main.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dinosaur Name Generator"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\fouadhanani\\miniconda3\\envs\\deep-learning\\lib\\site-packages\\tqdm\\auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import string\n",
    "import torch\n",
    "from torch.utils.data import DataLoader\n",
    "from torch import nn\n",
    "from torch import optim\n",
    "from torch.functional import F\n",
    "from torch.nn.utils.rnn import pad_sequence\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Hyper-parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 256\n",
    "n_epochs = 200\n",
    "learning_rate = 0.001\n",
    "hidden_size = 128\n",
    "embedding_dimensions = 16\n",
    "n_layers = 1\n",
    "keep_prob = 0.2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's start by reading the dinosaur names from the `dinosaur.txt` text file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"dinosaur.txt\", \"r\") as file:\n",
    "    names = file.read().lower().splitlines()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's create a list of list of characters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "EOS = \"EOS\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "split_names = []\n",
    "for name in names:\n",
    "    split_name = list(name) + [EOS]\n",
    "    split_names.append(split_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now let's create a vocabulary indicies."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "vocabulary = [EOS] + list(string.ascii_lowercase)\n",
    "idx2char = dict(enumerate(vocabulary))\n",
    "char2idx = {char: idx for idx, char in idx2char.items()}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "encoded_names = []\n",
    "for name in split_names:\n",
    "    encoded_name = torch.tensor([char2idx[char] for char in name])\n",
    "    encoded_names.append(encoded_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "padded_names = pad_sequence(encoded_names, batch_first=True, padding_value=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset and DataLoader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Dataset(torch.utils.data.Dataset):\n",
    "    def __init__(self, samples):\n",
    "        self.samples = samples\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.samples)\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        sample = self.samples[index]\n",
    "        X = sample[:-1]\n",
    "        y = sample[1:]\n",
    "        return X, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = Dataset(samples=padded_names)\n",
    "dataloader = DataLoader(dataset=dataset, batch_size=batch_size, shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Definition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Model(nn.Module):\n",
    "    def __init__(self, vocabulary_size, embedding_dimensions, hidden_size, n_layers, keep_prob):\n",
    "        super(Model, self).__init__()\n",
    "\n",
    "        self.vocabulary_size = vocabulary_size\n",
    "        self.embedding_dimensions = embedding_dimensions\n",
    "        self.hidden_size = hidden_size\n",
    "        self.n_layers = n_layers\n",
    "        self.keep_prob = keep_prob\n",
    "\n",
    "        self.embedding = nn.Embedding(num_embeddings=self.vocabulary_size, embedding_dim=self.embedding_dimensions)\n",
    "        self.lstm = nn.LSTM(input_size=self.embedding_dimensions, hidden_size=self.hidden_size, num_layers=self.n_layers, batch_first=True)\n",
    "        self.dropout = nn.Dropout(p=self.keep_prob)\n",
    "        self.fc = nn.Linear(in_features=self.hidden_size, out_features=self.vocabulary_size)\n",
    "\n",
    "    def forward(self, X, state_previous):\n",
    "        h_previous, c_previous = state_previous\n",
    "        embedding = self.embedding(X)\n",
    "        y_hat, state_current = self.lstm(embedding, (h_previous, c_previous))\n",
    "        y_hat = self.dropout(y_hat)\n",
    "        y_hat = self.fc(y_hat)\n",
    "\n",
    "        return y_hat, state_current"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Model(vocabulary_size=len(char2idx), embedding_dimensions=embedding_dimensions, hidden_size=hidden_size, n_layers=n_layers, keep_prob=keep_prob)\n",
    "criterion = nn.CrossEntropyLoss()\n",
    "optimizer = optim.Adam(params=model.parameters(), lr=learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1/200, 1/7]: loss = 0.293954461812973\n",
      "[1/200, 2/7]: loss = 0.2855260372161865\n",
      "[1/200, 3/7]: loss = 0.2949233949184418\n",
      "[1/200, 4/7]: loss = 0.2975930869579315\n",
      "[1/200, 5/7]: loss = 0.30316901206970215\n",
      "[1/200, 6/7]: loss = 0.3023665249347687\n",
      "[1/200, 7/7]: loss = 0.30097857117652893\n",
      "[2/200, 1/7]: loss = 0.282014399766922\n",
      "[2/200, 2/7]: loss = 0.28387564420700073\n",
      "[2/200, 3/7]: loss = 0.2986801564693451\n",
      "[2/200, 4/7]: loss = 0.29626426100730896\n",
      "[2/200, 5/7]: loss = 0.2904297411441803\n",
      "[2/200, 6/7]: loss = 0.29116541147232056\n",
      "[2/200, 7/7]: loss = 0.34009212255477905\n",
      "[3/200, 1/7]: loss = 0.28534165024757385\n",
      "[3/200, 2/7]: loss = 0.2910679876804352\n",
      "[3/200, 3/7]: loss = 0.2917312681674957\n",
      "[3/200, 4/7]: loss = 0.2900105118751526\n",
      "[3/200, 5/7]: loss = 0.2940482795238495\n",
      "[3/200, 6/7]: loss = 0.29527929425239563\n",
      "[3/200, 7/7]: loss = 0.25871288776397705\n",
      "[4/200, 1/7]: loss = 0.2911921739578247\n",
      "[4/200, 2/7]: loss = 0.29365137219429016\n",
      "[4/200, 3/7]: loss = 0.3045716881752014\n",
      "[4/200, 4/7]: loss = 0.29992249608039856\n",
      "[4/200, 5/7]: loss = 0.2959076166152954\n",
      "[4/200, 6/7]: loss = 0.2964005768299103\n",
      "[4/200, 7/7]: loss = 0.2537968158721924\n",
      "[5/200, 1/7]: loss = 0.30190855264663696\n",
      "[5/200, 2/7]: loss = 0.296481728553772\n",
      "[5/200, 3/7]: loss = 0.28912389278411865\n",
      "[5/200, 4/7]: loss = 0.2908405363559723\n",
      "[5/200, 5/7]: loss = 0.28305330872535706\n",
      "[5/200, 6/7]: loss = 0.28462228178977966\n",
      "[5/200, 7/7]: loss = 0.30085888504981995\n",
      "[6/200, 1/7]: loss = 0.29672959446907043\n",
      "[6/200, 2/7]: loss = 0.28496086597442627\n",
      "[6/200, 3/7]: loss = 0.2953548729419708\n",
      "[6/200, 4/7]: loss = 0.2908012270927429\n",
      "[6/200, 5/7]: loss = 0.2900483012199402\n",
