添加了对类型的详细介绍

zhoulingjun · Oct 12, 2015 · fc7d4fc · fc7d4fc
1 parent 377bf2b
commit fc7d4fc
Showing 1 changed file with 33 additions and 3 deletions.
diff --git a/09. theano/09.02 theano basics.ipynb b/09. theano/09.02 theano basics.ipynb
@@ -288,9 +288,39 @@
    "source": [
     "除了 `T.scalar()` 标量之外，`Theano` 中还有很多符号变量类型，这些都包含在 `tensor`（张量）子模块中，而且 `tensor` 中也有很多函数对它们进行操作。\n",
     "\n",
-    "- `T.matrix()` 矩阵\n",
-    "- `T.scalar()` 标量\n",
-    "- `T.vector()` 向量"
+    "- `T.scalar(name=None, dtype=config.floatX)` \n",
+    "    - 标量，shape - ()\n",
+    "- `T.vector(name=None, dtype=config.floatX)` \n",
+    "    - 向量，shape - (?,)\n",
+    "- `T.matrix(name=None, dtype=config.floatX)` \n",
+    "    - 矩阵，shape - (?,?)\n",
+    "- `T.row(name=None, dtype=config.floatX)` \n",
+    "    - 行向量，shape - (1,?)\n",
+    "- `T.col(name=None, dtype=config.floatX)` \n",
+    "    - 列向量，shape - (?,1)\n",
+    "- `T.tensor3(name=None, dtype=config.floatX)`\n",
+    "    - 3 维张量，shape - (?,?,?)\n",
+    "- `T.tensor4(name=None, dtype=config.floatX)`\n",
+    "    - 4 维张量，shape - (?,?,?,?)\n",
+    "\n",
+    "`shape` 中为 1 的维度支持 `broadcast` 机制。\n",
+    "\n",
+    "除了直接指定符号变量的类型（默认 `floatX`），还可以直接在每类前面加上一个字母来定义不同的类型：\n",
+    "\n",
+    "- `b` int8\n",
+    "- `w` int16\n",
+    "- `i` int32\n",
+    "- `d` float64\n",
+    "- `f` float32\n",
+    "- `c` complex64\n",
+    "- `z` complex128\n",
+    "\n",
+    "例如 `T.dvector()` 表示的就是一个 `float64` 型的向量。\n",
+    "\n",
+    "除此之外，还可以用它们的复数形式一次定义多个符号变量：\n",
+    "\n",
+    "    x,y,z = T.vectors('x','y','z')\n",
+    "    x,y,z = T.vectors(3)"
    ]
   },
   {