953, 209, 369]]
+t9 = torch.tensor(list_9)
+
+# 轴交换，列变行，行变成列
+print(t9.size())
+# 1和0代表维度顺序
+t9 = t9.permute(1, 0)
+print(t9)
+print(t9.size())
+
+# 原维度是(2, 3, 4)，改变下标顺序(1, 0, 2)，变成了([3, 2, 4])
+t10 = torch.randint(low=100, high=1000, size=(2, 3, 4))
+print(t10.permute(1, 0, 2).size())

机器学习/02pytorch基础语法/数据类型转换一.py

@@ -0,0 +1,9 @@
+import torch
+
+list_2 = [
+    [127, 128, 129, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+]
+t2 = torch.tensor(list_2, dtype=torch.int8)  # [-128 - 127]
+print(t2)

机器学习/02pytorch基础语法/数据类型转换二.py

@@ -0,0 +1,12 @@
+import torch
+
+
+list_2 = [
+    [127, 128, 3, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+]
+t2 = torch.tensor(list_2)  # [-128 - 127]
+print(t2.int())
+print(t2.double())
+print(t2.double().dtype)

机器学习/02pytorch基础语法/获取张量形状.py

@@ -0,0 +1,11 @@
+import torch
+
+list_3 = [
+    [127, 128, 3, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+    [1, 2, 3, 4, 5, 6],
+]
+t3 = torch.tensor(list_3)  # [-128 - 127]
+print(t3.shape)
+print(t3.size())
+