Day1_2_placeholder.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"Day1_2_placeholder","version":"0.3.2","provenance":[],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"}},"cells":[{"cell_type":"markdown","metadata":{"id":"aQ8_Dhi9XId1","colab_type":"text"},"source":["# placeholder  구현 하기"]},{"cell_type":"markdown","metadata":{"id":"rB8DvIZuXeBh","colab_type":"text"},"source":["### 긍정/부정 Make data (긍정: 1, 부정: 0)\n","\n","---\n","\n"]},{"cell_type":"code","metadata":{"id":"1KaV7-g_sUxD","colab_type":"code","colab":{}},"source":["sample = ['너 오늘 이뻐 보인다',\n","          '나는 오늘 기분이 더러워',\n","          '끝내주는데, 좋은 일이 있나봐',\n","          '나 좋은 일이 생겼어',\n","          '아 오늘 진짜 짜증나',\n","          '환상적인데, 정말 좋은거 같아']\n","\n","label = [[1], [0], [1], [1], [0], [1]]"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"avcrNuViXphK","colab_type":"text"},"source":["### load need modules"]},{"cell_type":"code","metadata":{"id":"BOg4fPGasi6o","colab_type":"code","colab":{}},"source":["import os\n","import tensorflow as tf\n","import numpy as np\n","\n","from tensorflow.keras import preprocessing"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"OVtVn1D0Xsq5","colab_type":"text"},"source":["### Data Preprocessing"]},{"cell_type":"code","metadata":{"id":"Uc1BvbDosk5n","colab_type":"code","outputId":"6aa81321-fb0c-4e02-d7ff-20d0e0eec7dd","executionInfo":{"status":"ok","timestamp":1559358796367,"user_tz":-540,"elapsed":671,"user":{"displayName":"조중현","photoUrl":"","userId":"04275587646849861321"}},"colab":{"base_uri":"https://localhost:8080/","height":68}},"source":["tokenizer = preprocessing.text.Tokenizer() \n","tokenizer.fit_on_texts(sample) \n","sequence = tokenizer.texts_to_sequences(sample) \n","\n","print(\"index text data : \\n\", sequence)\n","print(np.array(sequence).shape)"],"execution_count":3,"outputs":[{"output_type":"stream","text":["index text data : \n"," [[4, 1, 5, 6], [7, 1, 8, 9], [10, 2, 3, 11], [12, 2, 3, 13], [14, 1, 15, 16], [17, 18, 19, 20]]\n","(6, 4)\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"e6mqFlRmsmx2","colab_type":"code","outputId":"2ac2eb58-8cae-4ad5-9592-6ce3d303af58","executionInfo":{"status":"ok","timestamp":1559353861535,"user_tz":-540,"elapsed":636,"user":{"displayName":"조중현","photoUrl":"","userId":"04275587646849861321"}},"colab":{"base_uri":"https://localhost:8080/","height":71}},"source":["word_index = tokenizer.word_index\n","\n","print(\"index of each word : \\n\", word_index)"],"execution_count":0,"outputs":[{"output_type":"stream","text":["index of each word : \n"," {'오늘': 1, '좋은': 2, '일이': 3, '너': 4, '이뻐': 5, '보인다': 6, '나는': 7, '기분이': 8, '더러워': 9, '끝내주는데': 10, '있나봐': 11, '나': 12, '생겼어': 13, '아': 14, '진짜': 15, '짜증나': 16, '환상적인데': 17, '정말': 18, '좋은거': 19, '같아': 20}\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"exaWwfOaswHt","colab_type":"code","outputId":"18188913-42fe-4117-e30e-674f87654ca6","executionInfo":{"status":"ok","timestamp":1559353868018,"user_tz":-540,"elapsed":622,"user":{"displayName":"조중현","photoUrl":"","userId":"04275587646849861321"}},"colab":{"base_uri":"https://localhost:8080/","height":51}},"source":["print(\"label : \", label)\n","print(np.array(label).shape) # np 지원 api shape"],"execution_count":0,"outputs":[{"output_type":"stream","text":["label :  [[1], [0], [1], [1], [0], [1]]\n","(6, 1)\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"nGaCIz-oXxrb","colab_type":"text"},"source":["### feed_dict data processing\n","<b> 장점 : TensorFlow 외부에서 데이터 처리를 수행하여 Python으로 쉽게 처리 할 수 있다.  \n","<b> 단점 : 사용자는 종종 단일 스레드에서 데이터를 처리하고 실행 속도를 저하시키는 데이터 병목 현상을 만든다."]