From 1420b27cfb4f7519b09bbb304e9481ba992c58b9 Mon Sep 17 00:00:00 2001
From: Xv-M-S <1205507925@qq.com>
Date: Wed, 4 Sep 2024 17:24:33 +0800
Subject: [PATCH] =?UTF-8?q?MindSpore2.3.1=E7=89=88=E6=9C=AC=E5=9C=A8?=
 =?UTF-8?q?=E5=90=AF=E6=99=BA=E7=A4=BE=E5=8C=BA=E6=98=87=E8=85=BE=E7=8E=AF?=
 =?UTF-8?q?=E5=A2=83=E7=9A=84=E9=80=82=E9=85=8D,=E5=88=9D=E5=AD=A6?=
 =?UTF-8?q?=E6=95=99=E7=A8=8B=E4=BF=AE=E6=94=B9?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../beginner/accelerate_with_static_graph.md  | 30 +++++++++++--------
 tutorials/source_en/beginner/autograd.md      |  8 ++---
 tutorials/source_en/beginner/model.md         |  7 +++--
 tutorials/source_en/beginner/quick_start.md   |  5 +++-
 tutorials/source_en/beginner/train.md         |  4 ++-
 .../accelerate_with_static_graph.ipynb        | 30 +++++++++++--------
 .../source_zh_cn/beginner/autograd.ipynb      |  8 ++---
 tutorials/source_zh_cn/beginner/model.ipynb   |  7 +++--
 .../source_zh_cn/beginner/quick_start.ipynb   |  5 +++-
 tutorials/source_zh_cn/beginner/train.ipynb   |  8 +++--
 10 files changed, 70 insertions(+), 42 deletions(-)

diff --git a/tutorials/source_en/beginner/accelerate_with_static_graph.md b/tutorials/source_en/beginner/accelerate_with_static_graph.md
index 7d20ca3aa1..551cb10073 100644
--- a/tutorials/source_en/beginner/accelerate_with_static_graph.md
+++ b/tutorials/source_en/beginner/accelerate_with_static_graph.md
@@ -39,7 +39,8 @@ class Network(nn.Cell):
         return logits
 
 model = Network()
-input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))
+model.to_float(ms.float16)
+input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))
 output = model(input)
 print(output)
 ```
@@ -98,7 +99,8 @@ class Network(nn.Cell):
         return logits
 
 model = Network()
-input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))
+model.to_float(ms.float16)
+input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))
 output = model(input)
 print(output)
 ```
@@ -161,14 +163,15 @@ class Network(nn.Cell):
         logits = self.dense_relu_sequential(x)
         return logits
 
-input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))
+input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))
 
 @ms.jit  # Use the ms.jit decorator to make the decorated function run in static graph mode
-def run(x):
-    model = Network()
+def run(model, x):
     return model(x)
 
-output = run(input)
+model = Network()
+model.to_float(ms.float16)
+output = run(model, input)
 print(output)
 ```
 
@@ -216,14 +219,15 @@ class Network(nn.Cell):
         logits = self.dense_relu_sequential(x)
         return logits
 
-input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))
+input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))
 
-def run(x):
-    model = Network()
+def run(model, x):
     return model(x)
 
+model = Network()
+model.to_float(ms.float16)
 run_with_jit = ms.jit(run)  # Transforming a function to execute as a static graph by calling jit
-output = run_with_jit(input)
+output = run_with_jit(model, input)
 print(output)
 ```
 
@@ -272,8 +276,9 @@ class Network(nn.Cell):
         logits = self.dense_relu_sequential(x)
         return logits
 
-input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))
+input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))
 model = Network()
+model.to_float(ms.float16)
 output = model(input)
 print(output)
 ```
@@ -326,7 +331,8 @@ class Network(nn.Cell):
         return logits
 
 model = Network()
-input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))
+model.to_float(ms.float16)
+input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))
 output = model(input)
 print(output)
 ```
diff --git a/tutorials/source_en/beginner/autograd.md b/tutorials/source_en/beginner/autograd.md
index fa142dde22..01b7b9b66c 100644
--- a/tutorials/source_en/beginner/autograd.md
+++ b/tutorials/source_en/beginner/autograd.md
@@ -25,10 +25,10 @@ Computing graphs are a way to represent mathematical functions in a graph-theore
 In this model, $x$ is the input, $y$ is the correct value, and $w$ and $b$ are the parameters we need to optimize.
 
 ```python
-x = ops.ones(5, mindspore.float32)  # input tensor
-y = ops.zeros(3, mindspore.float32)  # expected output
-w = Parameter(Tensor(np.random.randn(5, 3), mindspore.float32), name='w') # weight
-b = Parameter(Tensor(np.random.randn(3,), mindspore.float32), name='b') # bias
+x = ops.ones(5, mindspore.float16)  # input tensor
+y = ops.zeros(3, mindspore.float16)  # expected output
+w = Parameter(Tensor(np.random.randn(5, 3), mindspore.float16), name='w') # weight
+b = Parameter(Tensor(np.random.randn(3,), mindspore.float16), name='b') # bias
 ```
 
