神经网络保存-导入

保存

python 复制代码
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
import gzip
# fashion_mnist=tf.keras.datasets.fashion_mnist
# (train_images,train_labels),(test_images,test_labels)=fashion_mnist.load_data()
 
#数据在个人资源里面,放到该文件目录中即可
def load_data():
#     dirname = os.path.join('datasets', 'fashion-mnist')
#     base = 'https://storage.googleapis.com/tensorflow/tf-ke ras-datasets/'
    files = [
      'train-labels-idx1-ubyte.gz', 'train-images-idx3-ubyte.gz',
      't10k-labels-idx1-ubyte.gz', 't10k-images-idx3-ubyte.gz'
    ]
 
    paths = []
    for fname in files:
        paths.append(fname)
 
    with gzip.open(paths[0], 'rb') as lbpath:
        y_train = np.frombuffer(lbpath.read(), np.uint8, offset=8)
 
    with gzip.open(paths[1], 'rb') as imgpath:
        x_train = np.frombuffer(
        imgpath.read(), np.uint8, offset=16).reshape(len(y_train), 28, 28)
 
    with gzip.open(paths[2], 'rb') as lbpath:
        y_test = np.frombuffer(lbpath.read(), np.uint8, offset=8)
 
    with gzip.open(paths[3], 'rb') as imgpath:
        x_test = np.frombuffer(
        imgpath.read(), np.uint8, offset=16).reshape(len(y_test), 28, 28)
 
    return (x_train, y_train), (x_test, y_test)
(x_train, y_train), (x_test, y_test)=load_data()

x_train=np.expand_dims(x_train,-1)

y_train_one_hot=tf.one_hot(y_train,10).numpy()
x_train=np.float32(x_train)

model=tf.keras.Sequential([
    tf.keras.layers.Conv2D(1,3,1),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(256,activation="relu"),
    tf.keras.layers.Dense(128,activation="relu"),
    tf.keras.layers.Dense(64,activation="relu"),
    tf.keras.layers.Dense(32,activation="relu"),
    tf.keras.layers.Dense(10,activation="softmax")
])
 
model.build(input_shape=[None,28,28,1])
model.summary()
model.compile(optimizer=tf.keras.optimizers.Adam(),loss=tf.keras.losses.CategoricalCrossentropy(),metrics=[tf.keras.losses.CategoricalCrossentropy()])

import os
checkpoint_path="training_1/cp.ckpt"
cp_callback=tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path,save_weights_only=True,verbose=1)


history=model.fit(x_train,y_train_one_hot,epochs=10,callbacks=[cp_callback])
LOSS=history.history["loss"]
plt.plot(LOSS)
plt.show()

导入

python 复制代码
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
import gzip
# fashion_mnist=tf.keras.datasets.fashion_mnist
# (train_images,train_labels),(test_images,test_labels)=fashion_mnist.load_data()
 
#数据在个人资源里面,放到该文件目录中即可
def load_data():
#     dirname = os.path.join('datasets', 'fashion-mnist')
#     base = 'https://storage.googleapis.com/tensorflow/tf-ke ras-datasets/'
    files = [
      'train-labels-idx1-ubyte.gz', 'train-images-idx3-ubyte.gz',
      't10k-labels-idx1-ubyte.gz', 't10k-images-idx3-ubyte.gz'
    ]
 
    paths = []
    for fname in files:
        paths.append(fname)
 
    with gzip.open(paths[0], 'rb') as lbpath:
        y_train = np.frombuffer(lbpath.read(), np.uint8, offset=8)
 
    with gzip.open(paths[1], 'rb') as imgpath:
        x_train = np.frombuffer(
        imgpath.read(), np.uint8, offset=16).reshape(len(y_train), 28, 28)
 
    with gzip.open(paths[2], 'rb') as lbpath:
        y_test = np.frombuffer(lbpath.read(), np.uint8, offset=8)
 
    with gzip.open(paths[3], 'rb') as imgpath:
        x_test = np.frombuffer(
        imgpath.read(), np.uint8, offset=16).reshape(len(y_test), 28, 28)
 
    return (x_train, y_train), (x_test, y_test)
(x_train, y_train), (x_test, y_test)=load_data()

x_test=np.expand_dims(x_test,-1)
model=tf.keras.Sequential([
    tf.keras.layers.Conv2D(1,3,1),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(256,activation="relu"),
    tf.keras.layers.Dense(128,activation="relu"),
    tf.keras.layers.Dense(64,activation="relu"),
    tf.keras.layers.Dense(32,activation="relu"),
    tf.keras.layers.Dense(10,activation="softmax")
])
model.build(input_shape=[None,28,28,1])
model.summary()
model.compile(optimizer=tf.keras.optimizers.Adam(),loss=tf.keras.losses.CategoricalCrossentropy(),metrics=[tf.keras.losses.CategoricalCrossentropy()])

checkpoint_path="training_1/cp.ckpt"
model.load_weights(checkpoint_path)

x_test=np.array(x_test,dtype=np.float32)
print(np.argmax(model.predict(x_test),axis=1))
print(y_test)
np.sum((y_test==np.argmax(model.predict(x_test),axis=1))*1)/y_test.shape[0]
相关推荐
币须赢18 分钟前
英伟达Thor芯片套件9月发货 “物理AI”有哪些?
大数据·人工智能
盼小辉丶24 分钟前
Transformer实战(18)——微调Transformer语言模型进行回归分析
深度学习·语言模型·回归·transformer
格林威27 分钟前
机器视觉检测如何使用360 度全景成像镜头进行AI 瑕疵检测
人工智能·深度学习·数码相机·机器学习·计算机视觉·视觉检测·相机
互联网之声1 小时前
崔传波教授:以科技与人文之光,点亮近视患者的清晰视界‌
人工智能
lily363926046a1 小时前
智联未来 点赋科技
大数据·人工智能
聚客AI1 小时前
🍬传统工程师转型:智能体架构师的技能图谱
人工智能·agent·mcp
lihuayong1 小时前
AI赋能金融研报自动化生成:智能体系统架构与实现
人工智能·金融研报自动化
架构师日志1 小时前
Google开源框架LangExtract实践(1)——Docker部署,免费、低碳、无需GPU、多种大模型灵活切换,绝对可用!
人工智能
数智顾问1 小时前
从理论到落地:神经网络稀疏化设计构架中网络剪枝的深度实践与创新
神经网络