      "[6/200, 6/7]: loss = 0.2908041775226593\n",
      "[6/200, 7/7]: loss = 0.3686819076538086\n",
      "[7/200, 1/7]: loss = 0.29153868556022644\n",
      "[7/200, 2/7]: loss = 0.2950311601161957\n",
      "[7/200, 3/7]: loss = 0.28506794571876526\n",
      "[7/200, 4/7]: loss = 0.2885074317455292\n",
      "[7/200, 5/7]: loss = 0.2894288897514343\n",
      "[7/200, 6/7]: loss = 0.2968287765979767\n",
      "[7/200, 7/7]: loss = 0.360557496547699\n",
      "[8/200, 1/7]: loss = 0.2895401418209076\n",
      "[8/200, 2/7]: loss = 0.2959439754486084\n",
      "[8/200, 3/7]: loss = 0.2875686287879944\n",
      "[8/200, 4/7]: loss = 0.28439652919769287\n",
      "[8/200, 5/7]: loss = 0.29702523350715637\n",
      "[8/200, 6/7]: loss = 0.29407232999801636\n",
      "[8/200, 7/7]: loss = 0.2743872106075287\n",
      "[9/200, 1/7]: loss = 0.285756915807724\n",
      "[9/200, 2/7]: loss = 0.29697710275650024\n",
      "[9/200, 3/7]: loss = 0.3005102276802063\n",
      "[9/200, 4/7]: loss = 0.2939208149909973\n",
      "[9/200, 5/7]: loss = 0.28675585985183716\n",
      "[9/200, 6/7]: loss = 0.2923739552497864\n",
      "[9/200, 7/7]: loss = 0.30059975385665894\n",
      "[10/200, 1/7]: loss = 0.2890412509441376\n",
      "[10/200, 2/7]: loss = 0.28767889738082886\n",
      "[10/200, 3/7]: loss = 0.28949955105781555\n",
      "[10/200, 4/7]: loss = 0.2856703996658325\n",
      "[10/200, 5/7]: loss = 0.2914277911186218\n",
      "[10/200, 6/7]: loss = 0.29424741864204407\n",
      "[10/200, 7/7]: loss = 0.31194597482681274\n",
      "[11/200, 1/7]: loss = 0.28448909521102905\n",
      "[11/200, 2/7]: loss = 0.2869381606578827\n",
      "[11/200, 3/7]: loss = 0.2996288537979126\n",
      "[11/200, 4/7]: loss = 0.2994784414768219\n",
      "[11/200, 5/7]: loss = 0.29289454221725464\n",
      "[11/200, 6/7]: loss = 0.28119805455207825\n",
      "[11/200, 7/7]: loss = 0.24451492726802826\n",
      "[12/200, 1/7]: loss = 0.2925521433353424\n",
      "[12/200, 2/7]: loss = 0.29323455691337585\n",
      "[12/200, 3/7]: loss = 0.2934505343437195\n",
      "[12/200, 4/7]: loss = 0.28512367606163025\n",
      "[12/200, 5/7]: loss = 0.29489243030548096\n",
      "[12/200, 6/7]: loss = 0.2832624614238739\n",
      "[12/200, 7/7]: loss = 0.30004680156707764\n",
      "[13/200, 1/7]: loss = 0.28923073410987854\n",
      "[13/200, 2/7]: loss = 0.30755338072776794\n",
      "[13/200, 3/7]: loss = 0.28447484970092773\n",
      "[13/200, 4/7]: loss = 0.2989928126335144\n",
      "[13/200, 5/7]: loss = 0.290216326713562\n",
      "[13/200, 6/7]: loss = 0.2889595329761505\n",
      "[13/200, 7/7]: loss = 0.22731228172779083\n",
      "[14/200, 1/7]: loss = 0.2939482033252716\n",
      "[14/200, 2/7]: loss = 0.2886256277561188\n",
      "[14/200, 3/7]: loss = 0.29246044158935547\n",
      "[14/200, 4/7]: loss = 0.2983122766017914\n",
      "[14/200, 5/7]: loss = 0.28547045588493347\n",
      "[14/200, 6/7]: loss = 0.2845615744590759\n",
      "[14/200, 7/7]: loss = 0.3124823570251465\n",
      "[15/200, 1/7]: loss = 0.2897251844406128\n",
      "[15/200, 2/7]: loss = 0.28974878787994385\n",
      "[15/200, 3/7]: loss = 0.2886839807033539\n",
      "[15/200, 4/7]: loss = 0.29386284947395325\n",
      "[15/200, 5/7]: loss = 0.2927779257297516\n",
      "[15/200, 6/7]: loss = 0.2804434895515442\n",
      "[15/200, 7/7]: loss = 0.33503997325897217\n",
      "[16/200, 1/7]: loss = 0.2917421758174896\n",
      "[16/200, 2/7]: loss = 0.2932729721069336\n",
      "[16/200, 3/7]: loss = 0.2861384451389313\n",
      "[16/200, 4/7]: loss = 0.2909391522407532\n",
      "[16/200, 5/7]: loss = 0.28999102115631104\n",
      "[16/200, 6/7]: loss = 0.28958675265312195\n",
      "[16/200, 7/7]: loss = 0.26851415634155273\n",
      "[17/200, 1/7]: loss = 0.2819612920284271\n",
      "[17/200, 2/7]: loss = 0.28911998867988586\n",
      "[17/200, 3/7]: loss = 0.29718688130378723\n",
      "[17/200, 4/7]: loss = 0.2941562831401825\n",
      "[17/200, 5/7]: loss = 0.2835847735404968\n",
      "[17/200, 6/7]: loss = 0.29462483525276184\n",
      "[17/200, 7/7]: loss = 0.31190025806427\n",
      "[18/200, 1/7]: loss = 0.288053423166275\n",
      "[18/200, 2/7]: loss = 0.2883942425251007\n",
      "[18/200, 3/7]: loss = 0.2920597195625305\n",
      "[18/200, 4/7]: loss = 0.2901364266872406\n",
      "[18/200, 5/7]: loss = 0.2881453037261963\n",
      "[18/200, 6/7]: loss = 0.29062795639038086\n",
      "[18/200, 7/7]: loss = 0.32661378383636475\n",
      "[19/200, 1/7]: loss = 0.28768426179885864\n",
      "[19/200, 2/7]: loss = 0.2918980121612549\n",
      "[19/200, 3/7]: loss = 0.29034215211868286\n",
      "[19/200, 4/7]: loss = 0.2961052358150482\n",
      "[19/200, 5/7]: loss = 0.2915671169757843\n",
      "[19/200, 6/7]: loss = 0.2952725887298584\n",
      "[19/200, 7/7]: loss = 0.2802121937274933\n",
      "[20/200, 1/7]: loss = 0.2909698486328125\n",
      "[20/200, 2/7]: loss = 0.28602245450019836\n",
      "[20/200, 3/7]: loss = 0.28537359833717346\n",
      "[20/200, 4/7]: loss = 0.28388968110084534\n",
      "[20/200, 5/7]: loss = 0.2872714400291443\n",
      "[20/200, 6/7]: loss = 0.2884473204612732\n",
      "[20/200, 7/7]: loss = 0.30307355523109436\n",
      "[21/200, 1/7]: loss = 0.291811466217041\n",
      "[21/200, 2/7]: loss = 0.2891762852668762\n",
      "[21/200, 3/7]: loss = 0.2902836501598358\n",
      "[21/200, 4/7]: loss = 0.276882529258728\n",
      "[21/200, 5/7]: loss = 0.29030799865722656\n",
      "[21/200, 6/7]: loss = 0.2912042438983917\n",
      "[21/200, 7/7]: loss = 0.28108492493629456\n",
      "[22/200, 1/7]: loss = 0.291240394115448\n",
      "[22/200, 2/7]: loss = 0.2951382100582123\n",
      "[22/200, 3/7]: loss = 0.28729912638664246\n",
      "[22/200, 4/7]: loss = 0.28470757603645325\n",
      "[22/200, 5/7]: loss = 0.28763747215270996\n",
      "[22/200, 6/7]: loss = 0.28409644961357117\n",
      "[22/200, 7/7]: loss = 0.33361315727233887\n",
      "[23/200, 1/7]: loss = 0.2813052833080292\n",
      "[23/200, 2/7]: loss = 0.30332136154174805\n",
      "[23/200, 3/7]: loss = 0.28649625182151794\n",
      "[23/200, 4/7]: loss = 0.28751498460769653\n",
      "[23/200, 5/7]: loss = 0.2909769117832184\n",
      "[23/200, 6/7]: loss = 0.2913718521595001\n",
      "[23/200, 7/7]: loss = 0.299986332654953\n",
      "[24/200, 1/7]: loss = 0.30769798159599304\n",
      "[24/200, 2/7]: loss = 0.29168644547462463\n",
      "[24/200, 3/7]: loss = 0.2985333204269409\n",
