Skip to content

B
book

PaddlePaddle / book

通知 17
Star 4
Fork 0
B
book
提交 2e55fd46 编写于 作者: C ceci3 提交者： Cheerego
浏览文件
操作

update fit a line (#684)

上级 c064457d
隐藏空白更改
内联 并排
Showing with 76 addition and 98 deletion
01.fit_a_line/README.cn.md
浏览文件 @ 2e55fd46
......@@ -236,24 +236,6 @@ def train_test(executor, program, reader, feeder, fetch_list):
count += 1 # 累加测试集中的样本数量
return [x_d / count for x_d in accumulated] # 计算平均损失
```
可以直接输出损失值来观察`训练进程`:
```python
train_prompt = "train cost"
test_prompt = "test cost"
print("%s', out %f" % (train_prompt, out))
print("%s', out %f" % (test_prompt, out))
```
除此之外，还可以通过画图，来展现`训练进程`：
```python
from paddle.utils.plot import ploter
plot_prompt = ploter(train_prompt, test_prompt)
```
### 训练主循环
......@@ -264,8 +246,11 @@ plot_prompt = ploter(train_prompt, test_prompt)
%matplotlib inline
params_dirname = "fit_a_line.inference.model"
feeder = fluid.DataFeeder(place=place, feed_list=[x, y])
naive_exe = fluid.Executor(place)
naive_exe.run(startup_program)
exe.run(startup_program)
train_prompt = "train cost"
test_prompt = "test cost"
from paddle.utils.plot import Ploter
plot_prompt = Ploter(train_prompt, test_prompt)
step = 0
exe_test = fluid.Executor(place)
......@@ -280,10 +265,12 @@ for pass_id in range(num_epochs):
avg_loss_value, = exe.run(main_program,
feed=feeder.feed(data_train),
fetch_list=[avg_loss])
if step % 10 == 0: # 每10个批次记录一下训练损失
if step % 10 == 0: # 每10个批次记录并输出一下训练损失
plot_prompt.append(train_prompt, step, avg_loss_value[0])
plot_prompt.plot()
if step % 100 == 0: # 每100批次记录一下测试损失
print("%s, Step %d, Cost %f" %
(train_prompt, step, avg_loss_value[0]))
if step % 100 == 0: # 每100批次记录并输出一下测试损失
test_metics = train_test(executor=exe_test,
program=test_program,
reader=test_reader,
......@@ -291,6 +278,8 @@ for pass_id in range(num_epochs):
feeder=feeder)
plot_prompt.append(test_prompt, step, test_metics[0])
plot_prompt.plot()
print("%s, Step %d, Cost %f" %
(test_prompt, step, test_metics[0]))
if test_metics[0] < 10.0: # 如果准确率达到要求，则停止训练
break
......@@ -316,6 +305,24 @@ inference_scope = fluid.core.Scope()
```
### 预测
保存图片
```python
def save_result(points1, points2):
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
x1 = [idx for idx in range(len(points1))]
y1 = points1
y2 = points2
l1 = plt.plot(x1, y1, 'r--', label='predictions')
l2 = plt.plot(x1, y2, 'g--', label='GT')
plt.plot(x1, y1, 'ro-', x1, y2, 'g+-')
plt.title('predictions VS GT')
plt.legend()
plt.savefig('./image/prediction_gt.png')
```
通过fluid.io.load_inference_model，预测器会从`params_dirname`中读取已经训练好的模型，来对从未遇见过的数据进行预测。
```python
......@@ -337,37 +344,19 @@ with fluid.scope_guard(inference_scope):
results = infer_exe.run(inference_program,
feed={feed_target_names[0]: numpy.array(infer_feat)},
fetch_list=fetch_targets) # 进行预测
```
#打印预测结果和标签并可视化结果
print("infer results: (House Price)")
for idx, val in enumerate(results[0]):
print("%d: %.2f" % (idx, val)) # 打印预测结果
保存图片
```python
def save_result(points1, points2):
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
x1 = [idx for idx in range(len(points1))]
y1 = points1
y2 = points2
l1 = plt.plot(x1, y1, 'r--', label='predictions')
l2 = plt.plot(x1, y2, 'g--', label='GT')
plt.plot(x1, y1, 'ro-', x1, y2, 'g+-')
plt.title('predictions VS GT')
plt.legend()
plt.savefig('./image/prediction_gt.png')
```
print("\nground truth:")
for idx, val in enumerate(infer_label):
print("%d: %.2f" % (idx, val)) # 打印标签值
打印预测结果和标签并可视化结果
```python
print("infer results: (House Price)")
for idx, val in enumerate(results[0]):
print("%d: %.2f" % (idx, val)) # 打印预测结果
save_result(results[0], infer_label) # 保存图片
```
print("\nground truth:")
for idx, val in enumerate(infer_label):
print("%d: %.2f" % (idx, val)) # 打印标签值
save_result(results[0], infer_label) # 保存图片
```
