|
| 1 | +.. _cn_api_fluid_io_normalize_program: |
| 2 | + |
| 3 | +normalize_program |
| 4 | +------------------------------- |
| 5 | + |
| 6 | + |
| 7 | +.. py:function:: paddle.static.normalize_program(program, feed_vars, fetch_vars) |
| 8 | +
|
| 9 | +
|
| 10 | +
|
| 11 | +
|
| 12 | +根据指定的 feed_vars 和 fetch_vars,优化 program。 |
| 13 | + |
| 14 | +参数: |
| 15 | + - **program** - 指定想要优化的 program。 |
| 16 | + - **feed_vars** (Variable | list[Variable]) – 模型的输入变量。 |
| 17 | + - **fetch_vars** (Variable | list[Variable]) – 模型的输出变量。 |
| 18 | + |
| 19 | +返回:优化之后的 program。 |
| 20 | + |
| 21 | +抛出异常: |
| 22 | + - ``TypeError`` – 如果 ``program`` 类型不是 ``Program``, 或 ``feed_vars``, ``fetch_vars`` 类型不是 Variable 或 list[Variable],则抛出异常。 |
| 23 | + |
| 24 | +**代码示例** |
| 25 | + |
| 26 | +.. code-block:: python |
| 27 | +
|
| 28 | + import paddle |
| 29 | +
|
| 30 | + paddle.enable_static() |
| 31 | +
|
| 32 | + path_prefix = "./infer_model" |
| 33 | +
|
| 34 | + # User defined network, here a softmax regession example |
| 35 | + image = paddle.static.data(name='img', shape=[None, 28, 28], dtype='float32') |
| 36 | + label = paddle.static.data(name='label', shape=[None, 1], dtype='int64') |
| 37 | + predict = paddle.static.nn.fc(image, 10, activation='softmax') |
| 38 | +
|
| 39 | + loss = paddle.nn.functional.cross_entropy(predict, label) |
| 40 | +
|
| 41 | + exe = paddle.static.Executor(paddle.CPUPlace()) |
| 42 | + exe.run(paddle.static.default_startup_program()) |
| 43 | +
|
| 44 | + # normalize main program. |
| 45 | + program = paddle.static.default_main_program() |
| 46 | + normalized_program = paddle.static.normalize_program(program, [image], [predict]) |
| 47 | +
|
0 commit comments