如何解决我在使用 Random Search Keras Tuner 进行优化时遇到运行时错误
我使用 Keras 调谐器对数字识别器数据集进行超参数调整，但出现错误
首先，我在 CNNHyperModel 类中创建了用于超参数调整的构建方法
第二个我使用 Conv2D,MaxPooling2D,Dropout 然后是神经网络
我已经导入了这个程序所需的库
class CNNHyperModel(HyperModel):
#def __init__(self,input_shape,num_classes):
#self.input_shape =input_shape
#self.num_classes =num_classes
def build(self,hp) :
model=keras.Sequential()
model.add( Conv2D(filters=hp.Choice('1Conv_num_classes',values=[32,64,128,256]),activation="relu",strides=1,padding='same',kernal_size=(3,3),input_shape=(28,28,1))
)
model.add(Conv2D(filters=hp.Choice("2Conv_num_classes",54,activation='relu',3)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Dropout(rate=hp.Float("1Dropout",min_value=0.0,max_value=0.5,step=0.05)))
model.add(Conv2D(filters=hp.Choice("3Conv_num_classes",3)))
model.add(Conv2D(filters=hp.Choice("4Conv_num_classes",2),strides=(2,2)))
model.add(DropOut(rate=hp.Float("2Dropout",step=0.05)))
model.add(Conv2d(filters=hp.Choice("5Conv_num_classes",3)))
model.add(Conv2D(filters=hp.Choice("6Conv_NUM_CLASSES",2)))
model.add(Dropout(rate=hp.Float("3Dropout",step=0.05)))
model.add(Flatten())
model.add(Dense(units=hp.Int("Dense",min_value=32,max_value=512,step=32),activation='relu'))
model.add(Dropout(rate=hp.Float("Dense_Dropout",step=0.05)))
model.add(Dense(units=hp.Int("2Dense",min_values=32,max_values=512,activation='relu'))
model.add(Dropout(rate=hp.Float("2Dense_Dropout",step=0.05)))
model.add(Dense(10,activation='sigmoid'))
"""model.compile(optimizer=keras.optimizers.Adam(
hp.Float(
"Learning_rate",min_value=le-4,max_value=le-2,sampling="LOG"
)
),"""
model.compie(optimizer="sgd",loss="sparse_categorical_crossentropy",metrics=['accuracy'])
return model
#hypermodel=CNNHyperModel((28,1),10)
hypermodel=CNNHyperModel()
tuner = RandomSearch(
hypermodel,objective='accuracy',max_trials=15,executions_per_trial=3,directory='my_dir',project_name='digit'
)
但是我遇到了运行时错误
Traceback (most recent call last):
File "/usr/local/lib/python3.7/dist-packages/kerastuner/engine/hypermodel.py",line 104,in build
model = self.hypermodel.build(hp)
File "/usr/local/lib/python3.7/dist-packages/kerastuner/engine/hypermodel.py",line 64,in _build_wrapper
return self._build(hp,*args,**kwargs)
File "<ipython-input-17-9b2a20a37331>",line 10,in build
activation="relu",1))
TypeError: __init__() missing 1 required positional argument: 'kernel_size'
Traceback (most recent call last):
File "/usr/local/lib/python3.7/dist-packages/kerastuner/engine/hypermodel.py",1))
TypeError: __init__() missing 1 required positional argument: 'kernel_size'
Invalid model 0/5
Invalid model 1/5
Invalid model 2/5
Invalid model 3/5
Invalid model 4/5
Invalid model 5/5
Traceback (most recent call last):
File "/usr/local/lib/python3.7/dist-packages/kerastuner/engine/hypermodel.py",1))
TypeError: __init__() missing 1 required positional argument: 'kernel_size'
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/kerastuner/engine/hypermodel.py in build(self,hp)
103 with maybe_distribute(self.distribution_strategy):
--> 104 model = self.hypermodel.build(hp)
105 except:
9 frames
TypeError: __init__() missing 1 required positional argument: 'kernel_size'
During handling of the above exception,another exception occurred:
RuntimeError Traceback (most recent call last)
/usr/local/lib/python3.7/dist-packages/kerastuner/engine/hypermodel.py in build(self,hp)
111 if i == self._max_fail_streak:
112 raise RuntimeError(
--> 113 'Too many failed attempts to build model.')
114 continue
115
RuntimeError: Too many failed attempts to build model.
解决方法
内核大小应该是 3x3 而不是 3 。即 kernel_size=(3,3) 。内核是一个矩阵，而不是一个数字。
,上面的代码有一些拼写错误，需要改进
拼写错误，如 kernal_size ->kernel_size
所以这里是具有相同改进的工作核心
import tensorflow as tf
import tensorflow.keras as keras
from tensorflow.keras.layers import ( Conv2D,MaxPooling2D,Dropout,Dense,Flatten)
from kerastuner.tuners import RandomSearch
from kerastuner.engine.hyperparameters import HyperParameters
from kerastuner import HyperModel
import pandas as pd
import numpy as np
class CNNHyperModel(HyperModel):
#def __init__(self,input_shape,num_classes):
#self.input_shape =input_shape
#self.num_classes =num_classes
def build(self,hp) :
model=keras.Sequential()
model.add( Conv2D(filters=hp.Int('1Conv_num_classes',default=32,min_value=32,step=16,max_value=256),activation="relu",strides=1,padding='same',kernel_size=(3,3),input_shape=(28,28,1))
)
model.add(Conv2D(filters=hp.Int("2Conv_num_classes",max_value=256,step=16),activation='relu',3)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Dropout(rate=hp.Float("1Dropout",min_value=0.0,max_value=0.5,step=0.05)))
model.add(Conv2D(filters=hp.Int("3Conv_num_classes",default=64,3)))
model.add(Conv2D(filters=hp.Int("4Conv_num_classes",2),strides=(2,2)))
model.add(Dropout(rate=hp.Float("2Dropout",step=0.05)))
model.add(Conv2D(filters=hp.Int("5Conv_num_classes",default=128,3)))
model.add(Conv2D(filters=hp.Int("6Conv_NUM_CLASSES",2)))
model.add(Dropout(rate=hp.Float("3Dropout",step=0.05)))
model.add(Flatten())
model.add(Dense(units=hp.Int("Dense",default=516,max_value=512,activation='relu'))
model.add(Dropout(rate=hp.Float("Dense_Dropout",step=0.05)))
model.add(Dense(units=hp.Int("2Dense",activation='relu'))
model.add(Dropout(rate=hp.Float("2Dense_Dropout",step=0.05)))
model.add(Dense(10,activation='sigmoid'))
"""model.compile(optimizer=keras.optimizers.Adam(
hp.Float(
"Learning_rate",min_value=le-4,max_value=le-2,sampling="LOG"
),loss="sparse_categorical_crossentropy",metrics=['accuracy'])
),"""
model.compile(optimizer="sgd",metrics=['accuracy'])
return model
#hypermodel=CNNHyperModel((28,1),10)
hypermodel=CNNHyperModel()
如您所见，我在 Conv2D 中传递了 strides=1,padding='same' 以进行更多优化
快乐编码
版权声明：本文内容由互联网用户自发贡献，该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容， 请发送邮件至 dio@foxmail.com 举报，一经查实，本站将立刻删除。
