tensorflow基础知识（CNN）

1.在cnn的嵌入层看到如下代码

embedding=tf.get_variable('embedding',[3,2])

这时创建了一个类似3x2矩阵的变量，而且这个变量是的。所以要进行初始化，或者赋值。

embedding=tf.get_variable('embedding',[3,2])
session = tf.Session()
session.run(tf.global_variables_initializer())

这两行代码就可以进行初始化，先捕捉变量再进行变量的初始化，结果如下：

2.在tensorflow中定义一个变量
如下：
`

x = tf.Variable(1)

此时定义了一个变量x，那个1是个0阶张量
但此时只是定义了一个结构 类型是 变量 x，初始值是1，单x现在在内存中还不是1.因为tensorflow实际是以graph图表结构的形式运行的，在执行sess.run（传入需要取值的节点）时才去计算该图表某个节点的值，在此之前的操作都是为了构建此graph的结构并没有真正的赋于实际的值。执行variable（1）时也就是只是定义结构（类型为变量，初始值为1）。只有执行变量初始化方法时才赋予其定义的值。
with tf.Session().as_default() as sess:
sess.run(tf.global_variables_initializer())
print(sess.run(x))`
3.

4.关于tf.nn.embedding_look()的使用说明，这个实在看textcnn工程发现的。现有分类样本如下，只举很小的一个例子。

体育	鲍勃库西奖归谁属？ NCAA最强控卫是坎巴还是弗神新浪体育讯如今，本赛季的NCAA进入到了末段，各项奖项的评选结果也即将出炉，其中评选最佳控卫的鲍勃-库西奖就将在下周最终四强战时公布，
鲍勃-库西奖是由奈史密斯篮球名人堂提供，旨在奖励年度最佳大学控卫。最终获奖的球员也即将在以下几名热门人选中产生。〈〈〈 NCAA疯狂三月专题主页上线，点击链接查看精彩内容吉梅尔-弗雷戴特，
杨百翰大学“弗神”吉梅尔-弗雷戴特一直都备受关注，他不仅仅是一名射手，他会用“终结对手脚踝”
娱乐	林俊杰为电影《夏日乐悠悠》献首唱HOLD住全场新浪娱乐讯 近日，林俊杰(微博)在成都举办歌友会，现场演唱了其为由马楚成(微博)执导的电影《夏日乐悠悠》创作的主题曲《LOVE YOU YOU》，
这是林俊杰首次公开演唱这首新创作的歌曲却在现场出乎意料的引发全场歌迷大合唱场面感人。目前电影《夏日乐悠悠》计划于9月30日在全国上映，该片由当红明星Angelababy(微博)、彭于晏(微博)、朱雨辰(微博)、周扬(微博)
游戏	《海湾战争-空中王者》歼-20保卫升空一款绚丽的战斗画面，横越海、陆、空三大战场，加上紧张的剧情，就构成了今天小编给各大玩家介绍的这款射击游戏：《海湾战争-空中王者》。剧情介绍：海湾地区储存了丰富的石油和天然气，加上地处亚非欧交界，历为兵家必争之地。经过新一次全球经济危机之后，海湾战争一触即发，中国也被卷入纷争之中。为了掌握实地军情，我军派出了有空中王者之称的歼-20，远赴海湾地区……更多精彩震撼感觉，立即下载该款游戏尽情体验吧。玩家交流才是王道，讯易游戏玩家交流中心 

a:现在开始读取数据，需要将样本进行进行向量化。上面共有三个标签，分别读取每个标签的第一行，通过如下代码实现：

def read_file(filename):
    """读取文件数据"""
    contents, labels = [], []
    with open_file(filename) as f:
        for line in f:
            try:
                label, content = line.strip().split('\t')
                if content:
                    contents.append(list(native_content(content)))
                    labels.append(native_content(label))
            except:
                pass
    return contents, labels

经过这段代码后 labels里包含[‘体育’,‘娱乐’,‘游戏’]

