未验证 提交 cdcfa69a 编写于 作者: Z zhouzj 提交者: GitHub

test=fix_docs_distillation (#823)

上级 db08f47d
......@@ -27,14 +27,14 @@ Select `ResNet50` as the teacher to perform distillation training on the student
```python
import paddleslim as slim
model = slim.models.MobileNet()
student_model = slim.models.MobileNet()
student_program = fluid.Program()
student_startup = fluid.Program()
with fluid.program_guard(student_program, student_startup):
image = fluid.data(
name='image', shape=[None] + [1, 28, 28], dtype='float32')
label = fluid.data(name='label', shape=[None, 1], dtype='int64')
out = model.net(input=image, class_dim=10)
out = student_model.net(input=image, class_dim=10)
cost = fluid.layers.cross_entropy(input=out, label=label)
avg_cost = fluid.layers.mean(x=cost)
acc_top1 = fluid.layers.accuracy(input=out, label=label, k=1)
......@@ -44,7 +44,7 @@ with fluid.program_guard(student_program, student_startup):
```python
model = slim.models.ResNet50()
teacher_model = slim.models.ResNet50()
teacher_program = fluid.Program()
teacher_startup = fluid.Program()
with fluid.program_guard(teacher_program, teacher_startup):
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册