      "[24/200, 4/7]: loss = 0.28775477409362793\n",
      "[24/200, 5/7]: loss = 0.2844657301902771\n",
      "[24/200, 6/7]: loss = 0.2978402078151703\n",
      "[24/200, 7/7]: loss = 0.32698357105255127\n",
      "[25/200, 1/7]: loss = 0.3013806939125061\n",
      "[25/200, 2/7]: loss = 0.2878180146217346\n",
      "[25/200, 3/7]: loss = 0.28919753432273865\n",
      "[25/200, 4/7]: loss = 0.28396302461624146\n",
      "[25/200, 5/7]: loss = 0.30119815468788147\n",
      "[25/200, 6/7]: loss = 0.2901487350463867\n",
      "[25/200, 7/7]: loss = 0.3189140856266022\n",
      "[26/200, 1/7]: loss = 0.29664894938468933\n",
      "[26/200, 2/7]: loss = 0.2897261679172516\n",
      "[26/200, 3/7]: loss = 0.2943899929523468\n",
      "[26/200, 4/7]: loss = 0.29320088028907776\n",
      "[26/200, 5/7]: loss = 0.29299449920654297\n",
      "[26/200, 6/7]: loss = 0.29646846652030945\n",
      "[26/200, 7/7]: loss = 0.24938607215881348\n",
      "[27/200, 1/7]: loss = 0.2940041422843933\n",
      "[27/200, 2/7]: loss = 0.31036701798439026\n",
      "[27/200, 3/7]: loss = 0.29555797576904297\n",
      "[27/200, 4/7]: loss = 0.2886093854904175\n",
      "[27/200, 5/7]: loss = 0.29302486777305603\n",
      "[27/200, 6/7]: loss = 0.2943485379219055\n",
      "[27/200, 7/7]: loss = 0.275386244058609\n",
      "[28/200, 1/7]: loss = 0.29573342204093933\n",
      "[28/200, 2/7]: loss = 0.2928965985774994\n",
      "[28/200, 3/7]: loss = 0.2926279902458191\n",
      "[28/200, 4/7]: loss = 0.2886531352996826\n",
      "[28/200, 5/7]: loss = 0.28704023361206055\n",
      "[28/200, 6/7]: loss = 0.2834785580635071\n",
      "[28/200, 7/7]: loss = 0.2972010374069214\n",
      "[29/200, 1/7]: loss = 0.2875388562679291\n",
      "[29/200, 2/7]: loss = 0.29363739490509033\n",
      "[29/200, 3/7]: loss = 0.2927853465080261\n",
      "[29/200, 4/7]: loss = 0.29302287101745605\n",
      "[29/200, 5/7]: loss = 0.2892987132072449\n",
      "[29/200, 6/7]: loss = 0.2931252717971802\n",
      "[29/200, 7/7]: loss = 0.3159697949886322\n",
      "[30/200, 1/7]: loss = 0.2863301634788513\n",
      "[30/200, 2/7]: loss = 0.2899453043937683\n",
      "[30/200, 3/7]: loss = 0.28620633482933044\n",
      "[30/200, 4/7]: loss = 0.2995840013027191\n",
      "[30/200, 5/7]: loss = 0.29678574204444885\n",
      "[30/200, 6/7]: loss = 0.28796982765197754\n",
      "[30/200, 7/7]: loss = 0.30729973316192627\n",
      "[31/200, 1/7]: loss = 0.29307013750076294\n",
      "[31/200, 2/7]: loss = 0.2903311848640442\n",
      "[31/200, 3/7]: loss = 0.286742627620697\n",
      "[31/200, 4/7]: loss = 0.30048441886901855\n",
      "[31/200, 5/7]: loss = 0.2896868586540222\n",
      "[31/200, 6/7]: loss = 0.29524216055870056\n",
      "[31/200, 7/7]: loss = 0.24070629477500916\n",
      "[32/200, 1/7]: loss = 0.28916317224502563\n",
      "[32/200, 2/7]: loss = 0.2941302955150604\n",
      "[32/200, 3/7]: loss = 0.2938304841518402\n",
      "[32/200, 4/7]: loss = 0.2875092327594757\n",
      "[32/200, 5/7]: loss = 0.2912902534008026\n",
      "[32/200, 6/7]: loss = 0.29400330781936646\n",
      "[32/200, 7/7]: loss = 0.2575823962688446\n",
      "[33/200, 1/7]: loss = 0.28949034214019775\n",
      "[33/200, 2/7]: loss = 0.2957257628440857\n",
      "[33/200, 3/7]: loss = 0.2859789729118347\n",
      "[33/200, 4/7]: loss = 0.2955840229988098\n",
      "[33/200, 5/7]: loss = 0.29036232829093933\n",
      "[33/200, 6/7]: loss = 0.2798243463039398\n",
      "[33/200, 7/7]: loss = 0.26229774951934814\n",
      "[34/200, 1/7]: loss = 0.28349682688713074\n",
      "[34/200, 2/7]: loss = 0.29831311106681824\n",
      "[34/200, 3/7]: loss = 0.28712716698646545\n",
      "[34/200, 4/7]: loss = 0.29446855187416077\n",
      "[34/200, 5/7]: loss = 0.29086920619010925\n",
      "[34/200, 6/7]: loss = 0.2939388155937195\n",
      "[34/200, 7/7]: loss = 0.23498092591762543\n",
      "[35/200, 1/7]: loss = 0.29636603593826294\n",
      "[35/200, 2/7]: loss = 0.2999703288078308\n",
      "[35/200, 3/7]: loss = 0.29222553968429565\n",
      "[35/200, 4/7]: loss = 0.2836742103099823\n",
      "[35/200, 5/7]: loss = 0.2877577245235443\n",
      "[35/200, 6/7]: loss = 0.2888639569282532\n",
      "[35/200, 7/7]: loss = 0.2238226979970932\n",
      "[36/200, 1/7]: loss = 0.2838572561740875\n",
      "[36/200, 2/7]: loss = 0.29216712713241577\n",
      "[36/200, 3/7]: loss = 0.2810277044773102\n",
      "[36/200, 4/7]: loss = 0.28763899207115173\n",
      "[36/200, 5/7]: loss = 0.28293663263320923\n",
      "[36/200, 6/7]: loss = 0.28924039006233215\n",
      "[36/200, 7/7]: loss = 0.29068610072135925\n",
      "[37/200, 1/7]: loss = 0.290882408618927\n",
      "[37/200, 2/7]: loss = 0.2946629822254181\n",
      "[37/200, 3/7]: loss = 0.2825123071670532\n",
      "[37/200, 4/7]: loss = 0.2938055098056793\n",
      "[37/200, 5/7]: loss = 0.2868901193141937\n",
      "[37/200, 6/7]: loss = 0.3004414439201355\n",
      "[37/200, 7/7]: loss = 0.27450940012931824\n",
      "[38/200, 1/7]: loss = 0.2906284034252167\n",
      "[38/200, 2/7]: loss = 0.28791770339012146\n",
      "[38/200, 3/7]: loss = 0.28241294622421265\n",
      "[38/200, 4/7]: loss = 0.29475125670433044\n",
      "[38/200, 5/7]: loss = 0.29132065176963806\n",
      "[38/200, 6/7]: loss = 0.2940727174282074\n",
      "[38/200, 7/7]: loss = 0.3157965838909149\n",
      "[39/200, 1/7]: loss = 0.2935337424278259\n",
      "[39/200, 2/7]: loss = 0.29648125171661377\n",
      "[39/200, 3/7]: loss = 0.28774625062942505\n",
      "[39/200, 4/7]: loss = 0.2929639518260956\n",
      "[39/200, 5/7]: loss = 0.2846514582633972\n",
      "[39/200, 6/7]: loss = 0.29687777161598206\n",
      "[39/200, 7/7]: loss = 0.24984413385391235\n",
      "[40/200, 1/7]: loss = 0.2903393805027008\n",
      "[40/200, 2/7]: loss = 0.2917343080043793\n",
      "[40/200, 3/7]: loss = 0.2906581461429596\n",
      "[40/200, 4/7]: loss = 0.29932457208633423\n",
      "[40/200, 5/7]: loss = 0.2887701094150543\n",
      "[40/200, 6/7]: loss = 0.29402145743370056\n",
      "[40/200, 7/7]: loss = 0.29188108444213867\n",
      "[41/200, 1/7]: loss = 0.2860333323478699\n",
      "[41/200, 2/7]: loss = 0.29241716861724854\n",
      "[41/200, 3/7]: loss = 0.29749390482902527\n",
      "[41/200, 4/7]: loss = 0.2899321913719177\n",
      "[41/200, 5/7]: loss = 0.28848180174827576\n",
      "[41/200, 6/7]: loss = 0.2961941361427307\n",
      "[41/200, 7/7]: loss = 0.32302558422088623\n",
      "[42/200, 1/7]: loss = 0.2920624613761902\n",
