seq2seq 训练时 feed 自己的数据

在这个文件加入以下代码https://github.com/tensorflow/models/blob/master/tutorials/rnn/translate/translate.py

def vectorize_data(data, word_idx): #word_idx >=1 ,frist is unknow-token    Q = []    for line in data:        ss = []        for word in line:            if word not in word_idx:                ss.append(0)            else:                ss.append(word_idx[word])        Q.append(ss)    return Qdef load_data(file):    with open(file) as f:        lines = f.readlines()        chinese_data = []        english_data = []        index = 0        for line in lines:            if line == "\n":                continue            words_list=[]            words = line.split(' ')            [words_list.append(word.strip("\n")) for word in words]            if index % 2 == 0:                chinese_data.append(words_list)            elif index%2 == 1:                english_data.append(words_list)            index+=1        return chinese_data,english_data  chinese_data, english_data = load_data('./data.txt')  _PAD = b"_PAD"  _GO = b"_GO"  _EOS = b"_EOS"  _UNK = b"_UNK"  PAD_ID = 0  GO_ID = 1  EOS_ID = 2  UNK_ID = 3  temp = reduce(lambda x, y: x + y, [story for story in chinese_data])  chinese_vocab = set(temp)  chinese_word_idx = dict((c, i + 4) for i, c in enumerate(chinese_vocab))  chinese_word_idx[_PAD]= PAD_ID  chinese_word_idx[_GO] = GO_ID  chinese_word_idx[_EOS] = EOS_ID  chinese_word_idx[_UNK] = UNK_ID  sentence_max_word_number_chinese = max(map(len, chinese_data))  temp = reduce(lambda x, y: x + y, [story for story in english_data])  english_vocab = set(temp)  english_word_idx = dict((c, i + 4) for i, c in enumerate(english_vocab))  english_word_idx[_PAD] = PAD_ID  english_word_idx[_GO] = GO_ID  english_word_idx[_EOS] = EOS_ID  english_word_idx[_UNK] = UNK_ID  sentence_max_word_number_english = max(map(len, english_data))  chinese_ids = vectorize_data(chinese_data,chinese_word_idx)  english_ids = vectorize_data(english_data, english_word_idx)  for line in english_ids:    line.append(EOS_ID)  data_set = [[] for _ in _buckets]  for chinese_line,english_line in zip(chinese_ids,english_ids):    for bucket_id, (source_size, target_size) in enumerate(_buckets):      if len(chinese_line) < source_size and len(english_line) < target_size:          data_set[bucket_id].append([chinese_line, english_line])          break  train_set = data_set # 替换原来的train_set

数据文件的样子

纽约 比 加州 早 三个 小时
New York is 3 hours ahead of California
但 这 没有 让 加州 变慢
but it does not make California slow
有人 22岁 毕业了
Someone graduated at the age of 22
但 等了 五年 才 找到 好的 工作
but waited 5 years before securing a good job
有人 25岁 当上 CEO
Someone became a CEO at 25
却 在 50岁 去世
and died at 50
然而 另一个人 50岁 当上 CEO
While another became a CEO at 50
然后 活到 90岁
and lived to 90 years
有人 依然 单身
Someone is still single
然而 也 有人 已经 结婚
while someone else got married
奥巴马 55岁 退休
Obama retires at 55
但 川普 70岁 开始
but Trump starts at 70
本来 世界上 每个人 在 自己的 时区 工作
Absolutely everyone in this world works based on their Time Zone
身边 有人 可能 看似 走 在 你 前面
People around you might seem to go ahead of you
有人 可能 看似 在 你 后面
some might seem to be behind you.
但 每个人 正在 以 他们的 速度 奔跑 在 他们 自己的 时区
But everyone is running their own RACE in their own TIME.
不要 嫉妒 或 嘲笑 他们
Don’t envy them or mock them
他们 在 他们的 时区 你 在 你的
They are in their TIME ZONE and you are in yours
生命 是 关于 等待 正确的 时机 行动
Life is about waiting for the right moment to act
所以 放轻松
So RELAX
你 没有 落后
You’re not LATE
你 没有 领先
You’re not EARLY
你 非常 准时 在 命运 为 你 安排 的 时区
You are very much ON TIME, and in your TIME ZONE Destiny set up for you.