},{"cell_type":"markdown","metadata":{"id":"L3sR9ltXX0XF","colab_type":"text"},"source":["### placeholder define"]},{"cell_type":"code","metadata":{"id":"bKHZmNHXsxwt","colab_type":"code","colab":{}},"source":["sequences = tf.placeholder(dtype=tf.int32, shape=[None, 4], name='sequences') # 길이가 최대 \n","labels = tf.placeholder(dtype=tf.int32, shape=[None,1], name='labels')"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"Pae7qGoYX39H","colab_type":"text"},"source":["### operation define"]},{"cell_type":"code","metadata":{"id":"xN9pG8Rts5FK","colab_type":"code","outputId":"85ccfa5f-195a-40c8-ba09-03864958da55","executionInfo":{"status":"ok","timestamp":1559359234036,"user_tz":-540,"elapsed":616,"user":{"displayName":"조중현","photoUrl":"","userId":"04275587646849861321"}},"colab":{"base_uri":"https://localhost:8080/","height":326}},"source":["print_op_sequence = tf.Print(sequences, [sequences])\n","print_op_labels = tf.Print(labels, [labels])"],"execution_count":5,"outputs":[{"output_type":"stream","text":["WARNING:tensorflow:From <ipython-input-5-c9a0d633b4dc>:1: Print (from tensorflow.python.ops.logging_ops) is deprecated and will be removed after 2018-08-20.\n","Instructions for updating:\n","Use tf.print instead of tf.Print. Note that tf.print returns a no-output operator that directly prints the output. Outside of defuns or eager mode, this operator will not be executed unless it is directly specified in session.run or used as a control dependency for other operators. This is only a concern in graph mode. Below is an example of how to ensure tf.print executes in graph mode:\n","```python\n","    sess = tf.Session()\n","    with sess.as_default():\n","        tensor = tf.range(10)\n","        print_op = tf.print(tensor)\n","        with tf.control_dependencies([print_op]):\n","          out = tf.add(tensor, tensor)\n","        sess.run(out)\n","    ```\n","Additionally, to use tf.print in python 2.7, users must make sure to import\n","the following:\n","\n","  `from __future__ import print_function`\n","\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"bst9d1gPYAu4","colab_type":"text"},"source":["### Session run\n","\n"]},{"cell_type":"code","metadata":{"id":"14RhMcOzuESy","colab_type":"code","outputId":"7ede04e0-746c-4279-8f22-cce4ee6df9b1","executionInfo":{"status":"ok","timestamp":1559359357638,"user_tz":-540,"elapsed":669,"user":{"displayName":"조중현","photoUrl":"","userId":"04275587646849861321"}},"colab":{"base_uri":"https://localhost:8080/","height":255}},"source":["with tf.Session() as sess:\n","  print(\"print sequences : \")\n","  print(sess.run(print_op_sequence, feed_dict={sequences: sequence}))\n","  print(\"print labels : \")\n","  print(sess.run(print_op_labels, feed_dict={labels: label}))"],"execution_count":6,"outputs":[{"output_type":"stream","text":["print sequences : \n","[[ 4  1  5  6]\n"," [ 7  1  8  9]\n"," [10  2  3 11]\n"," [12  2  3 13]\n"," [14  1 15 16]\n"," [17 18 19 20]]\n","print labels : \n","[[1]\n"," [0]\n"," [1]\n"," [1]\n"," [0]\n"," [1]]\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"z9mZLLfK6bJb","colab_type":"text"},"source":["### shuffle data"]},{"cell_type":"code","metadata":{"id":"-1NXaEDI6fjt","colab_type":"code","outputId":"f88a4e6d-fb9b-48f5-a29b-5ea3e149682d","executionInfo":{"status":"ok","timestamp":1556427747979,"user_tz":-540,"elapsed":852,"user":{"displayName":"ChangWook Jun","photoUrl":"https://lh6.googleusercontent.com/-NHLAhAbNfPg/AAAAAAAAAAI/AAAAAAAAAEc/NDLQ8z7-qns/s64/photo.jpg","userId":"00685987924881157185"}},"colab":{"base_uri":"https://localhost:8080/","height":261}},"source":["shuffle_sequence = []\n","shuffle_label = []\n","#############################################################\n","# random index를 만들어 낸다.