Simulation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Simulation\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### End2End Workflow of Task Completion \n",
    "--------------------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "code_folding": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "from util import *\n",
    "from dlg_manager import *\n",
    "from alg import *\n",
    "from agent import *\n",
    "from user_sim import *\n",
    "from state_tracker import *\n",
    "import random\n",
    "from config import *\n",
    "import numpy as np\n",
    "from nlg import *\n",
    "from six.moves import cPickle as pickle\n",
    "import IPython\n",
    "import matplotlib.pyplot as plt\n",
    "import copy, argparse, json\n",
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load Data\n",
    "-------------------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Action Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "code_folding": [
     0
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "keys = 11\n",
      "Sample of dict:\n",
      "- deny: 9\n",
      "- confirm_question: 2\n",
      "- multiple_choice: 6\n",
      "- greeting: 4\n",
      "- request: 0\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "act_set_path = './data/dia_acts.txt'\n",
    "act_set = text_to_dict(act_set_path)\n",
    "sample_dict(act_set)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### slot set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "code_folding": [
     0
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "keys = 29\n",
      "Sample of dict:\n",
      "- theater: 22\n",
      "- video_format: 24\n",
      "- numberofkids: 15\n",
      "- ticket: 27\n",
      "- seating: 19\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "slots_set_path = \"./data/slot_set.txt\"\n",
    "slot_set = text_to_dict(slots_set_path)\n",
    "sample_dict(slot_set)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### movie dic: info about movie"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "code_folding": [
     0
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "keys = 991\n",
      "Sample of dict:\n",
      "- 568: {'city': 'carbondale', 'theater': 'amc showplace carbondale 8', 'zip': '62901', 'critic_rating': 'good', 'date': 'tomorrow', 'state': 'illinois', 'starttime': '2:30pm', 'genre': 'action', 'moviename': 'london has fallen'}\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "movie_kb_path = \"./data/movie_kb.1k.p\"\n",
    "movie_kb = pickle.load(open(movie_kb_path, 'rb'), encoding=\"latin\")\n",
    "sample_dict(movie_kb, sample_size=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Language Generator (pretrained)\n",
    "-------------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#\n",
    "nlg_model_path = './data/trained_model/nlg/lstm_tanh_relu_[1468202263.38]_2_0.610.p'\n",
    "nlg_model = Nlg()\n",
    "nlg_model.load_nlg_model(nlg_model_path)\n",
    "diaact_nl_pairs_path = \"./data/nlg/dia_act_nl_pairs.v6.json\"\n",
    "nlg_model.load_predefine_act_nl_pairs(diaact_nl_pairs_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Params"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "code_folding": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['grad_clip', 'dia_slot_val', 'reg_cost', 'data_path', 'save_check_point', 'slot_rep', 'max_epochs', 'sdgtype', 'init_rnn', 'cv_fold', 'write_model_dir', 'valid_test', 'pretrained_model_path', 'check_point', 'decay_rate', 'feed_recurrence', 'hidden_size', 'activation_func', 'momentum', 'learning_rate', 'batch_size', 'act_set', 'smooth_eps', 'split_method', 'slot_set', 'eva_metric', 'model'])\n",
      "1\n",
      ".\\checkpoints\\template\\07102016\\\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "model_params = pickle.load(open(nlg_model_path, 'rb'), encoding='latin1')\n",
    "params = model_params['params']\n",
    "print(params.keys())\n",
    "print(params['split_method'])\n",
    "print(params['write_model_dir'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# User Simulator\n",
    "-------------------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### goal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "code_folding": [
     0
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "goals length: 128\n",
      "Sample the first goal: \n",
      "{'request_slots': {}, 'diaact': 'request', 'inform_slots': {'city': 'birmingham', 'numberofpeople': '1', 'theater': 'carmike summit 16', 'state': 'al', 'starttime': 'around 2pm', 'date': 'today', 'moviename': 'zootopia'}}\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "goal_file_path = './data/user_goals_first_turn_template.part.movie.v1.p'\n",
    "all_goal_set = pickle.load(open(goal_file_path, 'rb'), encoding=\"latin\")\n",
    "print(\"goals length: {}\".format(len(all_goal_set)))\n",
    "print(\"Sample the first goal: \\n{}\".format(all_goal_set[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split goal set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "code_folding": [
     0
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "102\n",
      "0\n",
      "26\n",
      "128\n"
     ]
    }
   ],
   "source": [
    "# split goal set\n",
    "split_fold = params.get('split_fold', 5)\n",
    "goal_set = {'train':[], 'valid':[], 'test':[], 'all':[]}\n",
    "for u_goal_id, u_goal in enumerate(all_goal_set):\n",
    "    if u_goal_id % split_fold == 1: goal_set['test'].append(u_goal)\n",
    "    else: goal_set['train'].append(u_goal)\n",
    "    goal_set['all'].append(u_goal)\n",
    "print(len(goal_set['train']))\n",
    "print(len(goal_set['valid']))\n",
    "print(len(goal_set['test']))\n",
    "print(len(goal_set['all']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### user simulator param"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {
    "code_folding": [],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#\n",
    "usersim_params = {}\n",
    "usersim_params['max_turn'] = 40\n",