`
contents [[‘鲍’, ‘勃’, ‘库’, ‘西’, ‘奖’, ‘归’, ‘谁’, ‘属’, ‘？’, ’ ', ‘N’, ‘C’, ‘A’, ‘A’, ‘最’, ‘强’, ‘控’, ‘卫’, ‘是’, ‘坎’, ‘巴’, ‘还’, ‘是’, ‘弗’, ‘神’, ‘新’, ‘浪’, ‘体’, ‘育’, ‘讯’, ‘如’, ‘今’, ‘，’, ‘本’, ‘赛’, ‘季’, ‘的’, ‘N’, ‘C’, ‘A’, ‘A’, ‘进’, ‘入’, ‘到’, ‘了’, ‘末’, ‘段’, ‘，’, ‘各’, ‘项’, ‘奖’, ‘项’, ‘的’, ‘评’, ‘选’, ‘结’, ‘果’, ‘也’, ‘即’, ‘将’, ‘出’, ‘炉’, ‘，’, ‘其’, ‘中’, ‘评’, ‘选’, ‘最’, ‘佳’, ‘控’, ‘卫’, ‘的’, ‘鲍’, ‘勃’, ‘-’, ‘库’, ‘西’, ‘奖’, ‘就’, ‘将’, ‘在’, ‘下’, ‘周’, ‘最’, ‘终’, ‘四’, ‘强’, ‘战’, ‘时’, ‘公’, ‘布’, ‘，’], [‘林’, ‘俊’, ‘杰’, ‘为’, ‘电’, ‘影’, ‘《’, ‘夏’, ‘日’, ‘乐’, ‘悠’, ‘悠’, ‘》’, ‘献’, ‘首’, ‘唱’, ‘H’, ‘O’, ‘L’, ‘D’, ‘住’, ‘全’, ‘场’, ‘新’, ‘浪’, ‘娱’, ‘乐’, ‘讯’, ’ ', ‘近’, ‘日’, ‘，’, ‘林’, ‘俊’, ‘杰’, ‘(’, ‘微’, ‘博’, ‘)’, ‘在’, ‘成’, ‘都’, ‘举’, ‘办’, ‘歌’, ‘友’, ‘会’, ‘，’, ‘现’, ‘场’, ‘演’, ‘唱’, ‘了’, ‘其’, ‘为’, ‘由’, ‘马’, ‘楚’, ‘成’, ‘(’, ‘微’, ‘博’, ‘)’, ‘执’, ‘导’, ‘的’, ‘电’, ‘影’, ‘《’, ‘夏’, ‘日’, ‘乐’, ‘悠’, ‘悠’, ‘》’, ‘创’, ‘作’, ‘的’, ‘主’, ‘题’, ‘曲’, ‘《’, ‘L’, ‘O’, ‘V’, ‘E’, ’ ‘, ‘Y’, ‘O’, ‘U’, ’ ‘, ‘Y’, ‘O’, ‘U’, ‘》’, ‘，’], [’《’, ‘海’, ‘湾’, ‘战’, ‘争’, ‘-’, ‘空’, ‘中’, ‘王’, ‘者’, ‘》’, ‘歼’, ‘-’, ‘2’, ‘0’, ‘保’, ‘卫’, ‘升’, ‘空’, ‘一’, ‘款’, ‘绚’, ‘丽’, ‘的’, ‘战’, ‘斗’, ‘画’, ‘面’, ‘，’, ‘横’, ‘越’, ‘海’, ‘、’, ‘陆’, ‘、’, ‘空’, ‘三’, ‘大’, ‘战’, ‘场’, ‘，’, ‘加’, ‘上’, ‘紧’, ‘张’, ‘的’, ‘剧’, ‘情’, ‘，’, ‘就’, ‘构’, ‘成’, ‘了’, ‘今’, ‘天’, ‘小’, ‘编’, ‘给’, ‘各’, ‘大’, ‘玩’, ‘家’, ‘介’, ‘绍’, ‘的’, ‘这’, ‘款’, ‘射’, ‘击’, ‘游’, ‘戏’, ‘：’, ‘《’, ‘海’, ‘湾’, ‘战’, ‘争’, ‘-’, ‘空’, ‘中’, ‘王’, ‘者’, ‘》’, ‘。’, ‘剧’, ‘情’, ‘介’, ‘绍’, ‘：’, ‘海’, ‘湾’, ‘地’, ‘区’, ‘储’, ‘存’, ‘了’, ‘丰’, ‘富’, ‘的’, ‘石’, ‘油’, ‘和’, ‘天’, ‘然’, ‘气’, ‘，’, ‘加’, ‘上’, ‘地’, ‘处’, ‘亚’, ‘非’, ‘欧’, ‘交’, ‘界’, ‘，’, ‘历’, ‘为’, ‘兵’, ‘家’, ‘必’, ‘争’, ‘之’, ‘地’, ‘。’, ‘经’, ‘过’, ‘新’, ‘一’, ‘次’, ‘全’, ‘球’, ‘经’, ‘济’, ‘危’, ‘机’, ‘之’, ‘后’, ‘，’, ‘海’, ‘湾’, ‘战’, ‘争’, ‘一’, ‘触’, ‘即’, ‘发’, ‘，’, ‘中’, ‘国’, ‘也’, ‘被’, ‘卷’, ‘入’, ‘纷’, ‘争’, ‘之’, ‘中’, ‘。’, ‘为’, ‘了’, ‘掌’, ‘握’, ‘实’, ‘地’, ‘军’, ‘情’, ‘，’, ‘我’, ‘军’, ‘派’, ‘出’, ‘了’, ‘有’, ‘空’, ‘中’, ‘王’, ‘者’, ‘之’, ‘称’, ‘的’, ‘歼’, ‘-’, ‘2’, ‘0’, ‘，’, ‘远’, ‘赴’, ‘海’, ‘湾’, ‘地’, ‘区’, ‘…’, ‘…’, ‘更’, ‘多’, ‘精’, ‘彩’, ‘震’, ‘撼’, ‘感’, ‘觉’, ‘，’, ‘立’, ‘即’, ‘下’, ‘载’, ‘该’, ‘款’, ‘游’, ‘戏’, ‘尽’, ‘情’, ‘体’, ‘验’, ‘吧’, ‘。’, ‘玩’, ‘家’, ‘交’, ‘流’, ‘才’, ‘是’, ‘王’, ‘道’, ‘，’, ‘讯’, ‘易’, ‘游’, ‘戏’, ‘玩’, ‘家’, ‘交’, ‘流’, ‘中’, ‘心’]]