 We construct the computing function based on the computing process described by the computing graphs.
diff --git a/tutorials/source_en/beginner/model.md b/tutorials/source_en/beginner/model.md
index db1aa65ace..78d5758136 100644
--- a/tutorials/source_en/beginner/model.md
+++ b/tutorials/source_en/beginner/model.md
@@ -42,6 +42,7 @@ After completing construction, instantiate the `Network` object and look at its
 
 ```python
 model = Network()
+model.to_float(mindspore.float16)
 print(model)
 ```
 
@@ -63,7 +64,7 @@ We construct an input data and call the model directly to obtain a two-dimension
 > The `model.construct()` method cannot be called directly.
 
 ```python
-X = ops.ones((1, 28, 28), mindspore.float32)
+X = ops.ones((1, 28, 28), mindspore.float16)
 logits = model(X)
 # print logits
 logits
@@ -91,7 +92,7 @@ Predicted class: [4]
 In this section, we decompose each layer of the neural network model constructed in the previous section. First we construct a random data (3 images of 28x28) with shape (3, 28, 28) and pass through each neural network layer in turn to observe its effect.
 
 ```python
-input_image = ops.ones((3, 28, 28), mindspore.float32)
+input_image = ops.ones((3, 28, 28), mindspore.float16)
 print(input_image.shape)
 ```
 
@@ -119,6 +120,7 @@ print(flat_image.shape)
 
 ```python
 layer1 = nn.Dense(in_channels=28*28, out_channels=20)
+layer1.to_float(mindspore.float16)
 hidden1 = layer1(flat_image)
 print(hidden1.shape)
 ```
@@ -178,6 +180,7 @@ seq_modules = nn.SequentialCell(
     nn.Dense(20, 10)
 )
 
+seq_modules.to_float(mindspore.float16)
 logits = seq_modules(input_image)
 print(logits.shape)
 ```
diff --git a/tutorials/source_en/beginner/quick_start.md b/tutorials/source_en/beginner/quick_start.md
index 26ea0701f1..2db0e6c494 100644
--- a/tutorials/source_en/beginner/quick_start.md
+++ b/tutorials/source_en/beginner/quick_start.md
@@ -72,7 +72,8 @@ def datapipe(dataset, batch_size):
     image_transforms = [
         vision.Rescale(1.0 / 255.0, 0),
         vision.Normalize(mean=(0.1307,), std=(0.3081,)),
-        vision.HWC2CHW()
+        vision.HWC2CHW(),
+        transforms.TypeCast(mindspore.float16) 
     ]
 
     label_transform = transforms.TypeCast(mindspore.int32)
@@ -140,6 +141,7 @@ class Network(nn.Cell):
         return logits
 
 model = Network()
+model.to_float(mindspore.float16)
 print(model)
 ```
 
@@ -306,6 +308,7 @@ There are two steps to load the saved weights:
 ```python
 # Instantiate a random initialized model
 model = Network()
+model.to_float(mindspore.float16)
 # Load checkpoint and load parameter to model
 param_dict = mindspore.load_checkpoint("model.ckpt")
 param_not_load, _ = mindspore.load_param_into_net(model, param_dict)
diff --git a/tutorials/source_en/beginner/train.md b/tutorials/source_en/beginner/train.md
index 98bda268d7..ca0e2b1849 100644
--- a/tutorials/source_en/beginner/train.md
+++ b/tutorials/source_en/beginner/train.md
@@ -35,7 +35,8 @@ def datapipe(path, batch_size):
     image_transforms = [
         vision.Rescale(1.0 / 255.0, 0),
         vision.Normalize(mean=(0.1307,), std=(0.3081,)),
-        vision.HWC2CHW()
+        vision.HWC2CHW(),
+        transforms.TypeCast(mindspore.float16)
     ]
     label_transform = transforms.TypeCast(mindspore.int32)
 
