{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import os\n", "\n", "import itertools\n", "import pickle\n", "import matplotlib.pyplot as plt\n", "\n", "import torch\n", "import torch.nn as nn\n", "import torch.nn.functional as F\n", "import torch.optim as optim\n", "import math \n", "\n", "import sys\n", "sys.path.append('../')\n", "import utils\n", "import wiki_utils\n", "%matplotlib inline" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "corpus = wiki_utils.Texts('./wikitext/')" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "batch_size = 128\n", "sequence_length = 30\n", "grad_clip = 0.1\n", "lr = 4.\n", "best_val_loss = None\n", "log_interval = 100" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "eval_batch_size = 128\n", "train_loader = wiki_utils.TextLoader(corpus.train, batch_size=batch_size)\n", "val_loader = wiki_utils.TextLoader(corpus.valid, batch_size=eval_batch_size)\n", "test_loader = wiki_utils.TextLoader(corpus.test, batch_size=eval_batch_size)" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [], "source": [ "class RNNModel(nn.Module):\n", "\n", " def __init__(self, rnn_type, ntoken, ninp, nhid, nlayers, dropout=0.5):\n", " super(RNNModel, self).__init__()\n", " self.drop = nn.Dropout(dropout)\n", " self.encoder = nn.Embedding(ntoken, ninp)\n", " if rnn_type == 'LSTM':\n", " self.rnn = nn.LSTM(ninp, nhid, nlayers, dropout=dropout)\n", " elif rnn_type == 'GRU':\n", " self.rnn = nn.GRU(ninp, nhid, nlayers, dropout=dropout)\n", " self.decoder = nn.Linear(nhid, ntoken)\n", "\n", " self.init_weights()\n", "\n", " self.rnn_type = rnn_type\n", " self.nhid = nhid\n", " self.nlayers = nlayers\n", "\n", " def init_weights(self):\n", " initrange = 0.1\n", " self.encoder.weight.data.uniform_(-initrange, initrange)\n", " self.decoder.bias.data.fill_(0)\n", " self.decoder.weight.data.uniform_(-initrange, initrange)\n", "\n", " def forward(self, x, hidden=None):\n", " emb = self.drop(self.encoder(x))\n", " output, hidden = self.rnn(emb, hidden)\n", " output = self.drop(output)\n", " decoded = self.decoder(output.view(output.size(0)*output.size(1), output.size(2)))\n", " return decoded.view(output.size(0), output.size(1), decoded.size(1)), hidden\n", "\n", " def init_hidden(self, bsz):\n", " weight = next(self.parameters()).data\n", " if self.rnn_type == 'LSTM':\n", " return (weight.new(self.nlayers, bsz, self.nhid).zero_(),\n", " weight.new(self.nlayers, bsz, self.nhid).zero_())\n", " else:\n", " return weight.new(self.nlayers, bsz, self.nhid).zero_()" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "def evaluate(data_loader):\n", " model.eval()\n", " total_loss = 0\n", " ntokens = len(corpus.dictionary)\n", " hidden = model.init_hidden(eval_batch_size)\n", " for i, (data, targets) in enumerate(data_loader):\n", " output, hidden = model(data)\n", " output_flat = output.view(-1, ntokens)\n", " total_loss += len(data) * criterion(output_flat, targets).item()\n", " return total_loss / len(data_loader)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [], "source": [ "def train():\n", " model.train()\n", " total_loss = 0\n", " ntokens = len(corpus.dictionary)\n", " for batch, (data, targets) in enumerate(train_loader):\n", " model.zero_grad()\n", " output, hidden = model(data)\n", " loss = criterion(output.view(-1, ntokens), targets)\n", " loss.backward()\n", "\n", " # `clip_grad_norm` helps prevent the exploding gradient problem in RNNs / LSTMs.