上面是contents包含的内容。
接着构建词汇表，准确来说是字汇表，我看的文章是字符集的文本分类。构建词汇表可以根据以下代码实现

def build_vocab(train_dir, vocab_dir, vocab_size=5000):
    train_dir就是上面样本的路径
    data_train, _ = read_file(train_dir)
    #data_train就是上面的contents
    all_data = []
    for content in data_train:   #content对应的就是标签后的那一行
        print('content',content)
        all_data.extend(content)
    print('all_data:',all_data)
    all_data的内容可以和data_train比较下
    all_data:all_data: ['鲍', '勃', '库', '西', '奖', '归', '谁', '属', '？', ' ', 'N', 'C', 'A', 'A', '最', '强', '控', '卫', '是', '坎', '巴', '还', '是', '弗', '神', '新', '浪', '体', '育', '讯', '如', '今', '，', '本', '赛', '季', '的', 'N', 'C', 'A', 'A', '进', '入', '到', '了', '末', '段', '，', '各', '项', '奖', '项', '的', '评', '选', '结', '果', '也', '即', '将', '出', '炉', '，', '其', '中', '评', '选', '最', '佳', '控', '卫', '的', '鲍', '勃', '-', '库', '西', '奖', '就', '将', '在', '下', '周', '最', '终', '四', '强', '战', '时', '公', '布', '，', '林', '俊', '杰', '为', '电', '影', '《', '夏', '日', '乐', '悠', '悠', '》', '献', '首', '唱', 'H', 'O', 'L', 'D', '住', '全', '场', '新', '浪', '娱', '乐', '讯', ' ', '近', '日', '，', '林', '俊', '杰', '(', '微', '博', ')', '在', '成', '都', '举', '办', '歌', '友', '会', '，', '现', '场', '演', '唱', '了', '其', '为', '由', '马', '楚', '成', '(', '微', '博', ')', '执', '导', '的', '电', '影', '《', '夏', '日', '乐', '悠', '悠', '》', '创', '作', '的', '主', '题', '曲', '《', 'L', 'O', 'V', 'E', ' ', 'Y', 'O', 'U', ' ', 'Y', 'O', 'U', '》', '，', '《', '海', '湾', '战', '争', '-', '空', '中', '王', '者', '》', '歼', '-', '2', '0', '保', '卫', '升', '空', '一', '款', '绚', '丽', '的', '战', '斗', '画', '面', '，', '横', '越', '海', '、', '陆', '、', '空', '三', '大', '战', '场', '，', '加', '上', '紧', '张', '的', '剧', '情', '，', '就', '构', '成', '了', '今', '天', '小', '编', '给', '各', '大', '玩', '家', '介', '绍', '的', '这', '款', '射', '击', '游', '戏', '：', '《', '海', '湾', '战', '争', '-', '空', '中', '王', '者', '》', '。', '剧', '情', '介', '绍', '：', '海', '湾', '地', '区', '储', '存', '了', '丰', '富', '的', '石', '油', '和', '天', '然', '气', '，', '加', '上', '地', '处', '亚', '非', '欧', '交', '界', '，', '历', '为', '兵', '家', '必', '争', '之', '地', '。', '经', '过', '新', '一', '次', '全', '球', '经', '济', '危', '机', '之', '后', '，', '海', '湾', '战', '争', '一', '触', '即', '发', '，', '中', '国', '也', '被', '卷', '入', '纷', '争', '之', '中', '。', '为', '了', '掌', '握', '实', '地', '军', '情', '，', '我', '军', '派', '出', '了', '有', '空', '中', '王', '者', '之', '称', '的', '歼', '-', '2', '0', '，', '远', '赴', '海', '湾', '地', '区', '…', '…', '更', '多', '精', '彩', '震', '撼', '感', '觉', '，', '立', '即', '下', '载', '该', '款', '游', '戏', '尽', '情', '体', '验', '吧', '。', '玩', '家', '交', '流', '才', '是', '王', '道', '，', '讯', '易', '游', '戏', '玩', '家', '交', '流', '中', '心']
    counter = Counter(all_data)
    counter的内容 Counter({'，': 18, '的': 10, '中': 7, '了': 6, '战': 6, '海': 6, '-': 5, '《': 5, '》': 5, '湾': 5, '争': 5, '空': 5, '地': 5, ' ': 4, 'A': 4, '为': 4, '悠': 4, 'O': 4, '王': 4, '情': 4, '家': 4, '。': 4, '之': 4, '奖': 3, '最': 3, '卫': 3, '是': 3, '新': 3, '讯': 3, '即': 3, '日': 3, '乐': 3, '场': 3, '成': 3, '者': 3, '一': 3, '款': 3, '玩': 3, '游': 3, '戏': 3, '交': 3, '鲍': 2, '勃': 2, '库': 2, '西': 2, 'N': 2, 'C': 2, '强': 2, '控': 2, '浪': 2, '体': 2, '今': 2, '入': 2, '各': 2, '项': 2, '评': 2, '选': 2, '也': 2, '将': 2, '出': 2, '其': 2, '就': 2, '在': 2, '下': 2, '林': 2, '俊': 2, '杰': 2, '电': 2, '影': 2, '夏': 2, '唱': 2, 'L': 2, '全': 2, '(': 2, '微': 2, '博': 2, ')': 2, 'Y': 2, 'U': 2, '歼': 2, '2': 2, '0': 2, '、': 2, '大': 2, '加': 2, '上': 2, '剧': 2, '天': 2, '介': 2, '绍': 2, '：': 2, '区': 2, '经': 2, '军': 2, '…': 2, '流': 2, '归': 1, '谁': 1, '属': 1, '？': 1, '坎': 1, '巴': 1, '还': 1, '弗': 1, '神': 1, '育': 1, '如': 1, '本': 1, '赛': 1, '季': 1, '进': 1, '到': 1, '末': 1, '段': 1, '结': 1, '果': 1, '炉': 1, '佳': 1, '周': 1, '终': 1, '四': 1, '时': 1, '公': 1, '布': 1, '献': 1, '首': 1, 'H': 1, 'D': 1, '住': 1, '娱': 1, '近': 1, '都': 1, '举': 1, '办': 1, '歌': 1, '友': 1, '会': 1, '现': 1, '演': 1, '由': 1, '马': 1, '楚': 1, '执': 1, '导': 1, '创': 1, '作': 1, '主': 1, '题': 1, '曲': 1, 'V': 1, 'E': 1, '保': 1, '升': 1, '绚': 1, '丽': 1, '斗': 1, '画': 1, '面': 1, '横': 1, '越': 1, '陆': 1, '三': 1, '紧': 1, '张': 1, '构': 1, '小': 1, '编': 1, '给': 1, '这': 1, '射': 1, '击': 1, '储': 1, '存': 1, '丰': 1, '富': 1, '石': 1, '油': 1, '和': 1, '然': 1, '气': 1, '处': 1, '亚': 1, '非': 1, '欧': 1, '界': 1, '历': 1, '兵': 1, '必': 1, '过': 1, '次': 1, '球': 1, '济': 1, '危': 1, '机': 1, '后': 1, '触': 1, '发': 1, '国': 1, '被': 1, '卷': 1, '纷': 1, '掌': 1, '握': 1, '实': 1, '我': 1, '派': 1, '有': 1, '称': 1, '远': 1, '赴': 1, '更': 1, '多': 1, '精': 1, '彩': 1, '震': 1, '撼': 1, '感': 1, '觉': 1, '立': 1, '载': 1, '该': 1, '尽': 1, '验': 1, '吧': 1, '才': 1, '道': 1, '易': 1, '心': 1})
    count_pairs = counter.most_common(vocab_size - 1)
    #vocab_size是不同字符的个数，自己定取多少个
 [('，', 18), ('的', 10), ('中', 7), ('了', 6), ('战', 6), ('海', 6), ('-', 5), ('《', 5), ('》', 5), ('湾', 5), ('争', 5), ('空', 5), ('地', 5), (' ', 4), ('A', 4), ('为', 4), ('悠', 4), ('O', 4), ('王', 4), ('情', 4), ('家', 4), ('。', 4), ('之', 4), ('奖', 3), ('最', 3), ('卫', 3), ('是', 3), ('新', 3), ('讯', 3), ('即', 3), ('日', 3), ('乐', 3), ('场', 3), ('成', 3), ('者', 3), ('一', 3), ('款', 3), ('玩', 3), ('游', 3), ('戏', 3), ('交', 3), ('鲍', 2), ('勃', 2), ('库', 2), ('西', 2), ('N', 2), ('C', 2), ('强', 2), ('控', 2), ('浪', 2), ('体', 2), ('今', 2), ('入', 2), ('各', 2), ('项', 2), ('评', 2), ('选', 2), ('也', 2), ('将', 2), ('出', 2), ('其', 2), ('就', 2), ('在', 2), ('下', 2), ('林', 2), ('俊', 2)]
 
    words, _ = list(zip(*count_pairs))
     