## 总结
在这章里，我们借助波士顿房价这一数据集，介绍了线性回归模型的基本概念，以及如何使用PaddlePaddle实现训练和测试的过程。很多的模型和技巧都是从简单的线性回归模型演化而来，因此弄清楚线性模型的原理和局限非常重要。
......
01.fit_a_line/index.cn.html
浏览文件 @ 2e55fd46
......@@ -278,24 +278,6 @@ def train_test(executor, program, reader, feeder, fetch_list):
count += 1 # 累加测试集中的样本数量
return [x_d / count for x_d in accumulated] # 计算平均损失
```
可以直接输出损失值来观察`训练进程`:
```python
train_prompt = "train cost"
test_prompt = "test cost"
print("%s', out %f" % (train_prompt, out))
print("%s', out %f" % (test_prompt, out))
```
除此之外，还可以通过画图，来展现`训练进程`：
```python
from paddle.utils.plot import ploter
plot_prompt = ploter(train_prompt, test_prompt)
```
### 训练主循环
......@@ -306,8 +288,11 @@ plot_prompt = ploter(train_prompt, test_prompt)
%matplotlib inline
params_dirname = "fit_a_line.inference.model"
feeder = fluid.DataFeeder(place=place, feed_list=[x, y])
naive_exe = fluid.Executor(place)
naive_exe.run(startup_program)
exe.run(startup_program)
train_prompt = "train cost"
test_prompt = "test cost"
from paddle.utils.plot import Ploter
plot_prompt = Ploter(train_prompt, test_prompt)
step = 0
exe_test = fluid.Executor(place)
......@@ -322,10 +307,12 @@ for pass_id in range(num_epochs):
avg_loss_value, = exe.run(main_program,
feed=feeder.feed(data_train),
fetch_list=[avg_loss])
if step % 10 == 0: # 每10个批次记录一下训练损失
if step % 10 == 0: # 每10个批次记录并输出一下训练损失
plot_prompt.append(train_prompt, step, avg_loss_value[0])
plot_prompt.plot()
if step % 100 == 0: # 每100批次记录一下测试损失
print("%s, Step %d, Cost %f" %
(train_prompt, step, avg_loss_value[0]))
if step % 100 == 0: # 每100批次记录并输出一下测试损失
test_metics = train_test(executor=exe_test,
program=test_program,
reader=test_reader,
......@@ -333,6 +320,8 @@ for pass_id in range(num_epochs):
feeder=feeder)
plot_prompt.append(test_prompt, step, test_metics[0])
plot_prompt.plot()
print("%s, Step %d, Cost %f" %
(test_prompt, step, test_metics[0]))
if test_metics[0] < 10.0: # 如果准确率达到要求，则停止训练
break
......@@ -358,6 +347,24 @@ inference_scope = fluid.core.Scope()
```
### 预测
保存图片
```python
def save_result(points1, points2):
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
x1 = [idx for idx in range(len(points1))]
y1 = points1
y2 = points2
l1 = plt.plot(x1, y1, 'r--', label='predictions')
l2 = plt.plot(x1, y2, 'g--', label='GT')
plt.plot(x1, y1, 'ro-', x1, y2, 'g+-')
plt.title('predictions VS GT')
plt.legend()
plt.savefig('./image/prediction_gt.png')
```
通过fluid.io.load_inference_model，预测器会从`params_dirname`中读取已经训练好的模型，来对从未遇见过的数据进行预测。
```python
......@@ -379,37 +386,19 @@ with fluid.scope_guard(inference_scope):
results = infer_exe.run(inference_program,
feed={feed_target_names[0]: numpy.array(infer_feat)},
fetch_list=fetch_targets) # 进行预测
```
#打印预测结果和标签并可视化结果
print("infer results: (House Price)")
for idx, val in enumerate(results[0]):
print("%d: %.2f" % (idx, val)) # 打印预测结果
保存图片
```python
def save_result(points1, points2):
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
x1 = [idx for idx in range(len(points1))]
y1 = points1
y2 = points2
l1 = plt.plot(x1, y1, 'r--', label='predictions')
l2 = plt.plot(x1, y2, 'g--', label='GT')
plt.plot(x1, y1, 'ro-', x1, y2, 'g+-')
plt.title('predictions VS GT')
plt.legend()
plt.savefig('./image/prediction_gt.png')
```
print("\nground truth:")
for idx, val in enumerate(infer_label):
print("%d: %.2f" % (idx, val)) # 打印标签值
打印预测结果和标签并可视化结果
```python
print("infer results: (House Price)")
for idx, val in enumerate(results[0]):
print("%d: %.2f" % (idx, val)) # 打印预测结果
save_result(results[0], infer_label) # 保存图片
```
print("\nground truth:")
for idx, val in enumerate(infer_label):
print("%d: %.2f" % (idx, val)) # 打印标签值
save_result(results[0], infer_label) # 保存图片
```
## 总结
在这章里，我们借助波士顿房价这一数据集，介绍了线性回归模型的基本概念，以及如何使用PaddlePaddle实现训练和测试的过程。很多的模型和技巧都是从简单的线性回归模型演化而来，因此弄清楚线性模型的原理和局限非常重要。
......
Markdown is supported
0% 或 .
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 或