[英]How to solve tensorflow Conv2DBackpropFilter error?
我一直在尝试使用tf-gpu没有成功。 我从 anaconda 安装了 cuda（不确定这是问题还是编码问题）。 该代码在不使用 gpu 和安装 cuda 的情况下正常工作。 但是在我安装tf gpu我收到此错误：
tensorflow.python.framework.errors_impl.InvalidArgumentError: Conv2DBackpropFilter: input depth must be evenly divisible by filter depth
[[{{node gradient_tape/sequential/conv2d_10/Conv2D/Conv2DBackpropFilter4}}]] [Op:__inference_train_function_1390]
Function call stack:
train_function
我的代码：
import tensorflow as tf
import matplotlib.pyplot as plt
import os
import numpy as np
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.preprocessing import image
from tensorflow.keras.optimizers import SGD,Adam
from tensorflow.keras.models import load_model
physical_devices = tf.config.list_physical_devices("GPU")
tf.config.experimental.set_memory_growth(physical_devices[0], True)
train = ImageDataGenerator(rescale=1 / 255, rotation_range=20,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
vertical_flip=True,
fill_mode='nearest',
brightness_range=(0.1,0.9))
validation = ImageDataGenerator(rescale=1 / 255)
test = ImageDataGenerator(rescale=1 / 255)
train_dataset = train.flow_from_directory('/raw-img/training', target_size=(200,200), batch_size=1,
class_mode='categorical')
validation_dataset = validation.flow_from_directory('/raw-img/validation', target_size=(200,200), batch_size=1,
class_mode='categorical')
test_dataset = test.flow_from_directory('/raw-img/testing', target_size=(200,200), batch_size=1,
class_mode='categorical')
tf.config.experimental.enable_mlir_graph_optimization()
model = tf.keras.models.Sequential([
tf.keras.layers.Conv2D(filters=32, kernel_size=(3,3), input_shape=(200,200,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.MaxPool2D(pool_size=(2, 2),padding='same'),
tf.keras.layers.Dropout(rate=.2),
#
tf.keras.layers.Conv2D(filters=32, kernel_size=(3,3), padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.Conv2D(filters=32,kernel_size= (3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.MaxPool2D(pool_size=(2, 2),padding='same'),
tf.keras.layers.Dropout(rate=.25),
#
tf.keras.layers.Conv2D(filters=64,kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.Conv2D(filters=64, kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.MaxPool2D(pool_size=(2, 2),padding='same'),
tf.keras.layers.Dropout(rate=.25),
#
tf.keras.layers.Conv2D(filters=128, kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.Conv2D(filters=128, kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.MaxPool2D(pool_size=(2, 2),padding='same'),
tf.keras.layers.Dropout(rate=.25),
#
tf.keras.layers.Conv2D(filters=128, kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.Conv2D(filters=128, kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.MaxPool2D(pool_size=(2, 2),padding='same'),
tf.keras.layers.Dropout(rate=.25),
#
tf.keras.layers.Conv2D(filters=256, kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.Conv2D(filters=256, kernel_size=(3,3),padding='same'),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.MaxPool2D(pool_size=(2, 2),padding='same'),
tf.keras.layers.Dropout(rate=.2),
#
tf.keras.layers.Flatten(),