    #自己参考zip的用法，zip返回的是个迭代器，里面是tuple组成的，因此用列表显示，在这里list(zip(*count_pairs))包含两个tuple，一个是词汇字符，一个是字符相对应所出现的频次
   
    
    print(' words', words)#words是tuple
   words  ('，', '的', '中', '了', '战', '海', '-', '《', '》', '湾', '争', '空', '地', ' ', 'A', '为', '悠', 'O', '王', '情', '家', '。', '之', '奖', '最', '卫', '是', '新', '讯', '即', '日', '乐', '场', '成', '者', '一', '款', '玩', '游', '戏', '交', '鲍', '勃', '库', '西', 'N', 'C', '强', '控', '浪', '体', '今', '入', '各', '项', '评', '选', '也', '将', '出', '其', '就', '在', '下', '林', '俊')

    # 添加一个 <PAD> 来将所有文本pad为同一长度
    words = ['<PAD>'] + list(words)
    open_file(vocab_dir, mode='w').write('\n'.join(words) + '\n')
    #vocab_dir是建立的字汇表的地址
    词汇表建立后如下图

b：建立词汇表后，将词汇表数字化，以及内容数字化.。词汇表数字化使用以下代码

def read_vocab(vocab_dir):
    """读取词汇表"""
    # words = open_file(vocab_dir).read().strip().split('\n')
    with open_file(vocab_dir) as fp:
        # 如果是py2 则每个值都转化为unicode
        words = [_.strip() for _ in fp.readlines()]
        print('words2', words)
    word_to_id = dict(zip(words, range(len(words))))
    print('word_to_id',word_to_id)
    return words, word_to_id

word_to_id {’’: 0, ‘，’: 1, ‘的’: 2, ‘中’: 3, ‘了’: 4, ‘战’: 5, ‘海’: 6, ‘-’: 7, ‘《’: 8, ‘》’: 9, ‘湾’: 10, ‘争’: 11, ‘空’: 12, ‘地’: 13, ‘’: 14, ‘A’: 15, ‘为’: 16, ‘悠’: 17, ‘O’: 18, ‘王’: 19, ‘情’: 20, ‘家’: 21, ‘。’: 22, ‘之’: 23, ‘奖’: 24, ‘最’: 25, ‘卫’: 26, ‘是’: 27, ‘新’: 28, ‘讯’: 29, ‘即’: 30, ‘日’: 31, ‘乐’: 32, ‘场’: 33, ‘成’: 34, ‘者’: 35, ‘一’: 36, ‘款’: 37, ‘玩’: 38, ‘游’: 39, ‘戏’: 40, ‘交’: 41, ‘鲍’: 42, ‘勃’: 43, ‘库’: 44, ‘西’: 45, ‘N’: 46, ‘C’: 47, ‘强’: 48, ‘控’: 49, ‘浪’: 50, ‘体’: 51, ‘今’: 52, ‘入’: 53, ‘各’: 54, ‘项’: 55, ‘评’: 56, ‘选’: 57, ‘也’: 58, ‘将’: 59, ‘出’: 60, ‘其’: 61, ‘就’: 62, ‘在’: 63, ‘下’: 64, ‘林’: 65, ‘俊’: 66}

c:将种类进行数字化

def read_category():
    """读取分类目录，固定"""
    categories = ['体育', '财经', '房产', '家居', '教育', '科技', '时尚', '时政', '游戏', '娱乐']

    categories = [native_content(x) for x in categories]

    cat_to_id = dict(zip(categories, range(len(categories))))


    return categories, cat_to_id

cat_to_id {‘体育’: 0, ‘财经’: 1, ‘房产’: 2, ‘家居’: 3, ‘教育’: 4, ‘科技’: 5, ‘时尚’: 6, ‘时政’: 7, ‘游戏’: 8, ‘娱乐’: 9}

d：在词汇表和种类都进行数字化后，现在进行文本内容，也就是上面contents的数字化

def process_file(filename, word_to_id, cat_to_id, max_length=600):
    """将文件转换为id表示"""
    contents, labels = read_file(filename)
    print('----------------------fengexian--------------------------------------------')
    print('contents',contents)
    print('labels',labels)
    data_id, label_id = [], []
    for i in range(len(contents)):
        data_id.append([word_to_id[x] for x in contents[i] if x in word_to_id])

        print('data_id',data_id)
        label_id.append(cat_to_id[labels[i]])
        print('label_id', label_id)
        print('length',len(data_id[0]))
    # 使用keras提供的pad_sequences来将文本pad为固定长度
    x_pad = kr.preprocessing.sequence.pad_sequences(data_id, 68)#x_pad可以看做二维矩阵，每一行都是每一行内容的数字化。长度可以调整，如果不够就用0补，这样每行的维数机一样，此处x_pad形式很重要
    print('x_pad',x_pad)
    y_pad = kr.utils.to_categorical(label_id, num_classes=len(cat_to_id))  # 将标签转换为one-hot表示
    print('y_pad',y_pad)