    "usersim_params['slot_err_prob'] = 0.0\n",
    "# slot_err_mode: 0 for slot_val only; 1 for three errs\n",
    "usersim_params['slot_err_mode'] = 0\n",
    "usersim_params['intent_err_prob'] = 0\n",
    "# run_mode: 0 for default NL; 1 for dia_act; 2 for both\n",
    "usersim_params['run_mode'] = 0\n",
    "# 0 for dia_act level; 1 for NL level\n",
    "usersim_params['act_level'] = 0\n",
    "# train/test/all; default is all\n",
    "usersim_params['learn_phase'] = 'all'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### a movie dictionary for user simulator - slot:possible values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "code_folding": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "keys = 20\n",
      "Sample of dict:\n",
      "- price: ['$20', '$10', 'cheapest', 'adult price is 8', '32']\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "movie_dict_path = './data/user/dicts.v3.p'\n",
    "movie_dictionary = pickle.load(open(movie_dict_path, 'rb'), encoding=\"latin\")\n",
    "samples = sample_dict(movie_dictionary, sample_size=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  Create a User"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "user = RuleSimulator(movie_dictionary, act_set, slot_set, goal_set, usersim_params)\n",
    "user.set_nlg_model(nlg_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Agent\n",
    "-------------------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## param"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "code_folding": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "pretrained model path = None\n"
     ]
    }
   ],
   "source": [
    "#\n",
    "agent_params = {}\n",
    "# maximum length of each dialog (default=20, 0=no maximum length)\n",
    "agent_params['max_turn'] = 100\n",
    "# Epsilon to determine stochasticity of epsilon-greedy agent policies\n",
    "agent_params['epsilon'] = 0\n",
    "# run_mode: 0 for default NL; 1 for dia_act; 2 for both\n",
    "agent_params['agent_run_mode'] = 3\n",
    "# 0 for dia_act level; 1 for NL level\n",
    "agent_params['agent_act_level'] = 0\n",
    "\n",
    "############### DQN #################\n",
    "# the size for experience replay\n",
    "agent_params['experience_replay_pool_size'] = 10000\n",
    "# # the hidden size for DQN\n",
    "agent_params['dqn_hidden_size'] = 60\n",
    "agent_params['batch_size'] = 16\n",
    "# # gamma for DQN\n",
    "agent_params['gamma'] = 0.9\n",
    "# # predict model for DQN\n",
    "agent_params['predict_mode'] = True\n",
    "agent_params['trained_model_path'] = params['pretrained_model_path']\n",
    "#####################################\n",
    "print(\"pretrained model path = {}\".format(agent_params['trained_model_path']))\n",
    "# 0: no warm start; 1: warm start for training\n",
    "agent_params['warm_start'] = 2\n",
    "# run_mode: 0 for NL; 1 for dia_act\n",
    "agent_params['cmd_input_mode'] = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### create an agent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "code_folding": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "agent_run_mode:3\n"
     ]
    }
   ],
   "source": [
    "# agent = RequestBasicsAgent(movie_kb, act_set, slot_set, agent_params)\n",
    "agent = AgentDQN(movie_kb, act_set, slot_set, agent_params)\n",
    "agent.set_nlg_model(nlg_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dialog Manager\n",
    "-------------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dlg_manager = DlgManager(agent, user, act_set, slot_set, movie_kb)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Running Episodes\n",
    "-------------------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Param"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#\n",
    "status = {'successes': 0, 'count': 0, 'cumulative_reward': 0}\n",
    "# the size of validation set\n",
    "simulation_epoch_size = 50\n",
    "# the number of epochs for warm start \n",
    "warm_start_epochs = 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {
    "code_folding": [
     0
    ],
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def run_episodes(num_episode, status, verbose = False, record_training_data = True):\n",
    "    num_success = 0\n",
    "    accu_reward = 0.\n",
    "    accu_episodes = 0.\n",
    "    for i in range(1, num_episode+1):\n",
    "#         print(\"\\nEpisode {} starts!\".format(i))\n",
    "        dlg_manager.init_episode()\n",
    "        done = False\n",
    "        while not done:\n",
    "            done, reward = dlg_manager.step()\n",
    "            accu_reward += reward\n",
    "            if done:\n",
    "                if reward > 0: num_success += 1\n",
    "                accu_episodes += dlg_manager.state_tracker.turn_count\n",
    "#         print(\"Progress: %s / %s, Success rate: %s / %s Avg reward: %.2f Avg turns: %.2f\" % (i+1, num_episode, num_success, i+1, float(accu_reward/(i+1)), float(accu_episodes)/(i+1)))\n",
    "    succ_rate = num_success/num_episode\n",
    "    if verbose: print(\"Success rate: %s / %s Avg reward: %.2f Avg turns: %.2f\" % (num_success, num_episode, succ_rate, float(accu_episodes)/num_episode))\n",
    "    status['successes'] += num_success\n",
    "    status['count'] += num_episode\n",
    "    return num_success, succ_rate"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Run !"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Success rate: 0 / 1 Avg reward: 0.00 Avg turns: 42.00\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<h3>Success Rate: 0.0</h3> "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_episodes = 1\n",
    "agt = 2\n",
    "_, succ_rate = run_episodes(num_episodes, status, verbose = True, record_training_data=False)\n",
    "display.HTML('<h3>{}</h3> '.format(\"Success Rate: {}\".format(succ_rate)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train & Eval\n",
    "-------------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def train_agent(num_episode = 100, num_batches = 100, batch_size = 10, status=status, eval_every = 10, show_every = 100):\n",