#
tf.keras.layers.Dense(units=1024),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.Dense(units=512),
tf.keras.layers.LeakyReLU( alpha=0.3),
tf.keras.layers.Dense(units=10, activation='softmax'),
])
print(model.summary())
rlronp=tf.keras.callbacks.ReduceLROnPlateau(monitor='val_loss',patience=3,verbose=1, factor=0.7)
es=tf.keras.callbacks.EarlyStopping(monitor="val_loss",patience=15,verbose=1,
restore_best_weights=True)
model.compile(loss='categorical_crossentropy', optimizer=SGD(learning_rate=0.003), metrics=['accuracy'])
model_fit = model.fit(train_dataset, epochs=100, batch_size=32, validation_data=validation_dataset,steps_per_epoch=len(train_dataset),validation_steps=len(validation_dataset),callbacks=[rlronp,es])
以下是我的 conda 环境中包含的软件包：
_tflow_select 2.1.0 gpu
abseil-cpp 20210324.2 hd77b12b_0
absl-py 0.13.0 py39haa95532_0
aiohttp 3.7.4 py39h2bbff1b_1
argon2-cffi 20.1.0 py39h2bbff1b_1
astor 0.8.1 py39haa95532_0
astunparse 1.6.3 py_0
async-timeout 3.0.1 py39haa95532_0
async_generator 1.10 pyhd3eb1b0_0
attrs 21.2.0 pyhd3eb1b0_0
backcall 0.2.0 pyhd3eb1b0_0
blas 1.0 mkl
bleach 3.3.1 pyhd3eb1b0_0
blinker 1.4 py39haa95532_0
brotlipy 0.7.0 py39h2bbff1b_1003
ca-certificates 2021.7.5 haa95532_1
cached-property 1.5.2 py_0
cachetools 4.2.2 pyhd3eb1b0_0
certifi 2021.5.30 py39haa95532_0
cffi 1.14.6 py39h2bbff1b_0
chardet 3.0.4 py39haa95532_1003
click 8.0.1 pyhd3eb1b0_0
colorama 0.4.4 pyhd3eb1b0_0
coverage 5.5 py39h2bbff1b_2
cryptography 3.4.7 py39h71e12ea_0
cudatoolkit 11.3.1 h59b6b97_2
cudnn 8.2.1 cuda11.3_0
cycler 0.10.0 py39haa95532_0
cython 0.29.24 py39hd77b12b_0
decorator 5.0.9 pyhd3eb1b0_0
defusedxml 0.7.1 pyhd3eb1b0_0
entrypoints 0.3 py39haa95532_0
flatbuffers 2.0.0 h6c2663c_0
freetype 2.10.4 hd328e21_0
gast 0.4.0 py_0
giflib 5.2.1 h62dcd97_0
google-auth 1.33.0 pyhd3eb1b0_0
google-auth-oauthlib 0.4.1 py_2
google-pasta 0.2.0 py_0
grpcio 1.36.1 py39hc60d5dd_1
h5py 3.2.1 py39h3de5c98_0
hdf5 1.10.6 h7ebc959_0
icc_rt 2019.0.0 h0cc432a_1
icu 68.1 h6c2663c_0
idna 2.10 pyhd3eb1b0_0
importlib-metadata 3.10.0 py39haa95532_0
importlib_metadata 3.10.0 hd3eb1b0_0
intel-openmp 2021.3.0 haa95532_3372
ipykernel 5.3.4 py39h7b7c402_0
ipython 7.25.0 py39h832f523_1 conda-forge
ipython_genutils 0.2.0 pyhd3eb1b0_1
jedi 0.18.0 py39haa95532_1
jinja2 3.0.1 pyhd3eb1b0_0
jpeg 9d h8ffe710_0 conda-forge
jsonschema 3.2.0 py_2
jupyter_client 6.1.12 pyhd3eb1b0_0
jupyter_core 4.7.1 py39haa95532_0
jupyterlab_pygments 0.1.2 py_0
keras-preprocessing 1.1.2 pyhd3eb1b0_0
kiwisolver 1.3.1 py39hd77b12b_0
krb5 1.18.2 hc04afaa_0
libclang 11.1.0 default_h5c34c98_1 conda-forge
libcurl 7.71.1 h2a8f88b_1
libpng 1.6.37 h2a8f88b_0
libprotobuf 3.14.0 h23ce68f_0
libsodium 1.0.18 h62dcd97_0
libssh2 1.9.0 h7a1dbc1_1
libtiff 4.2.0 hd0e1b90_0
lz4-c 1.9.3 h2bbff1b_0
m2w64-gcc-libgfortran 5.3.0 6
m2w64-gcc-libs 5.3.0 7
m2w64-gcc-libs-core 5.3.0 7
m2w64-gmp 6.1.0 2
m2w64-libwinpthread-git 5.0.0.4634.697f757 2
markdown 3.3.4 py39haa95532_0
markupsafe 2.0.1 py39h2bbff1b_0
matplotlib 3.3.4 py39haa95532_0
matplotlib-base 3.3.4 py39h49ac443_0
matplotlib-inline 0.1.2 pyhd8ed1ab_2 conda-forge
mistune 0.8.4 py39h2bbff1b_1000
mkl 2021.3.0 haa95532_524
mkl-service 2.4.0 py39h2bbff1b_0