    return x_pad, y_pad

contents [[‘鲍’, ‘勃’, ‘库’, ‘西’, ‘奖’, ‘归’, ‘谁’, ‘属’, ‘？’, ’ ', ‘N’, ‘C’, ‘A’, ‘A’, ‘最’, ‘强’, ‘控’, ‘卫’, ‘是’, ‘坎’, ‘巴’, ‘还’, ‘是’, ‘弗’, ‘神’, ‘新’, ‘浪’, ‘体’, ‘育’, ‘讯’, ‘如’, ‘今’, ‘，’, ‘本’, ‘赛’, ‘季’, ‘的’, ‘N’, ‘C’, ‘A’, ‘A’, ‘进’, ‘入’, ‘到’, ‘了’, ‘末’, ‘段’, ‘，’, ‘各’, ‘项’, ‘奖’, ‘项’, ‘的’, ‘评’, ‘选’, ‘结’, ‘果’, ‘也’, ‘即’, ‘将’, ‘出’, ‘炉’, ‘，’, ‘其’, ‘中’, ‘评’, ‘选’, ‘最’, ‘佳’, ‘控’, ‘卫’, ‘的’, ‘鲍’, ‘勃’, ‘-’, ‘库’, ‘西’, ‘奖’, ‘就’, ‘将’, ‘在’, ‘下’, ‘周’, ‘最’, ‘终’, ‘四’, ‘强’, ‘战’, ‘时’, ‘公’, ‘布’, ‘，’], [‘林’, ‘俊’, ‘杰’, ‘为’, ‘电’, ‘影’, ‘《’, ‘夏’, ‘日’, ‘乐’, ‘悠’, ‘悠’, ‘》’, ‘献’, ‘首’, ‘唱’, ‘H’, ‘O’, ‘L’, ‘D’, ‘住’, ‘全’, ‘场’, ‘新’, ‘浪’, ‘娱’, ‘乐’, ‘讯’, ’ ', ‘近’, ‘日’, ‘，’, ‘林’, ‘俊’, ‘杰’, ‘(’, ‘微’, ‘博’, ‘)’, ‘在’, ‘成’, ‘都’, ‘举’, ‘办’, ‘歌’, ‘友’, ‘会’, ‘，’, ‘现’, ‘场’, ‘演’, ‘唱’, ‘了’, ‘其’, ‘为’, ‘由’, ‘马’, ‘楚’, ‘成’, ‘(’, ‘微’, ‘博’, ‘)’, ‘执’, ‘导’, ‘的’, ‘电’, ‘影’, ‘《’, ‘夏’, ‘日’, ‘乐’, ‘悠’, ‘悠’, ‘》’, ‘创’, ‘作’, ‘的’, ‘主’, ‘题’, ‘曲’, ‘《’, ‘L’, ‘O’, ‘V’, ‘E’, ’ ‘, ‘Y’, ‘O’, ‘U’, ’ ‘, ‘Y’, ‘O’, ‘U’, ‘》’, ‘，’], [’《’, ‘海’, ‘湾’, ‘战’, ‘争’, ‘-’, ‘空’, ‘中’, ‘王’, ‘者’, ‘》’, ‘歼’, ‘-’, ‘2’, ‘0’, ‘保’, ‘卫’, ‘升’, ‘空’, ‘一’, ‘款’, ‘绚’, ‘丽’, ‘的’, ‘战’, ‘斗’, ‘画’, ‘面’, ‘，’, ‘横’, ‘越’, ‘海’, ‘、’, ‘陆’, ‘、’, ‘空’, ‘三’, ‘大’, ‘战’, ‘场’, ‘，’, ‘加’, ‘上’, ‘紧’, ‘张’, ‘的’, ‘剧’, ‘情’, ‘，’, ‘就’, ‘构’, ‘成’, ‘了’, ‘今’, ‘天’, ‘小’, ‘编’, ‘给’, ‘各’, ‘大’, ‘玩’, ‘家’, ‘介’, ‘绍’, ‘的’, ‘这’, ‘款’, ‘射’, ‘击’, ‘游’, ‘戏’, ‘：’, ‘《’, ‘海’, ‘湾’, ‘战’, ‘争’, ‘-’, ‘空’, ‘中’, ‘王’, ‘者’, ‘》’, ‘。’, ‘剧’, ‘情’, ‘介’, ‘绍’, ‘：’, ‘海’, ‘湾’, ‘地’, ‘区’, ‘储’, ‘存’, ‘了’, ‘丰’, ‘富’, ‘的’, ‘石’, ‘油’, ‘和’, ‘天’, ‘然’, ‘气’, ‘，’, ‘加’, ‘上’, ‘地’, ‘处’, ‘亚’, ‘非’, ‘欧’, ‘交’, ‘界’, ‘，’, ‘历’, ‘为’, ‘兵’, ‘家’, ‘必’, ‘争’, ‘之’, ‘地’, ‘。’, ‘经’, ‘过’, ‘新’, ‘一’, ‘次’, ‘全’, ‘球’, ‘经’, ‘济’, ‘危’, ‘机’, ‘之’, ‘后’, ‘，’, ‘海’, ‘湾’, ‘战’, ‘争’, ‘一’, ‘触’, ‘即’, ‘发’, ‘，’, ‘中’, ‘国’, ‘也’, ‘被’, ‘卷’, ‘入’, ‘纷’, ‘争’, ‘之’, ‘中’, ‘。’, ‘为’, ‘了’, ‘掌’, ‘握’, ‘实’, ‘地’, ‘军’, ‘情’, ‘，’, ‘我’, ‘军’, ‘派’, ‘出’, ‘了’, ‘有’, ‘空’, ‘中’, ‘王’, ‘者’, ‘之’, ‘称’, ‘的’, ‘歼’, ‘-’, ‘2’, ‘0’, ‘，’, ‘远’, ‘赴’, ‘海’, ‘湾’, ‘地’, ‘区’, ‘…’, ‘…’, ‘更’, ‘多’, ‘精’, ‘彩’, ‘震’, ‘撼’, ‘感’, ‘觉’, ‘，’, ‘立’, ‘即’, ‘下’, ‘载’, ‘该’, ‘款’, ‘游’, ‘戏’, ‘尽’, ‘情’, ‘体’, ‘验’, ‘吧’, ‘。’, ‘玩’, ‘家’, ‘交’, ‘流’, ‘才’, ‘是’, ‘王’, ‘道’, ‘，’, ‘讯’, ‘易’, ‘游’, ‘戏’, ‘玩’, ‘家’, ‘交’, ‘流’, ‘中’, ‘心’]]
labels [‘体育’, ‘娱乐’, ‘游戏’]
data_id [[42, 43, 44, 45, 24, 46, 47, 15, 15, 25, 48, 49, 26, 27, 27, 28, 50, 51, 29, 52, 1, 2, 46, 47, 15, 15, 53, 4, 1, 54, 55, 24, 55, 2, 56, 57, 58, 30, 59, 60, 1, 61, 3, 56, 57, 25, 49, 26, 2, 42, 43, 7, 44, 45, 24, 62, 59, 63, 64, 25, 48, 5, 1], [65, 66, 16, 8, 31, 32, 17, 17, 9, 18, 33, 28, 50, 32, 29, 31, 1, 65, 66, 63, 34, 1, 33, 4, 61, 16, 34, 2, 8, 31, 32, 17, 17, 9, 2, 8, 18, 18, 18, 9, 1], [8, 6, 10, 5, 11, 7, 12, 3, 19, 35, 9, 7, 26, 12, 36, 37, 2, 5, 1, 6, 12, 5, 33, 1, 2, 20, 1, 62, 34, 4, 52, 54, 38, 21, 2, 37, 39, 40, 8, 6, 10, 5, 11, 7, 12, 3, 19, 35, 9, 22, 20, 6, 10, 13, 4, 2, 1, 13, 41, 1, 16, 21, 11, 23, 13, 22, 28, 36, 23, 1, 6, 10, 5, 11, 36, 30, 1, 3, 58, 53, 11, 23, 3, 22, 16, 4, 13, 20, 1, 60, 4, 12, 3, 19, 35, 23, 2, 7, 1, 6, 10, 13, 1, 30, 64, 37, 39, 40, 20, 51, 22, 38, 21, 41, 27, 19, 1, 29, 39, 40, 38, 21, 41, 3]]
label_id [0, 9, 8]
x_pad [[ 0 0 0 0 0 42 43 44 45 24 46 47 15 15 25 48 49 26 27 27 28 50 51 29
52 1 2 46 47 15 15 53 4 1 54 55 24 55 2 56 57 58 30 59 60 1 61 3
56 57 25 49 26 2 42 43 7 44 45 24 62 59 63 64 25 48 5 1]
[ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 65 66 16 8 31 32 17 17 9 18 33 28 50 32 29 31 1 65 66 63 34
1 33 4 61 16 34 2 8 31 32 17 17 9 2 8 18 18 18 9 1]
[ 1 13 41 1 16 21 11 23 13 22 28 36 23 1 6 10 5 11 36 30 1 3 58 53
11 23 3 22 16 4 13 20 1 60 4 12 3 19 35 23 2 7 1 6 10 13 1 30
64 37 39 40 20 51 22 38 21 41 27 19 1 29 39 40 38 21 41 3]]
y_pad [[1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]
[0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]
[0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]
这里的x_pad和y_pad的形式是非常重要的
e：生成批次数据（批量梯度下降，每次更新部分权重）

