Xurui_Luo的博客

既然说 paritially known，那就想办法直接告诉后台shape到底什么样tf.shape 就可以获取当前张量的shape样例代码class Sampling(layers.Layer): def call(self, inputs): z_mean, z_log_var = inputs batch = tf.shape(z_mean)[0] # 获取第一维 dim = tf.shape(z_mean)[1] # 获取第二维 epsilon = tf.

参考 http://www.siyuanblog.com/?p=115207服务器端起：tensorboard –logdir=/data/serve_output/ —port=6006Mac端重定向: ssh -L 16006:127.0.0.1:6006 user@hostname 本地 localhost:16006:127.0.0.1 打开

转载 https://blog.youkuaiyun.com/jinbeibei0606/article/details/100771997

axis的取值个数等于矩阵的维度axis=i 表示第i维的数，按照i的变化进行相应操作，第i维对应位置的所有数，会最终变成1个数，故少了一个维度比如np.sum操作下面的例子假设了(4,3,2,3)维的矩阵，axis取值的顺序也就是(4,3,2,3)axis=0时，矩阵形状就变成(3,2,3)axis=1时，矩阵形状就变成(4,2,3)axis=2时，矩阵形状就变成(4,3,3)axis=3时，矩阵形状就变成(4,3,2)d = data = np.random.randint(0.

macOS：Mac OSX python ssl.SSLError：[SSL：CERTIFICATE_VERIFY_FAILED]证书验证失败/Applications/Python 3.7目录下，运行Install Certificates.command

tf.keras.callbacks.EarlyStopping( monitor='val_loss', min_delta=0, patience=0, verbose=0, mode='auto', baseline=None, restore_best_weights=False)monitor：监控的数据接口。keras定义了如下的数据接口可以直接使用：acc（accuracy），测试集的正确率loss，测试集的损失函数（误差）val_acc（val_accuracy

代码中添加np.set_printoptions(threshold=np.inf)

tf.clip_by_value(A, min, max)：输入一个张量A，把A中的每一个元素的值都压缩在min和max之间。小于min的让它等于min，大于max的元素的值等于max。import tensorflow as tf;import numpy as np; A = np.array([[1,1,2,4], [3,4,8,5]]) with tf.Session() as sess: print(sess.run(tf.clip_by_value(A, 2, 5)))输.

就是将原来用一维表示的特征如：分类问题共7类，用一维表示就是0,1,2,3,4,5,6中的某个数字来表示某一类而one-hot就吧它们变成[[1. 0. 0. 0. 0. 0. 0.] [0. 1. 0. 0. 0. 0. 0.] [0. 0. 1. 0. 0. 0. 0.] [0. 0. 0. 1. 0. 0. 0.] [0. 0. 0. 0. 1. 0. 0.] [0. 0. 0. 0. 0. 1. 0.] [0. 0. 0. 0. 0. 0. 1.]]具体代码样例：im

import tensorflow as tfA = [1, 3, 4, 5, 6]B = [1, 3, 4, 3, 2]correct_prediction = tf.equal(A, B)with tf.Session() as sess: # print(sess.run(tf.equal(A, B))) print(sess.run(tf.equal(A, B))) print(sess.run(tf.cast(correct_prediction, tf.flo

tf.equal(A, B)是对比这两个矩阵或者向量的相等的元素，如果是相等的那就返回True，反正返回False，返回的值的矩阵维度和A是一样的import tensorflow as tfimport numpy as np A = [[1,3,4,5,6]]B = [[1,3,4,3,2]] with tf.Session() as sess: print(sess.run(tf.equal(A, B)))[[ True True True False False]].

tf.cast：用于改变某个张量的数据类型import tensorflow as tf;import numpy as np; A = tf.convert_to_tensor(np.array([[1,1,2,4], [3,4,8,5]])) with tf.Session() as sess: print(A.dtype) b = tf.cast(A, tf.float32) print(b.dtype)输出<dtype: 'int64'><dtype: '

tf.argmax(vector, 1)：返回的是vector中的最大值的索引号，如果vector是一个向量，那就返回一个值，如果是一个矩阵，那就返回一个向量，这个向量的每一个维度都是相对应矩阵行的最大值元素的索引号。import tensorflow as tfimport numpy as np A = [[1,3,4,5,6]]B = [[1,3,4], [2,4,1]] with tf.Session() as sess: print(sess.run(tf.argmax(.

解决原代码# Load the data set(X_train, y_train), (X_test, y_test) = imdb.load_data(num_words=NUM_WORDS, index_from=INDEX_FROM)修改后import numpy as np# save np.loadnp_load_old = np.load# modify the default parameters of np.loadnp.load = lambda *a,**

参考在代码开头加上import requestsrequests.packages.urllib3.disable_warnings()import ssltry: _create_unverified_https_context = ssl._create_unverified_contextexcept AttributeError: # Legacy Python that doesn't verify HTTPS certificates by default