mkl_fft 1.3.0 py39h277e83a_2
mkl_random 1.2.2 py39hf11a4ad_0
msys2-conda-epoch 20160418 1
multidict 5.1.0 py39h2bbff1b_2
nbclient 0.5.3 pyhd3eb1b0_0
nbconvert 6.1.0 py39haa95532_0
nbformat 5.1.3 pyhd3eb1b0_0
nest-asyncio 1.5.1 pyhd3eb1b0_0
notebook 6.4.0 py39haa95532_0
numpy 1.20.3 py39ha4e8547_0
numpy-base 1.20.3 py39hc2deb75_0
oauthlib 3.1.1 pyhd3eb1b0_0
olefile 0.46 py_0
openssl 1.1.1k h2bbff1b_0
opt_einsum 3.3.0 pyhd3eb1b0_1
packaging 21.0 pyhd3eb1b0_0
pandocfilters 1.4.3 py39haa95532_1
parso 0.8.2 pyhd3eb1b0_0
pickleshare 0.7.5 pyhd3eb1b0_1003
pillow 8.3.1 py39h4fa10fc_0
pip 21.1.3 py39haa95532_0
powershell_shortcut 0.0.1 3
prometheus_client 0.11.0 pyhd3eb1b0_0
prompt-toolkit 3.0.17 pyh06a4308_0
protobuf 3.14.0 py39hd77b12b_1
pyasn1 0.4.8 py_0
pyasn1-modules 0.2.8 py_0
pycparser 2.20 py_2
pygments 2.9.0 pyhd3eb1b0_0
pyjwt 2.1.0 py39haa95532_0
pyopenssl 20.0.1 pyhd3eb1b0_1
pyparsing 2.4.7 pyhd3eb1b0_0
pyqt 5.12.3 py39hcbf5309_7 conda-forge
pyqt-impl 5.12.3 py39h415ef7b_7 conda-forge
pyqt5-sip 4.19.18 py39h415ef7b_7 conda-forge
pyqtchart 5.12 py39h415ef7b_7 conda-forge
pyqtwebengine 5.12.1 py39h415ef7b_7 conda-forge
pyreadline 2.1 py39haa95532_1
pyrsistent 0.18.0 py39h2bbff1b_0
pysocks 1.7.1 py39haa95532_0
python 3.9.5 h6244533_3
python-dateutil 2.8.2 pyhd3eb1b0_0
python-flatbuffers 1.12 pyhd3eb1b0_0
python_abi 3.9 2_cp39 conda-forge
pywin32 228 py39he774522_0
pywinpty 0.5.7 py39haa95532_0
pyzmq 20.0.0 py39hd77b12b_1
qt 5.12.9 h5909a2a_4 conda-forge
requests 2.25.1 pyhd3eb1b0_0
requests-oauthlib 1.3.0 py_0
rsa 4.7.2 pyhd3eb1b0_1
scipy 1.6.2 py39h66253e8_1
send2trash 1.5.0 pyhd3eb1b0_1
setuptools 52.0.0 py39haa95532_0
six 1.16.0 pyhd3eb1b0_0
snappy 1.1.8 h33f27b4_0
sqlite 3.36.0 h2bbff1b_0
tensorboard 2.5.0 py_0
tensorboard-plugin-wit 1.6.0 py_0
tensorflow 2.5.0 gpu_py39h7dc34a2_0
tensorflow-base 2.5.0 gpu_py39hb3da07e_0
tensorflow-estimator 2.5.0 pyh7b7c402_0
tensorflow-gpu 2.5.0 h17022bd_0
termcolor 1.1.0 py39haa95532_1
terminado 0.9.4 py39haa95532_0
testpath 0.5.0 pyhd3eb1b0_0
tk 8.6.10 he774522_0
tornado 6.1 py39h2bbff1b_0
traitlets 5.0.5 pyhd3eb1b0_0
typing-extensions 3.10.0.0 hd3eb1b0_0
typing_extensions 3.10.0.0 pyh06a4308_0
tzdata 2021a h52ac0ba_0
urllib3 1.26.6 pyhd3eb1b0_1
vc 14.2 h21ff451_1
vs2015_runtime 14.27.29016 h5e58377_2
wcwidth 0.2.5 py_0
webencodings 0.5.1 py39haa95532_1
werkzeug 1.0.1 pyhd3eb1b0_0
wheel 0.35.1 pyhd3eb1b0_0
win_inet_pton 1.1.0 py39haa95532_0
wincertstore 0.2 py39h2bbff1b_0
winpty 0.4.3 4
wrapt 1.12.1 py39h196d8e1_1
xz 5.2.5 h62dcd97_0
yarl 1.6.3 py39h2bbff1b_0
zeromq 4.3.3 ha925a31_3
zipp 3.5.0 pyhd3eb1b0_0
zlib 1.2.11 h62dcd97_4
zstd 1.4.9 h19a0ad4_0
Tensorflow ( https://www.tensorflow.org/api_docs/python/tf/config/experimental/enable_mlir_graph_optimization?hl=ru )：
“注意：基于 MLIR 的 TensorFlow 编译器正在积极开发中，缺少功能，请不要使用。此 API 仅用于开发和测试。”
我不确定问题是什么，但您应该尝试删除该行：
tf.config.experimental.enable_mlir_graph_optimization()
看看它是如何运行的。
使用 tensorflow Conv1D：如何解决错误“层“conv1d_9”的输入 0 与层不兼容：”？
[英]Using tensorflow Conv1D: how can I solve error "Input 0 of layer "conv1d_9" is incompatible with the layer: "?
声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.