def batch_iter(x, y, batch_size=1):
    """生成批次数据"""
    #batchs_size每批的训练大小
    print(1111111111111111111111)
    data_len = len(x)
    print(22222222222222222222222)
    print(len(x))
    num_batch = int((data_len - 1) / batch_size) + 1

    indices = np.random.permutation(np.arange(data_len))
    x_shuffle = x[indices]
    y_shuffle = y[indices]

    for i in range(num_batch):
        start_id = i * batch_size
        end_id = min((i + 1) * batch_size, data_len)
        yield x_shuffle[start_id:end_id], y_shuffle[start_id:end_id]
        x和y就是x_pad,  y_pad

x_shuffle[start_id:end_id]:
[[ 1 13 41 1 16 21 11 23 13 22 28 36 23 1 6 10 5 11 36 30 1 3 58 53
11 23 3 22 16 4 13 20 1 60 4 12 3 19 35 23 2 7 1 6 10 13 1 30
64 37 39 40 20 51 22 38 21 41 27 19 1 29 39 40 38 21 41 3]]

y_shuffle[start_id:end_id]:
[[0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]
到这里你可能就明白yield和return的关系和区别了，带yield的函数是一个生成器，而不是一个函数了，这个生成器有一个函数就是next函数，next就相当于“下一步”生成哪个数，这一次的next开始的地方是接着上一次的next停止的地方执行的，所以调用next的时候，生成器并不会从foo函数的开始执行，只是接着上一步停止的地方开始，然后遇到yield后，return出要生成的数，此步就结束。
---------------------
版权声明：本文为CSDN博主「冯爽朗」的原创文章，遵循CC 4.0 by-sa版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/mieleizhi0522/article/details/82142856
f：生成批次数据后

…下面说cnn的模型
第一层嵌入层

词向量映射

      with tf.device('/cpu:0'):
            embedding = tf.get_variable('embedding', [self.config.vocab_size, self.config.embedding_dim])#创建了一个66x6 的矩阵 变量 ：self.config.vocab_size：66
     self.config.embedding_dim：6.一个字向量的维度       
            embedding_inputs = tf.nn.embedding_lookup(embedding, self.input_x)
           
           具体使用方法如下
       

init = tf.global_variables_initializer()
embedding = tf.get_variable('embedding', [66, 6])
embedding_inputs = tf.nn.embedding_lookup(embedding, self.input_x)

---------------------------------------------------------------------
input_x,input_y=batch_iter(xx, yy, batch_size=1)
print('input_x',input_x)


session = tf.Session()
embedding = tf.get_variable('embedding', [66, 6])
init = tf.global_variables_initializer()
session.run(init)

print(session.run(embedding))
embedding_inputs = tf.nn.embedding_lookup(embedding, input_x)
print('embedding_inputs',session.run(embedding_inputs))

input_x
[[ 1 13 41 1 16 21 11 23 13 22 28 36 23 1 6 10 5 11 36 30 1 3 58 53
11 23 3 22 16 4 13 20 1 60 4 12 3 19 35 23 2 7 1 6 10 13 1 30
64 37 39 40 20 51 22 38 21 41 27 19 1 29 39 40 38 21 41 3]]