      "[42/200, 2/7]: loss = 0.29375433921813965\n",
      "[42/200, 3/7]: loss = 0.29219725728034973\n",
      "[42/200, 4/7]: loss = 0.29304829239845276\n",
      "[42/200, 5/7]: loss = 0.2869701683521271\n",
      "[42/200, 6/7]: loss = 0.2845532298088074\n",
      "[42/200, 7/7]: loss = 0.3704102337360382\n",
      "[43/200, 1/7]: loss = 0.2851055860519409\n",
      "[43/200, 2/7]: loss = 0.29845130443573\n",
      "[43/200, 3/7]: loss = 0.29259440302848816\n",
      "[43/200, 4/7]: loss = 0.28950753808021545\n",
      "[43/200, 5/7]: loss = 0.28548911213874817\n",
      "[43/200, 6/7]: loss = 0.29470503330230713\n",
      "[43/200, 7/7]: loss = 0.2917118966579437\n",
      "[44/200, 1/7]: loss = 0.2892548143863678\n",
      "[44/200, 2/7]: loss = 0.2847863435745239\n",
      "[44/200, 3/7]: loss = 0.2871401309967041\n",
      "[44/200, 4/7]: loss = 0.2908734381198883\n",
      "[44/200, 5/7]: loss = 0.2965785264968872\n",
      "[44/200, 6/7]: loss = 0.2975998818874359\n",
      "[44/200, 7/7]: loss = 0.3195573389530182\n",
      "[45/200, 1/7]: loss = 0.283507376909256\n",
      "[45/200, 2/7]: loss = 0.28412413597106934\n",
      "[45/200, 3/7]: loss = 0.2885209023952484\n",
      "[45/200, 4/7]: loss = 0.2933894097805023\n",
      "[45/200, 5/7]: loss = 0.2860543727874756\n",
      "[45/200, 6/7]: loss = 0.28444913029670715\n",
      "[45/200, 7/7]: loss = 0.2753748297691345\n",
      "[46/200, 1/7]: loss = 0.2955063581466675\n",
      "[46/200, 2/7]: loss = 0.2872909605503082\n",
      "[46/200, 3/7]: loss = 0.28247934579849243\n",
      "[46/200, 4/7]: loss = 0.28721073269844055\n",
      "[46/200, 5/7]: loss = 0.28634119033813477\n",
      "[46/200, 6/7]: loss = 0.28247496485710144\n",
      "[46/200, 7/7]: loss = 0.3417206406593323\n",
      "[47/200, 1/7]: loss = 0.2850775420665741\n",
      "[47/200, 2/7]: loss = 0.2916465103626251\n",
      "[47/200, 3/7]: loss = 0.2898895740509033\n",
      "[47/200, 4/7]: loss = 0.28656354546546936\n",
      "[47/200, 5/7]: loss = 0.286567747592926\n",
      "[47/200, 6/7]: loss = 0.2941320240497589\n",
      "[47/200, 7/7]: loss = 0.3120247721672058\n",
      "[48/200, 1/7]: loss = 0.28816458582878113\n",
      "[48/200, 2/7]: loss = 0.2826850414276123\n",
      "[48/200, 3/7]: loss = 0.29632312059402466\n",
      "[48/200, 4/7]: loss = 0.2817210257053375\n",
      "[48/200, 5/7]: loss = 0.2914310097694397\n",
      "[48/200, 6/7]: loss = 0.2920551300048828\n",
      "[48/200, 7/7]: loss = 0.28547123074531555\n",
      "[49/200, 1/7]: loss = 0.2861963212490082\n",
      "[49/200, 2/7]: loss = 0.2965417802333832\n",
      "[49/200, 3/7]: loss = 0.2929631173610687\n",
      "[49/200, 4/7]: loss = 0.2826037108898163\n",
      "[49/200, 5/7]: loss = 0.28259578347206116\n",
      "[49/200, 6/7]: loss = 0.28808122873306274\n",
      "[49/200, 7/7]: loss = 0.2770979106426239\n",
      "[50/200, 1/7]: loss = 0.2908633351325989\n",
      "[50/200, 2/7]: loss = 0.2778189480304718\n",
      "[50/200, 3/7]: loss = 0.29563942551612854\n",
      "[50/200, 4/7]: loss = 0.2812773585319519\n",
      "[50/200, 5/7]: loss = 0.2978403866291046\n",
      "[50/200, 6/7]: loss = 0.2913501262664795\n",
      "[50/200, 7/7]: loss = 0.28007781505584717\n",
      "[51/200, 1/7]: loss = 0.2941417098045349\n",
      "[51/200, 2/7]: loss = 0.28887200355529785\n",
      "[51/200, 3/7]: loss = 0.28482919931411743\n",
      "[51/200, 4/7]: loss = 0.29630741477012634\n",
      "[51/200, 5/7]: loss = 0.28862935304641724\n",
      "[51/200, 6/7]: loss = 0.2814653813838959\n",
      "[51/200, 7/7]: loss = 0.2910735011100769\n",
      "[52/200, 1/7]: loss = 0.295074462890625\n",
      "[52/200, 2/7]: loss = 0.2920418083667755\n",
      "[52/200, 3/7]: loss = 0.2877757251262665\n",
      "[52/200, 4/7]: loss = 0.2851603031158447\n",
      "[52/200, 5/7]: loss = 0.3027512729167938\n",
      "[52/200, 6/7]: loss = 0.28423649072647095\n",
      "[52/200, 7/7]: loss = 0.260965496301651\n",
      "[53/200, 1/7]: loss = 0.29064568877220154\n",
      "[53/200, 2/7]: loss = 0.29191163182258606\n",
      "[53/200, 3/7]: loss = 0.2951257824897766\n",
      "[53/200, 4/7]: loss = 0.28805774450302124\n",
      "[53/200, 5/7]: loss = 0.2939719557762146\n",
      "[53/200, 6/7]: loss = 0.29781511425971985\n",
      "[53/200, 7/7]: loss = 0.27636078000068665\n",
      "[54/200, 1/7]: loss = 0.2927425503730774\n",
      "[54/200, 2/7]: loss = 0.30159375071525574\n",
      "[54/200, 3/7]: loss = 0.2771952152252197\n",
      "[54/200, 4/7]: loss = 0.2914049029350281\n",
      "[54/200, 5/7]: loss = 0.2775573432445526\n",
      "[54/200, 6/7]: loss = 0.29199448227882385\n",
      "[54/200, 7/7]: loss = 0.28567883372306824\n",
      "[55/200, 1/7]: loss = 0.29230040311813354\n",
      "[55/200, 2/7]: loss = 0.2856164276599884\n",
      "[55/200, 3/7]: loss = 0.29112616181373596\n",
      "[55/200, 4/7]: loss = 0.2942938506603241\n",
      "[55/200, 5/7]: loss = 0.2865191400051117\n",
      "[55/200, 6/7]: loss = 0.2951858639717102\n",
      "[55/200, 7/7]: loss = 0.2812744081020355\n",
      "[56/200, 1/7]: loss = 0.28721514344215393\n",
      "[56/200, 2/7]: loss = 0.28956419229507446\n",
      "[56/200, 3/7]: loss = 0.2896929085254669\n",
      "[56/200, 4/7]: loss = 0.2839953899383545\n",
      "[56/200, 5/7]: loss = 0.28434279561042786\n",
      "[56/200, 6/7]: loss = 0.2876521646976471\n",
      "[56/200, 7/7]: loss = 0.2837349772453308\n",
      "[57/200, 1/7]: loss = 0.2876533269882202\n",
      "[57/200, 2/7]: loss = 0.29020094871520996\n",
      "[57/200, 3/7]: loss = 0.28668680787086487\n",
      "[57/200, 4/7]: loss = 0.2923823893070221\n",
      "[57/200, 5/7]: loss = 0.28901028633117676\n",
      "[57/200, 6/7]: loss = 0.2867141664028168\n",
      "[57/200, 7/7]: loss = 0.27664393186569214\n",
      "[58/200, 1/7]: loss = 0.2893697917461395\n",
      "[58/200, 2/7]: loss = 0.28579026460647583\n",
      "[58/200, 3/7]: loss = 0.2857665419578552\n",
      "[58/200, 4/7]: loss = 0.2893577516078949\n",
      "[58/200, 5/7]: loss = 0.2938409745693207\n",
      "[58/200, 6/7]: loss = 0.29278233647346497\n",
      "[58/200, 7/7]: loss = 0.2549276053905487\n",
      "[59/200, 1/7]: loss = 0.28204280138015747\n",
      "[59/200, 2/7]: loss = 0.29083219170570374\n",
      "[59/200, 3/7]: loss = 0.296364963054657\n",
      "[59/200, 4/7]: loss = 0.2941589653491974\n",
      "[59/200, 5/7]: loss = 0.29761698842048645\n",
      "[59/200, 6/7]: loss = 0.2958161234855652\n",