\n","#############################################################\n","print(random_index)\n","\n","#############################################################\n","# random index를 기반으로 shuffle sequence 와  \n","# shuffle label를 만든다. \n","#############################################################\n","with tf.Session() as sess:\n","  print(\"Print sequences : \")\n","  print(sess.run(print_op_sequence, feed_dict={sequences: shuffle_sequence}))\n","  print(\"Print labels : \")\n","  print(sess.run(print_op_labels, feed_dict={labels: shuffle_label}))"],"execution_count":0,"outputs":[{"output_type":"stream","text":["[5 3 0 1 4 2]\n","Print sequences : \n","[[17 18 19 20]\n"," [12  2  3 13]\n"," [ 4  1  5  6]\n"," [ 7  1  8  9]\n"," [14  1 15 16]\n"," [10  2  3 11]]\n","Print labels : \n","[[1]\n"," [1]\n"," [1]\n"," [0]\n"," [0]\n"," [1]]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"s0ebj-Ug-uoW","colab_type":"code","outputId":"d4483b6a-66f4-4cbb-e4cf-9d4fb9041793","executionInfo":{"status":"ok","timestamp":1559190595275,"user_tz":-540,"elapsed":788,"user":{"displayName":"ChangWook Jun","photoUrl":"https://lh6.googleusercontent.com/-NHLAhAbNfPg/AAAAAAAAAAI/AAAAAAAAAEc/NDLQ8z7-qns/s64/photo.jpg","userId":"00685987924881157185"}},"colab":{"base_uri":"https://localhost:8080/","height":287}},"source":["# ANSWER\n","\n","shuffle_sequence = []\n","shuffle_label = []\n","random_index = np.random.choice(range(len(sequence)), len(sequence), replace=False)\n","print(random_index)\n","\n","for i in random_index:\n","  shuffle_sequence.append(sequence[i])\n","  shuffle_label.append(label[i])\n","  \n","with tf.Session() as sess:\n","  print(\"Print sequences : \")\n","  print(sess.run(print_op_sequence, feed_dict={sequences: shuffle_sequence}))\n","  print(\"Print labels : \")\n","  print(sess.run(print_op_labels, feed_dict={labels: shuffle_label}))"],"execution_count":0,"outputs":[{"output_type":"stream","text":["[3 5 2 4 1 0]\n","Print sequences : \n","[[12  2  3 13]\n"," [17 18 19 20]\n"," [10  2  3 11]\n"," [14  1 15 16]\n"," [ 7  1  8  9]\n"," [ 4  1  5  6]]\n","Print labels : \n","[[1]\n"," [1]\n"," [1]\n"," [0]\n"," [0]\n"," [1]]\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"gNblO2xaPaKg","colab_type":"text"},"source":["### make batch & 3 step"]},{"cell_type":"code","metadata":{"id":"0Q-twkfIPjW1","colab_type":"code","outputId":"11925643-14ba-49ff-b770-a840efb27e73","executionInfo":{"status":"ok","timestamp":1559190684860,"user_tz":-540,"elapsed":828,"user":{"displayName":"ChangWook Jun","photoUrl":"https://lh6.googleusercontent.com/-NHLAhAbNfPg/AAAAAAAAAAI/AAAAAAAAAEc/NDLQ8z7-qns/s64/photo.jpg","userId":"00685987924881157185"}},"colab":{"base_uri":"https://localhost:8080/","height":341}},"source":["BATCH_SIZE = 2\n","\n","shuffle_sequence = []\n","shuffle_label = []\n","index = 0\n","\n","random_index = np.random.choice(range(len(sequence)), len(sequence), replace=False)\n","#############################################################\n","# batch size를 define되어 있는 값으로 만들어서                  \n","# 모든 데이터를 1epoch하도록 처리 한다.\n","# 위와 다른점은 배치 크기만큼 한 step씩 3번에 하도록 만든다.