cxxx17的博客

自己筛选了一些词的word2vec组成了新的word2vec文件，新文件内容如下：300, 200的 0.209092 -0.165459 -0.058054 0.281176 0.102982 0.099868 0.047287 ....是 0.088422 -0.220535 0.042321 0.280248 0.158567 0.022675 0.104318 ....gensim load word2vec文件时报错。检查发现是词数和文件头词的个数（即这里的300）不一致。...

尝试jieba.posseg.cut(text, HMM=False)

from sklearn.metrics import confusion_matrixy_true = ["cat", "ant", "cat", "cat", "ant", "bird"]y_pred = ["ant", "ant", "cat", "cat", "ant", "cat"]cm = confusion_matrix(y_true, y_pred, labels=["ant", "bird", "cat"])print(cm.shape) # (3,3)计算多分类的recall

1、import sysimport osif hasattr(sys.modules[__name__], '__file__'): _file_name = __file__else: _file_name = inspect.getfile(inspect.currentframe())CURRENT_FILE_PATH = os.path.realpath(_file_name)print(CURRENT_FILE_PATH)CURRENT_DIR = os.path.

import matplotlib.pyplot as plt改为：import matplotlibmatplotlib.use('AGG')import matplotlib.pyplot as plt

有一个现成的json文件，需要load成字典使用。with open("lexicon.json",'r') as f: loaded_dict = json.load(f)

# logits 的最后一维=类别数def sofmax(logits): e_x = np.exp(x) probs = e_x / np.sum(e_x, axis=-1, keepdims=True) return probs

使用# T是个整数tensornp.concatenate((align[1:],T.cpu().numpy()), axis=0)报错ValueError: all the input arrays must have same number of dimensions, but the array at index 0 has 1 dimension(s) and the array at index 1 has 0 dimension(s)查看align[1:]及T.cpu().numpy(

use conda python

# logits [batch_size, frame_length, vocab_size]logits = np.load('logits_array.npy')logits = torch.from_numpy(logits.transpose((1,0))).requires_grad_()报错：ValueError: axes don’t match arraylogits = torch.from_numpy(logits.transpose((1,0,2))).requires_gr

原代码# -*- coding: utf-8 -*-f = open(metadata_file)for line in f: name, _, transcript = line.split('|')即使在py文件的头加入了# -- coding: utf-8 --仍会报错，修改代码：f = open(metadata_file, encoding='utf-8')问题解决。...

在linux下运行 prepare_data.py 出现 Bug：SyntaxError: Non-ASCII character ‘\xe2’ in file prepare_data.py我的代码是在 windows 系统下 git clone下来的，windows的默认编码是ASCII，而linux下默认编码是utf-8，将prepare_data.py头部加上# -*- coding: utf-8 -*-即可。...

1、判断大小写string = 'some strings'string.islower()string.isupper()2、大小写互转string = 'some strings'string = string.upper()string = string.lower()

对于list，里面如果没有相同的元素，可以用堆排序，比较简单。improt heapqtest_list = [1,2,3,4,5,23,-1,7,14,123]idx = list(map(test_list.index, heapq.nlargest(k, test_list)))for i in ind: test_list[i] -= 1print(test_list)如果list中含有相同的元素，则上述方法失效，因为test_list.index返回的是最大值第一次出现的地方。i

目的：将字符串最后一个%删掉。为设置了最大替换次数text = (text[::-1].replace('%','',max=1))[::-1]报错*** TypeError: replace() takes no keyword arguments，改成：text = (text[::-1].replace('%','',1))[::-1]

os.removedirs(self.outputdir)时出现Error:Directory not emptyimport shutilshutil.rmtree(self.outputdir, ignore_errors=True)问题解决。python判断文件是否存在：import osos.path.isfile(file_path)或者os.path.exists(file_path)判断文件夹是否存在：os.path.exists(file_dir)创建文件夹：

啦啦啦

@[TOC]python import 前两级目录下的文件参照了link的方法。文件结构：aaa.pybbb/code/ccc.pyddd/eee.py在ccc.py文件中import syssys.path.append("../..")sys.path.extend([os.path.join(root, name) for root, dirs, _ in os.walk("../../") for name in dirs])from aaa import *一级目录的话，