van的博客

# Attention(batch_size,1,max_length).dot(batch_size,max_legth,embedding_size)==（batch_size,1,embedding_size）

from sklearn.cluster import KMeansimport numpy as np num_clusters = 2km_cluster = KMeans(n_clusters=num_clusters, max_iter=300, n_init=40, \ init='k-means++',n_jobs=-1)tfidf_...

1.读取词向量def getWordWeight(weightfile, a=1e-3): if a <=0: # when the parameter makes no sense, use unweighted a = 1.0 word2weight = {} with open(weightfile) as f: line...

def funcc(a): return a*ac=lambda:funcc(a)a=1c()output:1带参数构建lambda,在调用此函数之前，先预定义其参数。

tensorflow collectiontensorflow的collection提供一个全局的存储机制，不会受到变量名生存空间的影响。一处保存，到处可取。接口介绍#向collection中存数据tf.Graph.add_to_collection(name, value)#Stores value in the collection with the given name....

import tensorflow as tfwith tf.variable_scope('gg',reuse=tf.AUTO_REUSE): f=tf.placeholder(tf.int32,[None]) f_f=tf.to_float(f > 0) user_emb_w = tf.get_variable("user_emb_w", [2, 5]) ...

import tensorflow as tf#tf.gather axis,,,二维a=tf.constant([[1,2,3],[4,5,6]])#paramb=tf.constant([0])#indexc=tf.gather(a,b,axis=-1)b1=tf.constant([1])#indexc1=tf.gather(a,b1,axis=-1)b2=tf.constan...

#svdfrom PIL import Imageimport numpy as npdef rebuild_img(u, sigma, v, p): #p表示奇异值的百分比 m = len(u) n = len(v) A=np.zeros((m,n)) for i in range(int(len(sigma)*p)): #print(i)...

import tensorflow as tfwith tf.variable_scope("foo", reuse=tf.AUTO_REUSE):    user_emb_w = tf.get_variable("user_emb_w", [2, 5])with tf.Session() as sess:    sess.run(tf.global_variables_initializ...

groupby分组函数：　　返回值：返回重构格式的DataFrame，特别注意，groupby里面的字段内的数据重构后都会变成索引　　groupby(),一般和sun()一起使用，如下例：from pandas import Series,DataFrame a=[['Li','男','PE',98.],['Li','男','MATH',60.],['liu','男','MATH',...

Python 取numpy数组的某几行某几列方法直接分析，如原矩阵如下（1）：（1）我们要截取的矩阵（取其一三行，和三四列数据构成矩阵）为如下（2）：（2）错误分析：取 C 的1 3行，3 4 列，定义Z = [0,2] #定义行数d = [2,3] #定义列数#代码C_zd = C[z,d]12345则结果为：由结果分析取的是第一行第三列和第三行第四列的...

from hyperopt import fmin, tpe, hp, STATUS_OK, Trials, space_evalimport numpy as npdef func(param_dict): loss=(param_dict['x'] ** 2 - 20 * param_dict['x']) ret = { "loss": loss, ...

from time import sleepfrom tqdm import tqdmfor i in tqdm(range(10000)): sleep(0.01)