@@ -80,6 +81,7 @@ class Network(nn.Cell):
         return logits
 
 model = Network()
+model.to_float(mindspore.float16)
 ```
 
 ## Defining the Hyperparameter, Loss Function and Optimizer
diff --git a/tutorials/source_zh_cn/beginner/accelerate_with_static_graph.ipynb b/tutorials/source_zh_cn/beginner/accelerate_with_static_graph.ipynb
index 0c1018e20c..9c74996e3e 100644
--- a/tutorials/source_zh_cn/beginner/accelerate_with_static_graph.ipynb
+++ b/tutorials/source_zh_cn/beginner/accelerate_with_static_graph.ipynb
@@ -79,7 +79,8 @@
     "        return logits\n",
     "\n",
     "model = Network()\n",
-    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))\n",
+    "model.to_float(ms.float16)\n",
+    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))\n",
     "output = model(input)\n",
     "print(output)"
    ]
@@ -152,7 +153,8 @@
     "        return logits\n",
     "\n",
     "model = Network()\n",
-    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))\n",
+    "model.to_float(ms.float16)\n",
+    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))\n",
     "output = model(input)\n",
     "print(output)"
    ]
@@ -231,14 +233,15 @@
     "        logits = self.dense_relu_sequential(x)\n",
     "        return logits\n",
     "\n",
-    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))\n",
+    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))\n",
     "\n",
     "@ms.jit  # 使用ms.jit装饰器，使被装饰的函数以静态图模式运行\n",
-    "def run(x):\n",
-    "    model = Network()\n",
+    "def run(model, x):\n",
     "    return model(x)\n",
     "\n",
-    "output = run(input)\n",
+    "model = Network()\n",
+    "model.to_float(ms.float16)\n",
+    "output = run(model, input)\n",
     "print(output)"
    ]
   },
@@ -302,14 +305,15 @@
     "        logits = self.dense_relu_sequential(x)\n",
     "        return logits\n",
     "\n",
-    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))\n",
+    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))\n",
     "\n",
-    "def run(x):\n",
-    "    model = Network()\n",
+    "def run(model, x):\n",
     "    return model(x)\n",
     "\n",
+    "model = Network()\n",
+    "model.to_float(ms.float16)\n",
     "run_with_jit = ms.jit(run)  # 通过调用jit将函数转换为以静态图方式执行\n",
-    "output = run_with_jit(input)\n",
+    "output = run_with_jit(model, input)\n",
     "print(output)"
    ]
   },
@@ -374,8 +378,9 @@
     "        logits = self.dense_relu_sequential(x)\n",
     "        return logits\n",
     "\n",
-    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))\n",
+    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))\n",
     "model = Network()\n",
+    "model.to_float(ms.float16)\n",
     "output = model(input)\n",
     "print(output)"
    ]
@@ -444,7 +449,8 @@
     "        return logits\n",
     "\n",
     "model = Network()\n",
-    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float32))\n",
+    "model.to_float(ms.float16)\n",
+    "input = Tensor(np.ones([64, 1, 28, 28]).astype(np.float16))\n",
     "output = model(input)\n",
     "print(output)"
    ]
diff --git a/tutorials/source_zh_cn/beginner/autograd.ipynb b/tutorials/source_zh_cn/beginner/autograd.ipynb
index 8bacbd7fdb..07217a6d61 100644
--- a/tutorials/source_zh_cn/beginner/autograd.ipynb
+++ b/tutorials/source_zh_cn/beginner/autograd.ipynb
@@ -57,10 +57,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "x = ops.ones(5, mindspore.float32)  # input tensor\n",
-    "y = ops.zeros(3, mindspore.float32)  # expected output\n",
-    "w = Parameter(Tensor(np.random.randn(5, 3), mindspore.float32), name='w') # weight\n",
-    "b = Parameter(Tensor(np.random.randn(3,), mindspore.float32), name='b') # bias"