      "[59/200, 7/7]: loss = 0.28669974207878113\n",
      "[60/200, 1/7]: loss = 0.28428810834884644\n",
      "[60/200, 2/7]: loss = 0.2850668430328369\n",
      "[60/200, 3/7]: loss = 0.29082009196281433\n",
      "[60/200, 4/7]: loss = 0.29513511061668396\n",
      "[60/200, 5/7]: loss = 0.28141987323760986\n",
      "[60/200, 6/7]: loss = 0.2837020754814148\n",
      "[60/200, 7/7]: loss = 0.3178761303424835\n",
      "[61/200, 1/7]: loss = 0.2869790494441986\n",
      "[61/200, 2/7]: loss = 0.29358547925949097\n",
      "[61/200, 3/7]: loss = 0.2954505681991577\n",
      "[61/200, 4/7]: loss = 0.29352349042892456\n",
      "[61/200, 5/7]: loss = 0.2928380072116852\n",
      "[61/200, 6/7]: loss = 0.2858050763607025\n",
      "[61/200, 7/7]: loss = 0.25133323669433594\n",
      "[62/200, 1/7]: loss = 0.28054091334342957\n",
      "[62/200, 2/7]: loss = 0.28176090121269226\n",
      "[62/200, 3/7]: loss = 0.3021436035633087\n",
      "[62/200, 4/7]: loss = 0.2927241921424866\n",
      "[62/200, 5/7]: loss = 0.2873760163784027\n",
      "[62/200, 6/7]: loss = 0.28597214818000793\n",
      "[62/200, 7/7]: loss = 0.273227721452713\n",
      "[63/200, 1/7]: loss = 0.28769025206565857\n",
      "[63/200, 2/7]: loss = 0.29017552733421326\n",
      "[63/200, 3/7]: loss = 0.2981908321380615\n",
      "[63/200, 4/7]: loss = 0.29099327325820923\n",
      "[63/200, 5/7]: loss = 0.2907671630382538\n",
      "[63/200, 6/7]: loss = 0.2891843318939209\n",
      "[63/200, 7/7]: loss = 0.340664267539978\n",
      "[64/200, 1/7]: loss = 0.2908551096916199\n",
      "[64/200, 2/7]: loss = 0.2813662588596344\n",
      "[64/200, 3/7]: loss = 0.2808721363544464\n",
      "[64/200, 4/7]: loss = 0.29553505778312683\n",
      "[64/200, 5/7]: loss = 0.29134348034858704\n",
      "[64/200, 6/7]: loss = 0.2975815534591675\n",
      "[64/200, 7/7]: loss = 0.36946868896484375\n",
      "[65/200, 1/7]: loss = 0.29221734404563904\n",
      "[65/200, 2/7]: loss = 0.27915993332862854\n",
      "[65/200, 3/7]: loss = 0.30103209614753723\n",
      "[65/200, 4/7]: loss = 0.28458651900291443\n",
      "[65/200, 5/7]: loss = 0.28973349928855896\n",
      "[65/200, 6/7]: loss = 0.28514471650123596\n",
      "[65/200, 7/7]: loss = 0.26126164197921753\n",
      "[66/200, 1/7]: loss = 0.2876061797142029\n",
      "[66/200, 2/7]: loss = 0.2820088863372803\n",
      "[66/200, 3/7]: loss = 0.2811024785041809\n",
      "[66/200, 4/7]: loss = 0.2887893319129944\n",
      "[66/200, 5/7]: loss = 0.2941458523273468\n",
      "[66/200, 6/7]: loss = 0.2982914447784424\n",
      "[66/200, 7/7]: loss = 0.33765292167663574\n",
      "[67/200, 1/7]: loss = 0.29072925448417664\n",
      "[67/200, 2/7]: loss = 0.28351470828056335\n",
      "[67/200, 3/7]: loss = 0.2841511368751526\n",
      "[67/200, 4/7]: loss = 0.2899780869483948\n",
      "[67/200, 5/7]: loss = 0.28806033730506897\n",
      "[67/200, 6/7]: loss = 0.2831902801990509\n",
      "[67/200, 7/7]: loss = 0.2556227445602417\n",
      "[68/200, 1/7]: loss = 0.28678959608078003\n",
      "[68/200, 2/7]: loss = 0.2875068783760071\n",
      "[68/200, 3/7]: loss = 0.290785551071167\n",
      "[68/200, 4/7]: loss = 0.289698988199234\n",
      "[68/200, 5/7]: loss = 0.2846500277519226\n",
      "[68/200, 6/7]: loss = 0.2905203700065613\n",
      "[68/200, 7/7]: loss = 0.2999818027019501\n",
      "[69/200, 1/7]: loss = 0.2765171229839325\n",
      "[69/200, 2/7]: loss = 0.2876458168029785\n",
      "[69/200, 3/7]: loss = 0.2884550392627716\n",
      "[69/200, 4/7]: loss = 0.2825060784816742\n",
      "[69/200, 5/7]: loss = 0.2912609279155731\n",
      "[69/200, 6/7]: loss = 0.2902240455150604\n",
      "[69/200, 7/7]: loss = 0.25388458371162415\n",
      "[70/200, 1/7]: loss = 0.28391432762145996\n",
      "[70/200, 2/7]: loss = 0.281711608171463\n",
      "[70/200, 3/7]: loss = 0.28401535749435425\n",
      "[70/200, 4/7]: loss = 0.2879803776741028\n",
      "[70/200, 5/7]: loss = 0.2866947650909424\n",
      "[70/200, 6/7]: loss = 0.28628262877464294\n",
      "[70/200, 7/7]: loss = 0.264580100774765\n",
      "[71/200, 1/7]: loss = 0.28199249505996704\n",
      "[71/200, 2/7]: loss = 0.29053324460983276\n",
      "[71/200, 3/7]: loss = 0.28163424134254456\n",
      "[71/200, 4/7]: loss = 0.28195029497146606\n",
      "[71/200, 5/7]: loss = 0.28358137607574463\n",
      "[71/200, 6/7]: loss = 0.29105067253112793\n",
      "[71/200, 7/7]: loss = 0.28987646102905273\n",
      "[72/200, 1/7]: loss = 0.2882866859436035\n",
      "[72/200, 2/7]: loss = 0.28013846278190613\n",
      "[72/200, 3/7]: loss = 0.29961609840393066\n",
      "[72/200, 4/7]: loss = 0.2834509015083313\n",
      "[72/200, 5/7]: loss = 0.2819810211658478\n",
      "[72/200, 6/7]: loss = 0.2966001629829407\n",
      "[72/200, 7/7]: loss = 0.22614794969558716\n",
      "[73/200, 1/7]: loss = 0.284102201461792\n",
      "[73/200, 2/7]: loss = 0.2798410654067993\n",
      "[73/200, 3/7]: loss = 0.29437941312789917\n",
      "[73/200, 4/7]: loss = 0.292684942483902\n",
      "[73/200, 5/7]: loss = 0.2884831428527832\n",
      "[73/200, 6/7]: loss = 0.2941591739654541\n",
      "[73/200, 7/7]: loss = 0.21879498660564423\n",
      "[74/200, 1/7]: loss = 0.27915477752685547\n",
      "[74/200, 2/7]: loss = 0.2890269160270691\n",
      "[74/200, 3/7]: loss = 0.2889139652252197\n",
      "[74/200, 4/7]: loss = 0.2868245244026184\n",
      "[74/200, 5/7]: loss = 0.28333497047424316\n",
      "[74/200, 6/7]: loss = 0.293598473072052\n",
      "[74/200, 7/7]: loss = 0.23726597428321838\n",
      "[75/200, 1/7]: loss = 0.28328511118888855\n",
      "[75/200, 2/7]: loss = 0.2878604829311371\n",
      "[75/200, 3/7]: loss = 0.2949153184890747\n",
      "[75/200, 4/7]: loss = 0.28832364082336426\n",
      "[75/200, 5/7]: loss = 0.2928432524204254\n",
      "[75/200, 6/7]: loss = 0.29037564992904663\n",
      "[75/200, 7/7]: loss = 0.2946411073207855\n",
      "[76/200, 1/7]: loss = 0.28674736618995667\n",
      "[76/200, 2/7]: loss = 0.2838045060634613\n",
      "[76/200, 3/7]: loss = 0.29289188981056213\n",
      "[76/200, 4/7]: loss = 0.29174911975860596\n",
      "[76/200, 5/7]: loss = 0.2815043032169342\n",
      "[76/200, 6/7]: loss = 0.2920636236667633\n",
      "[76/200, 7/7]: loss = 0.2920471429824829\n",
      "[77/200, 1/7]: loss = 0.28378981351852417\n",
      "[77/200, 2/7]: loss = 0.2883739173412323\n",
      "[77/200, 3/7]: loss = 0.2870323359966278\n",