\n", " torch.nn.utils.clip_grad_norm_(model.parameters(), grad_clip)\n", " for p in model.parameters():\n", " p.data.add_(-lr, p.grad.data)\n", "\n", " total_loss += loss.item()\n", "\n", " if batch % log_interval == 0 and batch > 0:\n", " cur_loss = total_loss / log_interval\n", " print('| epoch {:3d} | {:5d}/{:5d} batches | lr {:02.2f} | loss {:5.2f} | ppl {:8.2f}'.format(\n", " epoch, batch, len(train_loader) // sequence_length, lr, cur_loss, math.exp(cur_loss)))\n", " total_loss = 0" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [], "source": [ "ntokens = len(corpus.dictionary)\n", "model = RNNModel('LSTM', ntokens, 128, 128, 2, 0.3)\n", "criterion = nn.CrossEntropyLoss()" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [], "source": [ "def generate(n=50, temp=1.):\n", " model.eval()\n", " x = torch.rand(1, 1).mul(ntokens).long()\n", " hidden = None\n", " out = []\n", " for i in range(n):\n", " output, hidden = model(x, hidden)\n", " s_weights = output.squeeze().data.div(temp).exp()\n", " s_idx = torch.multinomial(s_weights, 1)[0]\n", " x.data.fill_(s_idx)\n", " s = corpus.dictionary.idx2symbol[s_idx]\n", " out.append(s)\n", " return ''.join(out)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "sample:\n", " 4JμュaµსʻzCв\"ėMFôćÍプัвง;ŻśK&3=ã戦ュิ(*mュłžยtí.ńFgĐ£L \n", "\n", "| epoch 1 | 100/ 2807 batches | lr 4.00 | loss 3.60 | ppl 36.43\n", "| epoch 1 | 200/ 2807 batches | lr 4.00 | loss 3.28 | ppl 26.63\n", "| epoch 1 | 300/ 2807 batches | lr 4.00 | loss 3.25 | ppl 25.76\n", "| epoch 1 | 400/ 2807 batches | lr 4.00 | loss 3.22 | ppl 25.12\n", "| epoch 1 | 500/ 2807 batches | lr 4.00 | loss 3.21 | ppl 24.73\n", "| epoch 1 | 600/ 2807 batches | lr 4.00 | loss 3.08 | ppl 21.78\n", "| epoch 1 | 700/ 2807 batches | lr 4.00 | loss 2.98 | ppl 19.69\n", "| epoch 1 | 800/ 2807 batches | lr 4.00 | loss 2.89 | ppl 18.05\n", "| epoch 1 | 900/ 2807 batches | lr 4.00 | loss 2.78 | ppl 16.09\n", "| epoch 1 | 1000/ 2807 batches | lr 4.00 | loss 2.68 | ppl 14.56\n", "| epoch 1 | 1100/ 2807 batches | lr 4.00 | loss 2.58 | ppl 13.24\n", "| epoch 1 | 1200/ 2807 batches | lr 4.00 | loss 2.53 | ppl 12.54\n", "| epoch 1 | 1300/ 2807 batches | lr 4.00 | loss 2.48 | ppl 12.00\n", "| epoch 1 | 1400/ 2807 batches | lr 4.00 | loss 2.44 | ppl 11.44\n", "| epoch 1 | 1500/ 2807 batches | lr 4.00 | loss 2.40 | ppl 11.06\n", "| epoch 1 | 1600/ 2807 batches | lr 4.00 | loss 2.37 | ppl 10.69\n", "| epoch 1 | 1700/ 2807 batches | lr 4.00 | loss 2.33 | ppl 10.33\n", "| epoch 1 | 1800/ 2807 batches | lr 4.00 | loss 2.30 | ppl 10.01\n", "| epoch 1 | 1900/ 2807 batches | lr 4.00 | loss 2.28 | ppl 9.79\n", "| epoch 1 | 2000/ 2807 batches | lr 4.00 | loss 2.25 | ppl 9.49\n", "| epoch 1 | 2100/ 2807 batches | lr 4.00 | loss 2.23 | ppl 9.32\n", "| epoch 1 | 2200/ 2807 batches | lr 4.00 | loss 2.21 | ppl 9.10\n", "| epoch 1 | 2300/ 2807 batches | lr 4.00 | loss 2.20 | ppl 9.04\n", "| epoch 1 | 2400/ 2807 batches | lr 4.00 | loss 2.17 | ppl 8.80\n", "| epoch 1 | 2500/ 2807 batches | lr 4.00 | loss 2.16 | ppl 8.70\n", "| epoch 1 | 2600/ 2807 batches | lr 4.00 | loss 2.15 | ppl 8.55\n", "| epoch 1 | 2700/ 2807 batches | lr 4.00 | loss 2.13 | ppl 8.42\n", "| epoch 1 | 2800/ 2807 batches | lr 4.00 | loss 2.10 | ppl 8.20\n", "-----------------------------------------------------------------------------------------\n", "| end of epoch 1 | valid loss 1.93 | valid ppl 6.90\n", "-----------------------------------------------------------------------------------------\n", "sample:\n", " ond , Por hiirs as \" in jeritiviciaatt Dhois \" sur \n", "\n", "| epoch 2 | 100/ 2807 batches | lr 4.00 | loss 2.12 | ppl 8.30\n", "| epoch 2 | 200/ 2807 batches | lr 4.00 | loss 2.08 | ppl 7.99\n", "| epoch 2 | 300/ 2807 batches | lr 4.00 | loss 2.07 | ppl 7.89\n", "| epoch 2 | 400/ 2807 batches | lr 4.00 | loss 2.06 | ppl 7.83\n", "| epoch 2 | 500/ 2807 batches | lr 4.00 | loss 2.04 | ppl 7.72\n", "| epoch 2 | 600/ 2807 batches | lr 4.00 | loss 2.03 | ppl 7.63\n", "| epoch 2 | 700/ 2807 batches | lr 4.00 | loss 2.02 | ppl 7.57\n", "| epoch 2 | 800/ 2807 batches | lr 4.00 | loss 2.01 | ppl 7.49\n", "| epoch 2 | 900/ 2807 batches | lr 4.00 | loss 2.01 | ppl 7.43\n", "| epoch 2 | 1000/ 2807 batches | lr 4.00 | loss 2.00 | ppl 7.39\n", "| epoch 2 | 1100/ 2807 batches | lr 4.00 | loss 1.98 | ppl 7.27\n", "| epoch 2 | 1200/ 2807 batches | lr 4.00 | loss 1.98 | ppl 7.23\n", "| epoch 2 | 1300/ 2807 batches | lr 4.00 | loss 1.97 | ppl 7.16\n", "| epoch 2 | 1400/ 2807 batches | lr 4.00 | loss 1.95 | ppl 7.06\n", "| epoch 2 | 1500/ 2807 batches | lr 4.00 | loss 1.95 | ppl 7.04\n", "| epoch 2 | 1600/ 2807 batches | lr 4.00 | loss 1.95 | ppl 7.00\n", "| epoch 2 | 1700/ 2807 batches | lr 4.00 | loss 1.94 | ppl 6.94\n", "| epoch 2 | 1800/ 2807 batches | lr 4.00 | loss 1.93 | ppl 6.88\n", "| epoch 2 | 1900/ 2807 batches | lr 4.00 | loss 1.93 | ppl 6.92\n", "| epoch 2 | 2000/ 2807 batches | lr 4.00 | loss 1.92 | ppl 6.80\n", "| epoch 2 | 2100/ 2807 batches | lr 4.00 | loss 1.92 | ppl 6.81\n", "| epoch 2 | 2200/ 2807 batches | lr 4.00 | loss 1.91 | ppl 6.75\n", "| epoch 2 | 2300/ 2807 batches | lr 4.00 | loss 1.91 | ppl 6.77\n", "| epoch 2 | 2400/ 2807 batches | lr 4.00 | loss 1.89 | ppl 6.65\n", "| epoch 2 | 2500/ 2807 batches | lr 4.00 | loss 1.89 | ppl 6.63\n", "| epoch 2 | 2600/ 2807 batches | lr 4.00 | loss 1.89 | ppl 6.64\n", "| epoch 2 | 2700/ 2807 batches | lr 4.00 | loss 1.89 | ppl 6.61\n", "| epoch 2 | 2800/ 2807 batches | lr 4.00 | loss 1.87 | ppl 6.49\n", "-----------------------------------------------------------------------------------------\n", "| end of epoch 2 | valid loss 1.67 | valid ppl 5.30\n", "-----------------------------------------------------------------------------------------\n", "sample:\n", " the 1650 . It in the scited in be subfire n \n", "\n", "| epoch 3 | 100/ 2807 batches | lr 4.00 | loss 1.89 | ppl 6.61\n", "| epoch 3 | 200/ 2807 batches | lr 4.00 | loss 1.86 | ppl 6.42\n", "| epoch 3 | 300/ 2807 batches | lr 4.00 | loss 1.86 | ppl 6.42\n", "| epoch 3 | 400/ 2807 batches | lr 4.00 | loss 1.86 | ppl 6.40\n", "| epoch 3 | 500/ 2807 batches | lr 4.00 | loss 1.85 | ppl 6.38\n", "| epoch 3 | 600/ 2807 batches | lr 4.00 | loss 1.85 | ppl 6.33\n", "| epoch 3 | 700/ 2807 batches | lr 4.00 | loss 1.84 | ppl 6.32\n", "| epoch 3 | 800/ 2807 batches | lr 4.00 | loss 1.84 | ppl 6.29\n", "| epoch 3 | 900/ 2807 batches | lr 4.00 | loss 1.84 | ppl 6.28\n", "| epoch 3 | 1000/ 2807 batches | lr 4.00 | loss 1.84 | ppl 6.28\n", "| epoch 3 | 1100/ 2807 batches | lr 4.00 | loss 1.82 | ppl 6.20\n", "| epoch 3 | 1200/ 2807 batches | lr 4.00 | loss 1.83 | ppl 6.22\n", "| epoch 3 | 1300/ 2807 batches | lr 4.00 | loss 1.82 | ppl 6.18\n", "| epoch 3 | 1400/ 2807 batches | lr 4.00 | loss 1.81 | ppl 6.10\n", "| epoch 3 | 1500/ 2807 batches | lr 4.00 | loss 1.81 | ppl 6.12\n", "| epoch 3 | 1600/ 2807 batches | lr 4.00 | loss 1.81 | ppl 6.13\n", "| epoch 3 | 1700/ 2807 batches | lr 4.00 | loss 1.81 | ppl 6.09\n", "| epoch 3 | 