随机初始化的embedding
[[ 0.23691297 -0.13519788 -0.17290153 -0.06241937 -0.05596925 -0.14235374]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674 -0.10877597]
[-0.26434112 -0.22787322 0.15392077 0.13395855 0.09772652 -0.18564492]
[ 0.14206117 0.11317515 -0.05528121 -0.05800201 0.2836451 -0.11509615]
[-0.15899974 0.0658133 -0.28170696 -0.1541525 -0.07255393 0.13900462]
[ 0.08123055 0.2070345 -0.21851322 0.2779314 -0.08152774 0.24302483]
[-0.13986439 0.1926142 -0.2029559 0.26328248 -0.01662454 0.0631927 ]
[ 0.00827599 -0.19770452 -0.1522938 -0.02354962 0.04330006 0.25638068]
[-0.15717566 0.04652867 -0.02429664 0.23104036 0.03586441 0.08314246]
[-0.12403797 0.01842996 -0.24247035 0.16313967 0.01510769 -0.03565237]
[-0.1276895 -0.03677085 -0.19485027 0.11100221 0.06780049 -0.19717574]
[-0.13580988 0.18453643 -0.17972282 0.26238257 -0.0425182 0.04440424]
[-0.15336685 -0.05864498 0.28834653 0.24184489 0.21647209 -0.09705402]
[-0.2526745 -0.26769823 0.05841517 -0.27402672 0.2282387 0.00096136]
[ 0.13156486 -0.07441346 -0.09149754 0.12538543 -0.10951564 0.1413089 ]
[-0.15028904 -0.14594609 0.24763787 -0.10334942 0.15272108 -0.19012119]
[ 0.00184816 0.11910158 0.25898772 0.17714614 -0.21949929 -0.17058921]
[ 0.09135431 0.09987381 0.12331295 0.2502839 0.0403257 0.01253623]
[-0.09576726 -0.13063213 0.05545032 0.03134313 -0.06821208 0.07928216]
[-0.08161914 0.06568494 0.14552012 0.16656539 -0.1164633 -0.0806136 ]
[-0.10627581 0.06169692 0.24996006 0.07337654 0.14071411 0.04785666]
[ 0.25510186 -0.12553987 -0.24126157 0.21673346 0.14857164 0.25489032]
[-0.02202603 -0.1186287 0.21785933 0.0280619 0.0017722 0.14117366]
[-0.0535892 0.07938072 -0.2857826 -0.24757418 0.15214473 0.12090454]
[ 0.2737862 0.03695834 -0.06650086 -0.19313148 0.23660904 0.10879511]
[-0.28669882 0.07020518 0.06695697 -0.15798223 0.2875694 0.01970571]
[ 0.09032515 -0.15521571 0.0489772 0.18398878 0.22454399 0.0077832 ]
[ 0.08065709 0.19518784 -0.21418074 -0.0028725 -0.13971834 -0.01058358]
[-0.22876623 -0.12268238 -0.22381066 0.18612084 -0.1495206 0.12250948]
[ 0.1780909 0.10810417 -0.11891362 0.15357238 -0.21714506 -0.20010895]
[-0.08518472 -0.04857677 0.2497778 0.20756474 0.18140996 0.07720825]
[-0.21426567 -0.21975863 0.08462957 0.20638472 -0.05134472 -0.2534072 ]
[-0.14999406 -0.03410277 -0.28500742 0.05089357 -0.04791309 0.06126118]
[-0.26611164 0.1410932 -0.06462757 0.15332225 0.23302484 -0.12114364]
[-0.1494725 -0.08487259 -0.19930506 0.01653457 -0.2679908 0.0472686 ]
[ 0.06406471 0.05587634 -0.16891722 -0.09462978 0.04923108 0.13883916]
[-0.18711463 0.15864715 -0.15546672 0.20826581 0.2781158 -0.20220985]
[-0.06363125 0.12919554 0.14578372 -0.01874146 0.14159101 0.15668741]
[-0.11848497 -0.05116902 0.09591931 0.01752862 0.10756946 -0.04643519]
[-0.03719392 0.17145976 -0.18463153 -0.07408771 -0.16771841 0.17747995]
[ 0.22013861 0.16776392 -0.13306917 -0.04167722 -0.17991196 0.25754595]
[ 0.21247488 0.22724444 0.04058126 -0.01098627 0.1850268 0.04928353]
[-0.24639142 -0.04634091 0.00045529 -0.25679696 -0.05259308 -0.24200112]
[ 0.07113779 -0.25441384 -0.1824894 -0.13744785 -0.05465193 -0.28369966]
[ 0.1663951 0.05346683 -0.15225552 0.16807556 -0.15423337 -0.2456753 ]
[-0.03412658 0.08400601 0.1980876 -0.14623068 0.21056029 0.1800251 ]
[ 0.23165774 0.19597226 -0.2495565 -0.16574113 0.1525026 0.2566433 ]
[ 0.0630953 -0.2786023 -0.16081005 -0.02538711 -0.27419418 0.1025582 ]
[-0.25362533 0.02197579 -0.00617951 -0.14856662 0.11153242 0.23738497]
[-0.20182483 0.1876134 0.14892706 -0.15894248 0.10338148 0.28538036]
[-0.16415553 -0.2628631 -0.09699531 0.01421681 -0.01112854 0.09936133]
[-0.08882944 0.02432218 0.20564589 -0.11377952 0.1974028 0.16832677]
[ 0.25703806 -0.00071028 0.16990384 -0.10639983 -0.0831883 -0.09193347]
[-0.03947866 -0.17047392 -0.05723536 0.11617863 -0.1890268 0.18543178]
[ 0.11309478 -0.1983316 0.06863239 -0.14372447 -0.2759583 -0.05608061]
[-0.04162106 0.07707885 -0.27668318 -0.06008887 0.26514602 -0.11434726]
[-0.26868257 0.1296862 0.15483758 -0.15717937 0.21677113 0.19152117]
[-0.05864732 0.0278258 0.19932055 -0.0688611 -0.12320758 -0.12650473]
[-0.19307345 -0.08884493 -0.20212801 0.26287162 -0.25741225 0.25926608]
[ 0.24767327 0.25195622 -0.05632921 -0.23977941 -0.11114164 0.01533225]
[ 0.19079602 -0.16030592 -0.27824146 -0.2586743 0.13144979 -0.10577594]
[ 0.20931724 0.02251613 0.11316517 0.03618035 -0.04443905 -0.10311459]
[ 0.07290316 0.08613291 -0.15033028 -0.14176294 -0.03231186 0.05009356]
[ 0.06722745 0.0038574 -0.24184425 -0.26024234 -0.03536853 -0.01136014]
[ 0.13659015 0.17667219 0.26687944 -0.05812459 0.24382049 -0.02167881]
[-0.27565438 0.03991991 -0.15689795 -0.11174311 -0.1111488 0.18866181]]