      "[77/200, 4/7]: loss = 0.2883162796497345\n",
      "[77/200, 5/7]: loss = 0.2812686562538147\n",
      "[77/200, 6/7]: loss = 0.2840001583099365\n",
      "[77/200, 7/7]: loss = 0.253127783536911\n",
      "[78/200, 1/7]: loss = 0.28853291273117065\n",
      "[78/200, 2/7]: loss = 0.28103163838386536\n",
      "[78/200, 3/7]: loss = 0.2801310122013092\n",
      "[78/200, 4/7]: loss = 0.29044651985168457\n",
      "[78/200, 5/7]: loss = 0.2912044823169708\n",
      "[78/200, 6/7]: loss = 0.28636446595191956\n",
      "[78/200, 7/7]: loss = 0.31272295117378235\n",
      "[79/200, 1/7]: loss = 0.2892971336841583\n",
      "[79/200, 2/7]: loss = 0.28203579783439636\n",
      "[79/200, 3/7]: loss = 0.2877182960510254\n",
      "[79/200, 4/7]: loss = 0.28123658895492554\n",
      "[79/200, 5/7]: loss = 0.2920759320259094\n",
      "[79/200, 6/7]: loss = 0.2818191945552826\n",
      "[79/200, 7/7]: loss = 0.35042497515678406\n",
      "[80/200, 1/7]: loss = 0.29370734095573425\n",
      "[80/200, 2/7]: loss = 0.28498682379722595\n",
      "[80/200, 3/7]: loss = 0.28286460041999817\n",
      "[80/200, 4/7]: loss = 0.2832558751106262\n",
      "[80/200, 5/7]: loss = 0.29244834184646606\n",
      "[80/200, 6/7]: loss = 0.2884277403354645\n",
      "[80/200, 7/7]: loss = 0.297976553440094\n",
      "[81/200, 1/7]: loss = 0.2964780926704407\n",
      "[81/200, 2/7]: loss = 0.28593000769615173\n",
      "[81/200, 3/7]: loss = 0.28232628107070923\n",
      "[81/200, 4/7]: loss = 0.2812712490558624\n",
      "[81/200, 5/7]: loss = 0.2853846251964569\n",
      "[81/200, 6/7]: loss = 0.28787991404533386\n",
      "[81/200, 7/7]: loss = 0.32068660855293274\n",
      "[82/200, 1/7]: loss = 0.29015082120895386\n",
      "[82/200, 2/7]: loss = 0.2921352684497833\n",
      "[82/200, 3/7]: loss = 0.2830861806869507\n",
      "[82/200, 4/7]: loss = 0.29119738936424255\n",
      "[82/200, 5/7]: loss = 0.28938761353492737\n",
      "[82/200, 6/7]: loss = 0.2969305217266083\n",
      "[82/200, 7/7]: loss = 0.2721082270145416\n",
      "[83/200, 1/7]: loss = 0.2937929630279541\n",
      "[83/200, 2/7]: loss = 0.28722506761550903\n",
      "[83/200, 3/7]: loss = 0.2788979113101959\n",
      "[83/200, 4/7]: loss = 0.2894408106803894\n",
      "[83/200, 5/7]: loss = 0.27713316679000854\n",
      "[83/200, 6/7]: loss = 0.28849631547927856\n",
      "[83/200, 7/7]: loss = 0.28166377544403076\n",
      "[84/200, 1/7]: loss = 0.28103381395339966\n",
      "[84/200, 2/7]: loss = 0.2860414683818817\n",
      "[84/200, 3/7]: loss = 0.30093255639076233\n",
      "[84/200, 4/7]: loss = 0.2932688891887665\n",
      "[84/200, 5/7]: loss = 0.2935743033885956\n",
      "[84/200, 6/7]: loss = 0.2790524363517761\n",
      "[84/200, 7/7]: loss = 0.2996951937675476\n",
      "[85/200, 1/7]: loss = 0.29571375250816345\n",
      "[85/200, 2/7]: loss = 0.2884540557861328\n",
      "[85/200, 3/7]: loss = 0.27627992630004883\n",
      "[85/200, 4/7]: loss = 0.2781084179878235\n",
      "[85/200, 5/7]: loss = 0.28928443789482117\n",
      "[85/200, 6/7]: loss = 0.2943340241909027\n",
      "[85/200, 7/7]: loss = 0.2983788847923279\n",
      "[86/200, 1/7]: loss = 0.286456823348999\n",
      "[86/200, 2/7]: loss = 0.28559479117393494\n",
      "[86/200, 3/7]: loss = 0.2764107584953308\n",
      "[86/200, 4/7]: loss = 0.2860791087150574\n",
      "[86/200, 5/7]: loss = 0.2853885591030121\n",
      "[86/200, 6/7]: loss = 0.2933024764060974\n",
      "[86/200, 7/7]: loss = 0.26285770535469055\n",
      "[87/200, 1/7]: loss = 0.2876703143119812\n",
      "[87/200, 2/7]: loss = 0.2953214943408966\n",
      "[87/200, 3/7]: loss = 0.28032830357551575\n",
      "[87/200, 4/7]: loss = 0.2848832309246063\n",
      "[87/200, 5/7]: loss = 0.28792569041252136\n",
      "[87/200, 6/7]: loss = 0.27726301550865173\n",
      "[87/200, 7/7]: loss = 0.3014545440673828\n",
      "[88/200, 1/7]: loss = 0.29717352986335754\n",
      "[88/200, 2/7]: loss = 0.2763993740081787\n",
      "[88/200, 3/7]: loss = 0.27595943212509155\n",
      "[88/200, 4/7]: loss = 0.2807428538799286\n",
      "[88/200, 5/7]: loss = 0.2842457592487335\n",
      "[88/200, 6/7]: loss = 0.28103259205818176\n",
      "[88/200, 7/7]: loss = 0.2725432217121124\n",
      "[89/200, 1/7]: loss = 0.28336548805236816\n",
      "[89/200, 2/7]: loss = 0.2831225097179413\n",
      "[89/200, 3/7]: loss = 0.2785090208053589\n",
      "[89/200, 4/7]: loss = 0.29270732402801514\n",
      "[89/200, 5/7]: loss = 0.2822135090827942\n",
      "[89/200, 6/7]: loss = 0.2845155596733093\n",
      "[89/200, 7/7]: loss = 0.2696293890476227\n",
      "[90/200, 1/7]: loss = 0.2838228940963745\n",
      "[90/200, 2/7]: loss = 0.29131290316581726\n",
      "[90/200, 3/7]: loss = 0.28858327865600586\n",
      "[90/200, 4/7]: loss = 0.29136520624160767\n",
      "[90/200, 5/7]: loss = 0.2812716066837311\n",
      "[90/200, 6/7]: loss = 0.2786397933959961\n",
      "[90/200, 7/7]: loss = 0.2861770689487457\n",
      "[91/200, 1/7]: loss = 0.291454017162323\n",
      "[91/200, 2/7]: loss = 0.29461219906806946\n",
      "[91/200, 3/7]: loss = 0.2835957705974579\n",
      "[91/200, 4/7]: loss = 0.2778528332710266\n",
      "[91/200, 5/7]: loss = 0.2779484987258911\n",
      "[91/200, 6/7]: loss = 0.280498743057251\n",
      "[91/200, 7/7]: loss = 0.3090652525424957\n",
      "[92/200, 1/7]: loss = 0.2841838598251343\n",
      "[92/200, 2/7]: loss = 0.2868606150150299\n",
      "[92/200, 3/7]: loss = 0.29076796770095825\n",
      "[92/200, 4/7]: loss = 0.2901006042957306\n",
      "[92/200, 5/7]: loss = 0.28859663009643555\n",
      "[92/200, 6/7]: loss = 0.2903408110141754\n",
      "[92/200, 7/7]: loss = 0.2812461256980896\n",
      "[93/200, 1/7]: loss = 0.2885224521160126\n",
      "[93/200, 2/7]: loss = 0.2817896008491516\n",
      "[93/200, 3/7]: loss = 0.28186410665512085\n",
      "[93/200, 4/7]: loss = 0.28680551052093506\n",
      "[93/200, 5/7]: loss = 0.29993799328804016\n",
      "[93/200, 6/7]: loss = 0.29189106822013855\n",
      "[93/200, 7/7]: loss = 0.26069632172584534\n",
      "[94/200, 1/7]: loss = 0.29113340377807617\n",
      "[94/200, 2/7]: loss = 0.28780969977378845\n",
      "[94/200, 3/7]: loss = 0.28142258524894714\n",
      "[94/200, 4/7]: loss = 0.28723829984664917\n",
      "[94/200, 5/7]: loss = 0.28876182436943054\n",
      "[94/200, 6/7]: loss = 0.2867445945739746\n",
      "[94/200, 7/7]: loss = 0.2727409303188324\n",
      "[95/200, 1/7]: loss = 0.28589335083961487\n",