+    "x = ops.ones(5, mindspore.float16)  # input tensor\n",
+    "y = ops.zeros(3, mindspore.float16)  # expected output\n",
+    "w = Parameter(Tensor(np.random.randn(5, 3), mindspore.float16), name='w') # weight\n",
+    "b = Parameter(Tensor(np.random.randn(3,), mindspore.float16), name='b') # bias"
    ]
   },
   {
diff --git a/tutorials/source_zh_cn/beginner/model.ipynb b/tutorials/source_zh_cn/beginner/model.ipynb
index 2db7b97b0f..313bf74044 100644
--- a/tutorials/source_zh_cn/beginner/model.ipynb
+++ b/tutorials/source_zh_cn/beginner/model.ipynb
@@ -121,6 +121,7 @@
    ],
    "source": [
     "model = Network()\n",
+    "model.to_float(mindspore.float16)\n",
     "print(model)"
    ]
   },
@@ -151,7 +152,7 @@
     }
    ],
    "source": [
-    "X = ops.ones((1, 28, 28), mindspore.float32)\n",
+    "X = ops.ones((1, 28, 28), mindspore.float16)\n",
     "logits = model(X)\n",
     "# print logits\n",
     "logits"
@@ -213,7 +214,7 @@
     }
    ],
    "source": [
-    "input_image = ops.ones((3, 28, 28), mindspore.float32)\n",
+    "input_image = ops.ones((3, 28, 28), mindspore.float16)\n",
     "print(input_image.shape)"
    ]
   },
@@ -285,6 +286,7 @@
    ],
    "source": [
     "layer1 = nn.Dense(in_channels=28*28, out_channels=20)\n",
+    "layer1.to_float(mindspore.float16)\n",
     "hidden1 = layer1(flat_image)\n",
     "print(hidden1.shape)"
    ]
@@ -391,6 +393,7 @@
     "    nn.Dense(20, 10)\n",
     ")\n",
     "\n",
+    "seq_modules.to_float(mindspore.float16)\n",
     "logits = seq_modules(input_image)\n",
     "print(logits.shape)"
    ]
diff --git a/tutorials/source_zh_cn/beginner/quick_start.ipynb b/tutorials/source_zh_cn/beginner/quick_start.ipynb
index a07153387a..eb52799dec 100644
--- a/tutorials/source_zh_cn/beginner/quick_start.ipynb
+++ b/tutorials/source_zh_cn/beginner/quick_start.ipynb
@@ -138,7 +138,8 @@
     "    image_transforms = [\n",
     "        vision.Rescale(1.0 / 255.0, 0),\n",
     "        vision.Normalize(mean=(0.1307,), std=(0.3081,)),\n",
-    "        vision.HWC2CHW()\n",
+    "        vision.HWC2CHW(),\n",
+    "        transforms.TypeCast(mindspore.float16) \n",
     "    ]\n",
     "    label_transform = transforms.TypeCast(mindspore.int32)\n",
     "\n",
@@ -266,6 +267,7 @@
     "        return logits\n",
     "\n",
     "model = Network()\n",
+    "model.to_float(mindspore.float16)\n",
     "print(model)"
    ]
   },
@@ -510,6 +512,7 @@
    "source": [
     "# Instantiate a random initialized model\n",
     "model = Network()\n",
+    "model.to_float(mindspore.float16)\n",
     "# Load checkpoint and load parameter to model\n",
     "param_dict = mindspore.load_checkpoint(\"model.ckpt\")\n",
     "param_not_load, _ = mindspore.load_param_into_net(model, param_dict)\n",
diff --git a/tutorials/source_zh_cn/beginner/train.ipynb b/tutorials/source_zh_cn/beginner/train.ipynb
index b11e86e108..3b8bbaeb60 100644
--- a/tutorials/source_zh_cn/beginner/train.ipynb
+++ b/tutorials/source_zh_cn/beginner/train.ipynb
@@ -85,7 +85,8 @@
     "    image_transforms = [\n",
     "        vision.Rescale(1.0 / 255.0, 0),\n",
     "        vision.Normalize(mean=(0.1307,), std=(0.3081,)),\n",
-    "        vision.HWC2CHW()\n",
+    "        vision.HWC2CHW(),\n",
+    "        transforms.TypeCast(mindspore.float16)\n",
     "    ]\n",
     "    label_transform = transforms.TypeCast(mindspore.int32)\n",
     "\n",
@@ -139,7 +140,8 @@
     "        logits = self.dense_relu_sequential(x)\n",
     "        return logits\n",
     "\n",
-    "model = Network()"
+    "model = Network()\n",
+    "model.to_float(mindspore.float16)"
    ]
   },
   {
@@ -464,4 +466,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
\ No newline at end of file
+}
-- 
Gitee