tensor

Author: yu2nshuai

README.md

@@ -1 +1,2 @@
-# TensorFlow2_Keras_API_Jupyter_demo
+# TensorFlow2_Keras_API_Jupyter_demo
+Tensorflow2 Keras 官方API文档demo，Jupyter版本学习笔记整理，在官方demo的基础上，丰富了一些样例。

tensor.ipynb

@@ -152,12 +152,255 @@
     "np.array(rank_2_tensor)"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array([[1., 2.],\n",
+       "       [3., 4.],\n",
+       "       [5., 6.]], dtype=float16)"
+      ]
+     },
+     "execution_count": 9,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "rank_2_tensor.numpy()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tf.Tensor(\n",
+      "[[ 6  8]\n",
+      " [10 12]], shape=(2, 2), dtype=int32) \n",
+      "\n",
+      "tf.Tensor(\n",
+      "[[-4 -4]\n",
+      " [-4 -4]], shape=(2, 2), dtype=int32) \n",
+      "\n",
+      "tf.Tensor(\n",
+      "[[ 5 12]\n",
+      " [21 32]], shape=(2, 2), dtype=int32) \n",
+      "\n",
+      "tf.Tensor(\n",
+      "[[0.2        0.33333333]\n",
+      " [0.42857143 0.5       ]], shape=(2, 2), dtype=float64) \n",
+      "\n",
+      "tf.Tensor(\n",
+      "[[19 22]\n",
+      " [43 50]], shape=(2, 2), dtype=int32) \n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "#张量通常包含浮点型和整型数据，但是还有许多其他数据类型，包括：\n",
+    "#复杂的数值\n",
+    "#字符串\n",
+    "#您可以对张量执行基本数学运算，包括加法、逐元素乘法和矩阵乘法。\n",
+    "\n",
+    "a = tf.constant([\n",
+    "    [1,2],\n",
+    "    [3,4]\n",
+    "],dtype=tf.int32)\n",
+    "\n",
+    "b = tf.constant([\n",
+    "    [5,6],\n",
+    "    [7,8]\n",
+    "],dtype=tf.int32)\n",
+    "\n",
+    "\n",
+    "print(tf.add(a,b),\"\\n\")\n",
+    "print(tf.subtract(a,b),\"\\n\")\n",
+    "print(tf.multiply(a,b),\"\\n\")\n",
+    "print(tf.divide(a,b),\"\\n\")\n",
+    "print(tf.matmul(a,b),\"\\n\")\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 45,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tf.Tensor(10.0, shape=(), dtype=float32) \n",
+      "\n",
+      "tf.Tensor([ 5. 10.], shape=(2,), dtype=float32) \n",
+      "\n",
+      "tf.Tensor([ 6.  7.  8.  9. 10.], shape=(5,), dtype=float32) \n",
+      "\n",
+      "tf.Tensor([1 1 1 1 1], shape=(5,), dtype=int64) \n",
+      "\n",
+      "tf.Tensor([4 4], shape=(2,), dtype=int64) \n",
+      "\n",
+      "tf.Tensor(\n",
+      "[[0.01165623 0.03168492 0.08612853 0.23412165 0.63640857]\n",
+      " [0.01165623 0.03168492 0.08612853 0.23412165 0.63640857]], shape=(2, 5), dtype=float32)\n"
+     ]
+    }
+   ],
+   "source": [
+    "#各种运算都可以使用张量。\n",
+    "#tf.reduce_max 是 TensorFlow 中的一个函数，用于计算张量（tensor）中的最大值。它可以沿着张量的一个或多个维度进行操作。\n",
+    "a = tf.constant([\n",
+    "    [1,2,3,4,5],\n",
+    "    [6,7,8,9,10]\n",
+    "],dtype=tf.float32)\n",
+    "\n",
+    "# Find the largest value\n",
+    "print(tf.reduce_max(a),\"\\n\") #全局最大值\n",
+    "print(tf.reduce_max(a,axis=1),\"\\n\") #1轴/行最大值\n",
+    "print(tf.reduce_max(a,axis=0),\"\\n\") #0轴/列最大值\n",
+    "\n",
+    "# Find the index of the largest value\n",
+    "print(tf.math.argmax(a,0),\"\\n\") #每列最大值索引\n",
+    "print(tf.math.argmax(a,1),\"\\n\") #行最大值索引\n",
+    "\n",
+    "# Compute the softmax\n",
+    "#tf.nn.softmax 函数用于计算张量中每个元素的 softmax 激活值。\n",