\n","#############################################################"],"execution_count":0,"outputs":[{"output_type":"stream","text":["Print sequences : \n","[[12  2  3 13]\n"," [ 7  1  8  9]]\n","Print labels : \n","[[1]\n"," [0]]\n","Print sequences : \n","[[10  2  3 11]\n"," [17 18 19 20]]\n","Print labels : \n","[[1]\n"," [1]]\n","Print sequences : \n","[[14  1 15 16]\n"," [ 4  1  5  6]]\n","Print labels : \n","[[0]\n"," [1]]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"R2Cg_aDuD05H","colab_type":"code","colab":{}},"source":["# ANSWER\n","\n","BATCH_SIZE = 2\n","\n","shuffle_sequence = []\n","shuffle_label = []\n","index = 0\n","\n","random_index = np.random.choice(range(len(sequence)), len(sequence), replace=False)\n","  \n","for i in random_index:\n","  shuffle_sequence.append(sequence[i])\n","  shuffle_label.append(label[i])\n","  index += 1\n","  \n","  if index==BATCH_SIZE:\n","    with tf.Session() as sess:\n","      print(\"Print sequences : \")\n","      print(sess.run(print_op_sequence, feed_dict={sequences: shuffle_sequence}))\n","      print(\"Print labels : \")\n","      print(sess.run(print_op_labels, feed_dict={labels: shuffle_label}))\n","      index = 0\n","      shuffle_sequence.clear()\n","      shuffle_label.clear()  "],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"BFUs9mTRuTKu","colab_type":"code","colab":{}},"source":["import numpy as np"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"PedKLiJDuZwT","colab_type":"code","colab":{}},"source":["a = [1,2,3,4,5]"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"XbTD1s5uuVv1","colab_type":"code","outputId":"8630d49e-24da-4e7d-fb8c-cc7375405ab5","executionInfo":{"status":"ok","timestamp":1559353764212,"user_tz":-540,"elapsed":614,"user":{"displayName":"조중현","photoUrl":"","userId":"04275587646849861321"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["np.random.choice(range(len(a)), len(a), replace=False)"],"execution_count":0,"outputs":[{"output_type":"execute_result","data":{"text/plain":["array([2, 4, 0, 1, 3])"]},"metadata":{"tags":[]},"execution_count":6}]},{"cell_type":"markdown","metadata":{"id":"q8g7nqTG6U1Z","colab_type":"text"},"source":["###  2 epoch"]},{"cell_type":"code","metadata":{"id":"hFEBbkjQB_cs","colab_type":"code","outputId":"21502518-3125-4b53-c461-bee5fedbe7f7","executionInfo":{"status":"ok","timestamp":1559191623980,"user_tz":-540,"elapsed":671,"user":{"displayName":"ChangWook Jun","photoUrl":"https://lh6.googleusercontent.com/-NHLAhAbNfPg/AAAAAAAAAAI/AAAAAAAAAEc/NDLQ8z7-qns/s64/photo.jpg","userId":"00685987924881157185"}},"colab":{"base_uri":"https://localhost:8080/","height":665}},"source":["EPOCH = 2\n","BATCH_SIZE = 2\n","\n","shuffle_sequence = []\n","shuffle_label = []\n","index = 0\n","\n","random_index = np.random.choice(range(len(sequence)), len(sequence), replace=False)\n","\n","for j in range(EPOCH):\n","  for i in random_index:\n","    shuffle_sequence.append(sequence[i])\n","    shuffle_label.append(label[i])\n","    index += 1\n","\n","    if index==BATCH_SIZE:\n","      with tf.Session() as sess:\n","        print(\"Print sequences : \")\n","        print(sess.run(print_op_sequence, feed_dict={sequences: shuffle_sequence}))\n","        print(\"Print labels : \")\n","        print(sess.run(print_op_labels, feed_dict={labels: shuffle_label}))\n","        index = 0\n","        shuffle_sequence.clear()\n","        shuffle_label.clear()  "],"execution_count":0,"outputs":[{"output_type":"stream","text":["Print sequences : \n","[[10  2  3 11]\n"," [12  2  3 13]]\n","Print labels : \n","[[1]\n"," [1]]\n","Print sequences : \n","[[ 7  1  8  9]\n"," [14  1 15 16]]\n","Print labels : \n","[[0]\n"," [0]]\n","Print sequences : \n","[[17 18 19 20]\n"," [ 4  1  5  6]]\n","Print labels : \n","[[1]\n"," [1]]\n","Print sequences : \n","[[10  2  3 11]\n"," [12  2  3 13]]\n","Print labels : \n","[[1]\n"," [1]]\n","Print sequences : \n","[[ 7  1  8  9]\n"," [14  1 15 16]]\n","Print labels : \n","[[0]\n"," [0]]\n","Print sequences : \n","[[17 18 19 20]\n"," [ 4  1  5  6]]\n","Print labels : \n","[[1]\n"," [1]]\n"],"name":"stdout"}]}]}