无法加载库cudnn_cnn_infer64_8.dll错误代码126

如何解决无法加载库cudnn_cnn_infer64_8.dll错误代码126

和我一样的“错误”。尽管我已经重新编译了tensorflow-GPU2.6.0的"Cuda version: 11.5 cuDNN version: 8.3“。当我将cudnn版本改为8.2，但将cuda版本保留为11.5时，“错误”就消失了。(需要重新编译)所以我认为这个错误一定是"cuDNN“造成的。

解决方法

Could not load library cudnn_cnn_infer64_8.dll. Error code 126
Please make sure cudnn_cnn_infer64_8.dll is in your library path!

当我尝试在图形处理器上使用TensorFlow时，我一直收到这个错误，我已经按照说明多次安装了CUDA、cuDNN和所有的驱动程序。但似乎什么都不起作用。如果我使用notebook，那么TensorFlow使用CPU，通过VS代码notebook扩展，我可以使用gpu，但它会在第一个时期停止会话，当我试图将其作为一个普通python文件运行时。发生上述错误。

完整的终端输出：

Found 14630 validated image filenames belonging to 3 classes.
Found 1500 validated image filenames belonging to 3 classes.
2021-11-08 11:03:58.000354: I tensorflow/core/platform/cpu_feature_guard.cc:151] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX AVX2
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-11-08 11:03:58.603592: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1525] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 2775 MB memory:  -> device: 0, name: NVIDIA GeForce GTX 1050 Ti, pci bus id: 0000:01:00.0, compute capability: 6.1        
Epoch 1/10
2021-11-08 11:04:07.306011: I tensorflow/stream_executor/cuda/cuda_dnn.cc:366] Loaded cuDNN version 8300
Could not load library cudnn_cnn_infer64_8.dll. Error code 126
Please make sure cudnn_cnn_infer64_8.dll is in your library path!
E:\MyWorkSpace\animal_detect>

代码片段：

import tensorflow as tf 
from tensorflow.keras.preprocessing.image import ImageDataGenerator 
from tensorflow.keras import layers 
from tensorflow.keras import Model 
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.applications.vgg16 import VGG16
import pandas as pd
import numpy as np

train_df = pd.read_csv('train.csv')
test_df = pd.read_csv('test.csv')
train_gen = ImageDataGenerator(rescale = 1./255.,rotation_range = 40, width_shift_range = 0.2, height_shift_range = 0.2, shear_range = 0.2, zoom_range = 0.2, horizontal_flip = True)
test_gen = ImageDataGenerator( rescale = 1.0/255. )
train_set = train_gen.flow_from_dataframe(train_df,x_col='loc',y_col='label',batch_size=20,target_size=(224,224))
test_set = train_gen.flow_from_dataframe(test_df,x_col='loc',y_col='label',batch_size=20,target_size=(224,224))
base_model = VGG16(input_shape = (224, 224, 3),
include_top = False,
weights = 'imagenet')
for layer in base_model.layers:
    layer.trainable = False
x = layers.Flatten()(base_model.output)
x = layers.Dense(512, activation='relu')(x)
x = layers.Dropout(0.5)(x)
x = layers.Dense(3, activation='sigmoid')(x)

model = tf.keras.models.Model(base_model.input, x)

model.compile(optimizer = tf.keras.optimizers.RMSprop(learning_rate=0.0001), loss = 'categorical_crossentropy',metrics = ['acc'])
vgghist = model.fit(train_set, validation_data = test_set, steps_per_epoch = 100, epochs = 10)

Jupyter-notebook、VS code notebook扩展和普通python文件都使用了相同的代码。

设备规格：

处理器:英特尔i5图形处理器: Nvidia Geforce 1050ti

Cuda版本: 11.5 cuDNN版本: 8.3