1800/ 2807 batches | lr 4.00 | loss 1.80 | ppl 6.06\n", "| epoch 3 | 1900/ 2807 batches | lr 4.00 | loss 1.81 | ppl 6.13\n", "| epoch 3 | 2000/ 2807 batches | lr 4.00 | loss 1.80 | ppl 6.06\n", "| epoch 3 | 2100/ 2807 batches | lr 4.00 | loss 1.80 | ppl 6.08\n", "| epoch 3 | 2200/ 2807 batches | lr 4.00 | loss 1.80 | ppl 6.05\n", "| epoch 3 | 2300/ 2807 batches | lr 4.00 | loss 1.80 | ppl 6.07\n", "| epoch 3 | 2400/ 2807 batches | lr 4.00 | loss 1.79 | ppl 5.98\n", "| epoch 3 | 2500/ 2807 batches | lr 4.00 | loss 1.79 | ppl 5.98\n", "| epoch 3 | 2600/ 2807 batches | lr 4.00 | loss 1.79 | ppl 6.01\n", "| epoch 3 | 2700/ 2807 batches | lr 4.00 | loss 1.79 | ppl 5.99\n", "| epoch 3 | 2800/ 2807 batches | lr 4.00 | loss 1.78 | ppl 5.90\n", "-----------------------------------------------------------------------------------------\n", "| end of epoch 3 | valid loss 1.57 | valid ppl 4.79\n", "-----------------------------------------------------------------------------------------\n", "sample:\n", " I4 : He for Peyhaling that the storyon \n", "\n", "| epoch 4 | 100/ 2807 batches | lr 4.00 | loss 1.80 | ppl 6.03\n", "| epoch 4 | 200/ 2807 batches | lr 4.00 | loss 1.77 | ppl 5.88\n", "| epoch 4 | 300/ 2807 batches | lr 4.00 | loss 1.77 | ppl 5.90\n", "| epoch 4 | 400/ 2807 batches | lr 4.00 | loss 1.77 | ppl 5.89\n", "| epoch 4 | 500/ 2807 batches | lr 4.00 | loss 1.77 | ppl 5.88\n", "| epoch 4 | 600/ 2807 batches | lr 4.00 | loss 1.77 | ppl 5.84\n", "| epoch 4 | 700/ 2807 batches | lr 4.00 | loss 1.77 | ppl 5.85\n", "| epoch 4 | 800/ 2807 batches | lr 4.00 | loss 1.76 | ppl 5.83\n", "| epoch 4 | 900/ 2807 batches | lr 4.00 | loss 1.76 | ppl 5.83\n", "| epoch 4 | 1000/ 2807 batches | lr 4.00 | loss 1.76 | ppl 5.84\n", "| epoch 4 | 1100/ 2807 batches | lr 4.00 | loss 1.75 | ppl 5.78\n", "| epoch 4 | 1200/ 2807 batches | lr 4.00 | loss 1.76 | ppl 5.81\n", "| epoch 4 | 1300/ 2807 batches | lr 4.00 | loss 1.76 | ppl 5.78\n", "| epoch 4 | 1400/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.71\n", "| epoch 4 | 1500/ 2807 batches | lr 4.00 | loss 1.75 | ppl 5.74\n", "| epoch 4 | 1600/ 2807 batches | lr 4.00 | loss 1.75 | ppl 5.75\n", "| epoch 4 | 1700/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.72\n", "| epoch 4 | 1800/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.71\n", "| epoch 4 | 1900/ 2807 batches | lr 4.00 | loss 1.76 | ppl 5.78\n", "| epoch 4 | 2000/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.71\n", "| epoch 4 | 2100/ 2807 batches | lr 4.00 | loss 1.75 | ppl 5.75\n", "| epoch 4 | 2200/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.71\n", "| epoch 4 | 2300/ 2807 batches | lr 4.00 | loss 1.75 | ppl 5.75\n", "| epoch 4 | 2400/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.67\n", "| epoch 4 | 2500/ 2807 batches | lr 4.00 | loss 1.73 | ppl 5.66\n", "| epoch 4 | 2600/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.72\n", "| epoch 4 | 2700/ 2807 batches | lr 4.00 | loss 1.74 | ppl 5.71\n", "| epoch 4 | 2800/ 2807 batches | lr 4.00 | loss 1.73 | ppl 5.62\n", "-----------------------------------------------------------------------------------------\n", "| end of epoch 4 | valid loss 1.52 | valid ppl 4.56\n", "-----------------------------------------------------------------------------------------\n", "sample:\n", " of Hill Performational diversomen , alfoir Soase \n", "\n", "| epoch 5 | 100/ 2807 batches | lr 4.00 | loss 1.75 | ppl 5.75\n", "| epoch 5 | 200/ 2807 batches | lr 