+    "#Softmax 函数将输入的张量转换为一个新的张量，其中每个元素的取值范围在 0 到 1 之间，并且所有元素之和为 1。这使得 softmax 激活函数特别适合用于多分类问题中的输出层，其中每个输出节点表示一个类别的概率。\n",
+    "print(tf.nn.softmax(a))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 48,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<tf.Tensor: shape=(3, 2, 4, 5), dtype=float32, numpy=\n",
+       "array([[[[0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.]],\n",
+       "\n",
+       "        [[0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.]]],\n",
+       "\n",
+       "\n",
+       "       [[[0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.]],\n",
+       "\n",
+       "        [[0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.]]],\n",
+       "\n",
+       "\n",
+       "       [[[0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.]],\n",
+       "\n",
+       "        [[0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.],\n",
+       "         [0., 0., 0., 0., 0.]]]], dtype=float32)>"
+      ]
+     },
+     "execution_count": 48,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "##张量有形状。下面是几个相关术语：\n",
+    "\n",
+    "#形状：张量的每个轴的长度（元素数量）。\n",
+    "#秩：张量轴数。标量的秩为 0，向量的秩为 1，矩阵的秩为 2。\n",
+    "#轴或维度：张量的一个特殊维度。\n",
+    "#大小：张量的总项数，即形状矢量元素的乘积\n",
+    "#注：虽然您可能会看到“二维张量”之类的表述，但 2 秩张量通常并不是用来描述二维空间。\n",
+    "\n",
+    "rank_4_tensor = tf.zeros([3,2,4,5])\n",
+    "rank_4_tensor"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 49,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Type of every element: <dtype: 'float32'>\n",
+      "Number of axes: 4\n",
+      "Shape of tensor: (3, 2, 4, 5)\n",
+      "Elements along axis 0 of tensor: 3\n",
+      "Elements along the last axis of tensor: 5\n",
+      "Total number of elements (3*2*4*5):  120\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(\"Type of every element:\", rank_4_tensor.dtype)\n",
+    "print(\"Number of axes:\", rank_4_tensor.ndim)\n",
+    "print(\"Shape of tensor:\", rank_4_tensor.shape)\n",
+    "print(\"Elements along axis 0 of tensor:\", rank_4_tensor.shape[0])\n",
+    "print(\"Elements along the last axis of tensor:\", rank_4_tensor.shape[-1])\n",
+    "print(\"Total number of elements (3*2*4*5): \", tf.size(rank_4_tensor).numpy())"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 52,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tf.Tensor(4, shape=(), dtype=int32) \n",
+      "\n",
+      "tf.Tensor([3 2 4 5], shape=(4,), dtype=int32) \n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "#但请注意，Tensor.ndim 和 Tensor.shape 特性不返回 Tensor 对象。如果您需要 Tensor，请使用 tf.rank 或 tf.shape 函数。这种差异不易察觉，但在构建计算图时（稍后）可能非常重要。\n",
+    "print(tf.rank(rank_4_tensor),'\\n')\n",
+    "print(tf.shape(rank_4_tensor),'\\n')"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
    "metadata": {},
    "outputs": [],
-   "source": []
+   "source": [
+    "#虽然通常用索引来指代轴，但是您始终要记住每个轴的含义。轴一般按照从全局到局部的顺序进行排序：首先是批次轴，随后是空间维度，最后是每个位置的特征。这样，在内存中，特征向量就会位于连续的区域。\n",
+    "# https://tensorflow.google.cn/guide/tensor?hl=zh-cn   结合2个图理解\n",
+    "# 3 2 4 5\n",
+    "# 3 - Batch\n",
+    "# 2 - Height\n",
+    "# 4 - Width\n",
+    "# 5 - Features"
+   ]
   }
  ],
  "metadata": {