      "[95/200, 2/7]: loss = 0.2836611568927765\n",
      "[95/200, 3/7]: loss = 0.28538456559181213\n",
      "[95/200, 4/7]: loss = 0.28311687707901\n",
      "[95/200, 5/7]: loss = 0.2885342538356781\n",
      "[95/200, 6/7]: loss = 0.2885778546333313\n",
      "[95/200, 7/7]: loss = 0.3047714829444885\n",
      "[96/200, 1/7]: loss = 0.28280743956565857\n",
      "[96/200, 2/7]: loss = 0.275338739156723\n",
      "[96/200, 3/7]: loss = 0.2810760736465454\n",
      "[96/200, 4/7]: loss = 0.291573703289032\n",
      "[96/200, 5/7]: loss = 0.28971290588378906\n",
      "[96/200, 6/7]: loss = 0.28125956654548645\n",
      "[96/200, 7/7]: loss = 0.3186536431312561\n",
      "[97/200, 1/7]: loss = 0.29023054242134094\n",
      "[97/200, 2/7]: loss = 0.2800845801830292\n",
      "[97/200, 3/7]: loss = 0.28579387068748474\n",
      "[97/200, 4/7]: loss = 0.2836954891681671\n",
      "[97/200, 5/7]: loss = 0.2798275947570801\n",
      "[97/200, 6/7]: loss = 0.2859970033168793\n",
      "[97/200, 7/7]: loss = 0.2851124405860901\n",
      "[98/200, 1/7]: loss = 0.29123547673225403\n",
      "[98/200, 2/7]: loss = 0.2913306653499603\n",
      "[98/200, 3/7]: loss = 0.28225451707839966\n",
      "[98/200, 4/7]: loss = 0.2865754961967468\n",
      "[98/200, 5/7]: loss = 0.28825172781944275\n",
      "[98/200, 6/7]: loss = 0.29337817430496216\n",
      "[98/200, 7/7]: loss = 0.28706279397010803\n",
      "[99/200, 1/7]: loss = 0.28635451197624207\n",
      "[99/200, 2/7]: loss = 0.2885991036891937\n",
      "[99/200, 3/7]: loss = 0.2943730652332306\n",
      "[99/200, 4/7]: loss = 0.2830442786216736\n",
      "[99/200, 5/7]: loss = 0.28975746035575867\n",
      "[99/200, 6/7]: loss = 0.28824958205223083\n",
      "[99/200, 7/7]: loss = 0.26233941316604614\n",
      "[100/200, 1/7]: loss = 0.29071763157844543\n",
      "[100/200, 2/7]: loss = 0.2884368896484375\n",
      "[100/200, 3/7]: loss = 0.2870805859565735\n",
      "[100/200, 4/7]: loss = 0.28197208046913147\n",
      "[100/200, 5/7]: loss = 0.28631123900413513\n",
      "[100/200, 6/7]: loss = 0.2745245397090912\n",
      "[100/200, 7/7]: loss = 0.23570086061954498\n",
      "[101/200, 1/7]: loss = 0.28614673018455505\n",
      "[101/200, 2/7]: loss = 0.279592365026474\n",
      "[101/200, 3/7]: loss = 0.27996769547462463\n",
      "[101/200, 4/7]: loss = 0.2862517833709717\n",
      "[101/200, 5/7]: loss = 0.29400521516799927\n",
      "[101/200, 6/7]: loss = 0.28026577830314636\n",
      "[101/200, 7/7]: loss = 0.25634732842445374\n",
      "[102/200, 1/7]: loss = 0.27403947710990906\n",
      "[102/200, 2/7]: loss = 0.2852616608142853\n",
      "[102/200, 3/7]: loss = 0.2767794132232666\n",
      "[102/200, 4/7]: loss = 0.28196239471435547\n",
      "[102/200, 5/7]: loss = 0.28851255774497986\n",
      "[102/200, 6/7]: loss = 0.2895342707633972\n",
      "[102/200, 7/7]: loss = 0.28839704394340515\n",
      "[103/200, 1/7]: loss = 0.278438001871109\n",
      "[103/200, 2/7]: loss = 0.27931883931159973\n",
      "[103/200, 3/7]: loss = 0.2879628539085388\n",
      "[103/200, 4/7]: loss = 0.28552860021591187\n",
      "[103/200, 5/7]: loss = 0.28582921624183655\n",
      "[103/200, 6/7]: loss = 0.27633729577064514\n",
      "[103/200, 7/7]: loss = 0.31642529368400574\n",
      "[104/200, 1/7]: loss = 0.2831566035747528\n",
      "[104/200, 2/7]: loss = 0.2882251441478729\n",
      "[104/200, 3/7]: loss = 0.2781570255756378\n",
      "[104/200, 4/7]: loss = 0.2828834056854248\n",
      "[104/200, 5/7]: loss = 0.278502881526947\n",
      "[104/200, 6/7]: loss = 0.2865416407585144\n",
      "[104/200, 7/7]: loss = 0.28357818722724915\n",
      "[105/200, 1/7]: loss = 0.28330689668655396\n",
      "[105/200, 2/7]: loss = 0.29126474261283875\n",
      "[105/200, 3/7]: loss = 0.29302600026130676\n",
      "[105/200, 4/7]: loss = 0.28823044896125793\n",
      "[105/200, 5/7]: loss = 0.28163477778434753\n",
      "[105/200, 6/7]: loss = 0.2935471832752228\n",
      "[105/200, 7/7]: loss = 0.25287988781929016\n",
      "[106/200, 1/7]: loss = 0.27988743782043457\n",
      "[106/200, 2/7]: loss = 0.28610965609550476\n",
      "[106/200, 3/7]: loss = 0.2897875905036926\n",
      "[106/200, 4/7]: loss = 0.2846865952014923\n",
      "[106/200, 5/7]: loss = 0.28085461258888245\n",
      "[106/200, 6/7]: loss = 0.2949773967266083\n",
      "[106/200, 7/7]: loss = 0.3123137950897217\n",
      "[107/200, 1/7]: loss = 0.2851790487766266\n",
      "[107/200, 2/7]: loss = 0.28689467906951904\n",
      "[107/200, 3/7]: loss = 0.2746347188949585\n",
      "[107/200, 4/7]: loss = 0.28130874037742615\n",
      "[107/200, 5/7]: loss = 0.28438326716423035\n",
      "[107/200, 6/7]: loss = 0.2895447909832001\n",
      "[107/200, 7/7]: loss = 0.3167588710784912\n",
      "[108/200, 1/7]: loss = 0.2778235673904419\n",
      "[108/200, 2/7]: loss = 0.2898387014865875\n",
      "[108/200, 3/7]: loss = 0.28028011322021484\n",
      "[108/200, 4/7]: loss = 0.2879020571708679\n",
      "[108/200, 5/7]: loss = 0.28951945900917053\n",
      "[108/200, 6/7]: loss = 0.2922591269016266\n",
      "[108/200, 7/7]: loss = 0.2862534523010254\n",
      "[109/200, 1/7]: loss = 0.28767701983451843\n",
      "[109/200, 2/7]: loss = 0.27962687611579895\n",
      "[109/200, 3/7]: loss = 0.29197433590888977\n",
      "[109/200, 4/7]: loss = 0.2838667035102844\n",
      "[109/200, 5/7]: loss = 0.2857823073863983\n",
      "[109/200, 6/7]: loss = 0.2881287634372711\n",
      "[109/200, 7/7]: loss = 0.23394708335399628\n",
      "[110/200, 1/7]: loss = 0.2806819975376129\n",
      "[110/200, 2/7]: loss = 0.2844517230987549\n",
      "[110/200, 3/7]: loss = 0.28560134768486023\n",
      "[110/200, 4/7]: loss = 0.28994014859199524\n",
      "[110/200, 5/7]: loss = 0.2834734618663788\n",
      "[110/200, 6/7]: loss = 0.28389662504196167\n",
      "[110/200, 7/7]: loss = 0.24531768262386322\n",
      "[111/200, 1/7]: loss = 0.28140774369239807\n",
      "[111/200, 2/7]: loss = 0.28168442845344543\n",
      "[111/200, 3/7]: loss = 0.2853807508945465\n",
      "[111/200, 4/7]: loss = 0.28722697496414185\n",
      "[111/200, 5/7]: loss = 0.29232439398765564\n",
      "[111/200, 6/7]: loss = 0.2841097116470337\n",
      "[111/200, 7/7]: loss = 0.2741694450378418\n",
      "[112/200, 1/7]: loss = 0.2795947194099426\n",
      "[112/200, 2/7]: loss = 0.2811245620250702\n",
      "[112/200, 3/7]: loss = 0.2804502844810486\n",
      "[112/200, 4/7]: loss = 0.2848035395145416\n",