    "    succ_rates = []\n",
    "    num_succ = 0\n",
    "    epochs = list(range(eval_every, num_episode+1, eval_every))\n",
    "    for n_epo in epochs:\n",
    "        # collect experience\n",
    "        run_episodes(num_episode=eval_every, status=status, record_training_data = True)\n",
    "        # train\n",
    "        agent.train(batch_size=batch_size, num_batches=num_batches, show_every = show_every)\n",
    "        # eval\n",
    "        curr_num_succ, _ = run_episodes(num_episode=eval_every, status=status, record_training_data = False)\n",
    "        num_succ += curr_num_succ\n",
    "        succ_rates.append(num_succ/n_epo)\n",
    "        \n",
    "    return {\"x\": epochs, 'success_rate': succ_rates}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on : 21\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "cur bellman err 0.0035, experience replay pool 21\n",
      "Train on : 63\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "decay_rate  0.999\n",
      "learning_rate =  0.001\n",
      "cur bellman err 2.2556, experience replay pool 63\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-119-657054a83e45>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0magent\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mclear_exp_pool\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0msucc_rates\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtrain_agent\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnum_episode\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m100\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnum_batches\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m10\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mshow_every\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m10\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0meval_every\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m<ipython-input-117-a6db0a300518>\u001b[0m in \u001b[0;36mtrain_agent\u001b[0;34m(num_episode, num_batches, batch_size, status, eval_every, show_every)\u001b[0m\n\u001b[1;32m      9\u001b[0m         \u001b[0magent\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mbatch_size\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnum_batches\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mnum_batches\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mshow_every\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mshow_every\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     10\u001b[0m         \u001b[0;31m# eval\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 11\u001b[0;31m         \u001b[0mcurr_num_succ\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrun_episodes\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnum_episode\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0meval_every\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mstatus\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mstatus\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrecord_training_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     12\u001b[0m         \u001b[0mnum_succ\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mcurr_num_succ\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m         \u001b[0msucc_rates\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnum_succ\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0mn_epo\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-110-0ecd4df12b1b>\u001b[0m in \u001b[0;36mrun_episodes\u001b[0;34m(num_episode, status, verbose, record_training_data)\u001b[0m\n\u001b[1;32m      8\u001b[0m         \u001b[0mdone\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m         \u001b[0;32mwhile\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mdone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 10\u001b[0;31m             \u001b[0mdone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreward\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdlg_manager\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstep\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     11\u001b[0m             \u001b[0maccu_reward\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mreward\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mdone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dengl11/Dropbox/1_2017_Spring/CS_234/_CS234_Project_/Github/dlg_manager/dlg_manager.py\u001b[0m in \u001b[0;36mstep\u001b[0;34m(self, record_training_data)\u001b[0m\n\u001b[1;32m     94\u001b[0m         \u001b[0;31m########################################################################\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     95\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msys_action\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstate_tracker\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdialog_history_dictionaries\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 96\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0muser_action\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mepisode_done\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdialog_status\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0muser\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msys_action\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     97\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreward\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreward_function\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdialog_status\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     98\u001b[0m         \u001b[0;31m########################################################################\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dengl11/Dropbox/1_2017_Spring/CS_234/_CS234_Project_/Github/user_sim/rule_simulator.py\u001b[0m in \u001b[0;36mnext\u001b[0;34m(self, system_action)\u001b[0m\n\u001b[1;32m    