embedding_inputs #这是cnn模型第二层的输入向量，可以看到这是一个三维的
[[[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[-0.2526745 -0.26769823 0.05841517 -0.27402672 0.2282387
0.00096136]
[ 0.21247488 0.22724444 0.04058126 -0.01098627 0.1850268
0.04928353]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[ 0.00184816 0.11910158 0.25898772 0.17714614 -0.21949929
-0.17058921]
[ 0.25510186 -0.12553987 -0.24126157 0.21673346 0.14857164
0.25489032]
[-0.13580988 0.18453643 -0.17972282 0.26238257 -0.0425182
0.04440424]
[-0.0535892 0.07938072 -0.2857826 -0.24757418 0.15214473
0.12090454]
[-0.2526745 -0.26769823 0.05841517 -0.27402672 0.2282387
0.00096136]
[-0.02202603 -0.1186287 0.21785933 0.0280619 0.0017722
0.14117366]
[-0.22876623 -0.12268238 -0.22381066 0.18612084 -0.1495206
0.12250948]
[-0.18711463 0.15864715 -0.15546672 0.20826581 0.2781158
-0.20220985]
[-0.0535892 0.07938072 -0.2857826 -0.24757418 0.15214473
0.12090454]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[-0.13986439 0.1926142 -0.2029559 0.26328248 -0.01662454
0.0631927 ]
[-0.1276895 -0.03677085 -0.19485027 0.11100221 0.06780049
-0.19717574]
[ 0.08123055 0.2070345 -0.21851322 0.2779314 -0.08152774
0.24302483]
[-0.13580988 0.18453643 -0.17972282 0.26238257 -0.0425182
0.04440424]
[-0.18711463 0.15864715 -0.15546672 0.20826581 0.2781158
-0.20220985]
[-0.08518472 -0.04857677 0.2497778 0.20756474 0.18140996
0.07720825]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[ 0.14206117 0.11317515 -0.05528121 -0.05800201 0.2836451
-0.11509615]
[-0.19307345 -0.08884493 -0.20212801 0.26287162 -0.25741225
0.25926608]
[-0.03947866 -0.17047392 -0.05723536 0.11617863 -0.1890268
0.18543178]
[-0.13580988 0.18453643 -0.17972282 0.26238257 -0.0425182
0.04440424]
[-0.0535892 0.07938072 -0.2857826 -0.24757418 0.15214473
0.12090454]
[ 0.14206117 0.11317515 -0.05528121 -0.05800201 0.2836451
-0.11509615]
[-0.02202603 -0.1186287 0.21785933 0.0280619 0.0017722
0.14117366]
[ 0.00184816 0.11910158 0.25898772 0.17714614 -0.21949929
-0.17058921]
[-0.15899974 0.0658133 -0.28170696 -0.1541525 -0.07255393
0.13900462]
[-0.2526745 -0.26769823 0.05841517 -0.27402672 0.2282387
0.00096136]
[-0.10627581 0.06169692 0.24996006 0.07337654 0.14071411
0.04785666]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[ 0.19079602 -0.16030592 -0.27824146 -0.2586743 0.13144979
-0.10577594]
[-0.15899974 0.0658133 -0.28170696 -0.1541525 -0.07255393
0.13900462]
[-0.15336685 -0.05864498 0.28834653 0.24184489 0.21647209
-0.09705402]
[ 0.14206117 0.11317515 -0.05528121 -0.05800201 0.2836451
-0.11509615]
[-0.08161914 0.06568494 0.14552012 0.16656539 -0.1164633
-0.0806136 ]
[ 0.06406471 0.05587634 -0.16891722 -0.09462978 0.04923108
0.13883916]
[-0.0535892 0.07938072 -0.2857826 -0.24757418 0.15214473
0.12090454]
[-0.26434112 -0.22787322 0.15392077 0.13395855 0.09772652
-0.18564492]
[ 0.00827599 -0.19770452 -0.1522938 -0.02354962 0.04330006
0.25638068]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[-0.13986439 0.1926142 -0.2029559 0.26328248 -0.01662454
0.0631927 ]
[-0.1276895 -0.03677085 -0.19485027 0.11100221 0.06780049
-0.19717574]
[-0.2526745 -0.26769823 0.05841517 -0.27402672 0.2282387
0.00096136]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[-0.08518472 -0.04857677 0.2497778 0.20756474 0.18140996
0.07720825]
[ 0.13659015 0.17667219 0.26687944 -0.05812459 0.24382049
-0.02167881]
[-0.06363125 0.12919554 0.14578372 -0.01874146 0.14159101
0.15668741]
[-0.03719392 0.17145976 -0.18463153 -0.07408771 -0.16771841
0.17747995]
[ 0.22013861 0.16776392 -0.13306917 -0.04167722 -0.17991196
0.25754595]
[-0.10627581 0.06169692 0.24996006 0.07337654 0.14071411
0.04785666]
[-0.08882944 0.02432218 0.20564589 -0.11377952 0.1974028
0.16832677]
[-0.02202603 -0.1186287 0.21785933 0.0280619 0.0017722
0.14117366]
[-0.11848497 -0.05116902 0.09591931 0.01752862 0.10756946
-0.04643519]
[ 0.25510186 -0.12553987 -0.24126157 0.21673346 0.14857164
0.25489032]
[ 0.21247488 0.22724444 0.04058126 -0.01098627 0.1850268
0.04928353]
[ 0.08065709 0.19518784 -0.21418074 -0.0028725 -0.13971834
-0.01058358]
[-0.08161914 0.06568494 0.14552012 0.16656539 -0.1164633
-0.0806136 ]
[-0.24571183 -0.05539428 0.15108624 -0.28432798 -0.22648674
-0.10877597]
[ 0.1780909 0.10810417 -0.11891362 0.15357238 -0.21714506
-0.20010895]
[-0.03719392 0.17145976 -0.18463153 -0.07408771 -0.16771841
0.17747995]
[ 0.22013861 0.16776392 -0.13306917 -0.04167722 -0.17991196
0.25754595]
[-0.11848497 -0.05116902 0.09591931 0.01752862 0.10756946
-0.04643519]
[ 0.25510186 -0.12553987 -0.24126157 0.21673346 0.14857164
0.25489032]
[ 0.21247488 0.22724444 0.04058126 -0.01098627 0.1850268
0.04928353]
[ 0.14206117 0.11317515 -0.05528121 -0.05800201 0.2836451
-0.11509615]]]