4.00 | loss 1.73 | ppl 5.61\n", "| epoch 5 | 300/ 2807 batches | lr 4.00 | loss 1.73 | ppl 5.64\n", "| epoch 5 | 400/ 2807 batches | lr 4.00 | loss 1.73 | ppl 5.63\n", "| epoch 5 | 500/ 2807 batches | lr 4.00 | loss 1.73 | ppl 5.62\n", "| epoch 5 | 600/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.60\n", "| epoch 5 | 700/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.61\n", "| epoch 5 | 800/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.58\n", "| epoch 5 | 900/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.60\n", "| epoch 5 | 1000/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.61\n", "| epoch 5 | 1100/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.57\n", "| epoch 5 | 1200/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.58\n", "| epoch 5 | 1300/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.56\n", "| epoch 5 | 1400/ 2807 batches | lr 4.00 | loss 1.70 | ppl 5.49\n", "| epoch 5 | 1500/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.53\n", "| epoch 5 | 1600/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.54\n", "| epoch 5 | 1700/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.51\n", "| epoch 5 | 1800/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.51\n", "| epoch 5 | 1900/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.58\n", "| epoch 5 | 2000/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.51\n", "| epoch 5 | 2100/ 2807 batches | lr 4.00 | loss 1.72 | ppl 5.56\n", "| epoch 5 | 2200/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.54\n", "| epoch 5 | 2300/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.55\n", "| epoch 5 | 2400/ 2807 batches | lr 4.00 | loss 1.70 | ppl 5.48\n", "| epoch 5 | 2500/ 2807 batches | lr 4.00 | loss 1.70 | ppl 5.49\n", "| epoch 5 | 2600/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.54\n", "| epoch 5 | 2700/ 2807 batches | lr 4.00 | loss 1.71 | ppl 5.52\n", "| epoch 5 | 2800/ 2807 batches | lr 4.00 | loss 1.70 | ppl 5.45\n", "-----------------------------------------------------------------------------------------\n", "| end of epoch 5 | valid loss 1.49 | valid ppl 4.42\n", "-----------------------------------------------------------------------------------------\n", "sample:\n", " housed to the Jlow , for works an game became or \n", "\n" ] } ], "source": [ "with torch.no_grad():\n", " print('sample:\\n', generate(50), '\\n')\n", "\n", "for epoch in range(1, 6):\n", " train()\n", " val_loss = evaluate(val_loader)\n", " print('-' * 89)\n", " print('| end of epoch {:3d} | valid loss {:5.2f} | valid ppl {:8.2f}'.format(\n", " epoch, val_loss, math.exp(val_loss)))\n", " print('-' * 89)\n", " if not best_val_loss or val_loss < best_val_loss:\n", " best_val_loss = val_loss\n", " else:\n", " # Anneal the learning rate if no improvement has been seen in the validation dataset.\n", " lr /= 4.0\n", " with torch.no_grad():\n", " print('sample:\\n', generate(50), '\\n')\n" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [], "source": [ "t1 = generate(10000, 1.)\n", "t15 = generate(10000, 1.5)\n", "t075 = generate(10000, 0.75)\n", "with open('./generated075.txt', 'w') as outf:\n", " outf.write(t075)\n", "with open('./generated1.txt', 'w') as outf:\n", " outf.write(t1)\n", "with open('./generated15.txt', 'w') as outf:\n", " outf.write(t15)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "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": "3.6.8" } }, "nbformat": 4, "nbformat_minor": 2 }