162\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    163\u001b[0m         \u001b[0;31m# add NL to dia_act\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 164\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0madd_nl_to_action\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresponse_action\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    165\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mresponse_action\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mspisode_done\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdialog_status\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    166\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dengl11/Dropbox/1_2017_Spring/CS_234/_CS234_Project_/Github/user_sim/user_simulator.py\u001b[0m in \u001b[0;36madd_nl_to_action\u001b[0;34m(self, user_action)\u001b[0m\n\u001b[1;32m     11\u001b[0m         \"\"\"\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 13\u001b[0;31m         \u001b[0muser_nlg_sentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnlg_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mconvert_diaact_to_nl\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0muser_action\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'usr'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     14\u001b[0m         \u001b[0muser_action\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'nl'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0muser_nlg_sentence\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dengl11/Dropbox/1_2017_Spring/CS_234/_CS234_Project_/Github/nlg/nlg.py\u001b[0m in \u001b[0;36mconvert_diaact_to_nl\u001b[0;34m(self, dia_act, turn_msg)\u001b[0m\n\u001b[1;32m     62\u001b[0m             \u001b[0msentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m\"Oh sorry, there is no ticket available.\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     63\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 64\u001b[0;31m         \u001b[0;32mif\u001b[0m \u001b[0mboolean_in\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0msentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtranslate_diaact\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdia_act\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     65\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0msentence\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dengl11/Dropbox/1_2017_Spring/CS_234/_CS234_Project_/Github/nlg/nlg.py\u001b[0m in \u001b[0;36mtranslate_diaact\u001b[0;34m(self, dia_act)\u001b[0m\n\u001b[1;32m    168\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    169\u001b[0m         \u001b[0;31m#pred_ys, pred_words = nlg_model['model'].forward(inverse_word_dict, dia_act_rep, nlg_model['params'], predict_model=True)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 170\u001b[0;31m         \u001b[0mpred_ys\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpred_words\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbeam_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minverse_word_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdia_act_rep\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mparams\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mpredict_model\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    171\u001b[0m         \u001b[0mpred_sentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m' '\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjoin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpred_words\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    172\u001b[0m         \u001b[0msentence\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpost_process\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpred_sentence\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdia_act\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'inform_slots'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mslot_dict\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dengl11/Dropbox/1_2017_Spring/CS_234/_CS234_Project_/Github/nlg/lstm_decoder_tanh.py\u001b[0m in \u001b[0;36mbeam_forward\u001b[0;34m(self, dict, Xs, params, **kwargs)\u001b[0m\n\u001b[1;32m    288\u001b[0m                 \u001b[0mhout_prev\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mIFOGf\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0md\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;36m3\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0md\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mcell_out\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    289\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 290\u001b[0;31m                 \u001b[0mY\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhout_prev\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mWd\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mbd\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    291\u001b[0m                 \u001b[0mmaxes\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mamax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mY\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0maxis\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mkeepdims\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    292\u001b[0m                 \u001b[0me\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mexp\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mY\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mmaxes\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;31m# for numerical stability shift into good numerical range\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "agent.clear_exp_pool()\n",
    "succ_rates = train_agent(num_episode = 100, num_batches = 10, show_every = 10, eval_every = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABYgAAAH9CAYAAACwdZjWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11ac49518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_learning_curve(succ_rates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}