      "[112/200, 5/7]: loss = 0.28286072611808777\n",
      "[112/200, 6/7]: loss = 0.2794616222381592\n",
      "[112/200, 7/7]: loss = 0.2555137276649475\n",
      "[113/200, 1/7]: loss = 0.2811848521232605\n",
      "[113/200, 2/7]: loss = 0.2767009437084198\n",
      "[113/200, 3/7]: loss = 0.28769174218177795\n",
      "[113/200, 4/7]: loss = 0.2869396507740021\n",
      "[113/200, 5/7]: loss = 0.28267747163772583\n",
      "[113/200, 6/7]: loss = 0.2821839153766632\n",
      "[113/200, 7/7]: loss = 0.2961953282356262\n",
      "[114/200, 1/7]: loss = 0.28608888387680054\n",
      "[114/200, 2/7]: loss = 0.2829960882663727\n",
      "[114/200, 3/7]: loss = 0.28550612926483154\n",
      "[114/200, 4/7]: loss = 0.2863049805164337\n",
      "[114/200, 5/7]: loss = 0.2784552574157715\n",
      "[114/200, 6/7]: loss = 0.2861425280570984\n",
      "[114/200, 7/7]: loss = 0.2876492440700531\n",
      "[115/200, 1/7]: loss = 0.2737275958061218\n",
      "[115/200, 2/7]: loss = 0.28862884640693665\n",
      "[115/200, 3/7]: loss = 0.28960224986076355\n",
      "[115/200, 4/7]: loss = 0.2835431396961212\n",
      "[115/200, 5/7]: loss = 0.28257685899734497\n",
      "[115/200, 6/7]: loss = 0.2838198244571686\n",
      "[115/200, 7/7]: loss = 0.2922583222389221\n",
      "[116/200, 1/7]: loss = 0.2881128787994385\n",
      "[116/200, 2/7]: loss = 0.2819450795650482\n",
      "[116/200, 3/7]: loss = 0.28254255652427673\n",
      "[116/200, 4/7]: loss = 0.28795871138572693\n",
      "[116/200, 5/7]: loss = 0.2849256992340088\n",
      "[116/200, 6/7]: loss = 0.2835095226764679\n",
      "[116/200, 7/7]: loss = 0.3196246027946472\n",
      "[117/200, 1/7]: loss = 0.2785574495792389\n",
      "[117/200, 2/7]: loss = 0.29116615653038025\n",
      "[117/200, 3/7]: loss = 0.2806781828403473\n",
      "[117/200, 4/7]: loss = 0.2861572206020355\n",
      "[117/200, 5/7]: loss = 0.28977087140083313\n",
      "[117/200, 6/7]: loss = 0.28622856736183167\n",
      "[117/200, 7/7]: loss = 0.37762969732284546\n",
      "[118/200, 1/7]: loss = 0.2868407666683197\n",
      "[118/200, 2/7]: loss = 0.2927737534046173\n",
      "[118/200, 3/7]: loss = 0.28411775827407837\n",
      "[118/200, 4/7]: loss = 0.284614622592926\n",
      "[118/200, 5/7]: loss = 0.28374794125556946\n",
      "[118/200, 6/7]: loss = 0.292881578207016\n",
      "[118/200, 7/7]: loss = 0.28731098771095276\n",
      "[119/200, 1/7]: loss = 0.28406113386154175\n",
      "[119/200, 2/7]: loss = 0.2931702435016632\n",
      "[119/200, 3/7]: loss = 0.28756260871887207\n",
      "[119/200, 4/7]: loss = 0.2857491374015808\n",
      "[119/200, 5/7]: loss = 0.2815391421318054\n",
      "[119/200, 6/7]: loss = 0.28410422801971436\n",
      "[119/200, 7/7]: loss = 0.30410051345825195\n",
      "[120/200, 1/7]: loss = 0.2828492522239685\n",
      "[120/200, 2/7]: loss = 0.2730760872364044\n",
      "[120/200, 3/7]: loss = 0.2776390016078949\n",
      "[120/200, 4/7]: loss = 0.2944369614124298\n",
      "[120/200, 5/7]: loss = 0.28320571780204773\n",
      "[120/200, 6/7]: loss = 0.28459078073501587\n",
      "[120/200, 7/7]: loss = 0.3355048596858978\n",
      "[121/200, 1/7]: loss = 0.27573162317276\n",
      "[121/200, 2/7]: loss = 0.2737356126308441\n",
      "[121/200, 3/7]: loss = 0.28621459007263184\n",
      "[121/200, 4/7]: loss = 0.2883327007293701\n",
      "[121/200, 5/7]: loss = 0.2849638760089874\n",
      "[121/200, 6/7]: loss = 0.2868322432041168\n",
      "[121/200, 7/7]: loss = 0.2782977819442749\n",
      "[122/200, 1/7]: loss = 0.2759643495082855\n",
      "[122/200, 2/7]: loss = 0.2764088809490204\n",
      "[122/200, 3/7]: loss = 0.2890041172504425\n",
      "[122/200, 4/7]: loss = 0.28465738892555237\n",
      "[122/200, 5/7]: loss = 0.29089584946632385\n",
      "[122/200, 6/7]: loss = 0.282948762178421\n",
      "[122/200, 7/7]: loss = 0.2588231861591339\n",
      "[123/200, 1/7]: loss = 0.29122230410575867\n",
      "[123/200, 2/7]: loss = 0.2820156216621399\n",
      "[123/200, 3/7]: loss = 0.2824011743068695\n",
      "[123/200, 4/7]: loss = 0.2828855812549591\n",
      "[123/200, 5/7]: loss = 0.28586140275001526\n",
      "[123/200, 6/7]: loss = 0.29292941093444824\n",
      "[123/200, 7/7]: loss = 0.2633453905582428\n",
      "[124/200, 1/7]: loss = 0.27383899688720703\n",
      "[124/200, 2/7]: loss = 0.2834804356098175\n",
      "[124/200, 3/7]: loss = 0.2859758138656616\n",
      "[124/200, 4/7]: loss = 0.2895961105823517\n",
      "[124/200, 5/7]: loss = 0.2895979583263397\n",
      "[124/200, 6/7]: loss = 0.28266212344169617\n",
      "[124/200, 7/7]: loss = 0.21848835051059723\n",
      "[125/200, 1/7]: loss = 0.28038489818573\n",
      "[125/200, 2/7]: loss = 0.2827218472957611\n",
      "[125/200, 3/7]: loss = 0.29160091280937195\n",
      "[125/200, 4/7]: loss = 0.2794254422187805\n",
      "[125/200, 5/7]: loss = 0.2778044044971466\n",
      "[125/200, 6/7]: loss = 0.2771684229373932\n",
      "[125/200, 7/7]: loss = 0.28976282477378845\n",
      "[126/200, 1/7]: loss = 0.280013769865036\n",
      "[126/200, 2/7]: loss = 0.28354260325431824\n",
      "[126/200, 3/7]: loss = 0.2766720652580261\n",
      "[126/200, 4/7]: loss = 0.28097808361053467\n",
      "[126/200, 5/7]: loss = 0.28099969029426575\n",
      "[126/200, 6/7]: loss = 0.2843294143676758\n",
      "[126/200, 7/7]: loss = 0.2988879382610321\n",
      "[127/200, 1/7]: loss = 0.2734904885292053\n",
      "[127/200, 2/7]: loss = 0.2754567861557007\n",
      "[127/200, 3/7]: loss = 0.2858208417892456\n",
      "[127/200, 4/7]: loss = 0.28821539878845215\n",
      "[127/200, 5/7]: loss = 0.2884577214717865\n",
      "[127/200, 6/7]: loss = 0.28781697154045105\n",
      "[127/200, 7/7]: loss = 0.2672937512397766\n",
      "[128/200, 1/7]: loss = 0.281453400850296\n",
      "[128/200, 2/7]: loss = 0.28104451298713684\n",
      "[128/200, 3/7]: loss = 0.2840452492237091\n",
      "[128/200, 4/7]: loss = 0.28620368242263794\n",
      "[128/200, 5/7]: loss = 0.27743691205978394\n",
      "[128/200, 6/7]: loss = 0.2822185754776001\n",
      "[128/200, 7/7]: loss = 0.31151625514030457\n",
      "[129/200, 1/7]: loss = 0.27947089076042175\n",
      "[129/200, 2/7]: loss = 0.28153735399246216\n",
      "[129/200, 3/7]: loss = 0.2854230999946594\n",
      "[129/200, 4/7]: loss = 0.28806009888648987\n",
      "[129/200, 5/7]: loss = 0.2856537997722626\n",
      "[129/200, 6/7]: loss = 0.282679945230484\n",
      "[129/200, 7/7]: loss = 0.2557474672794342\n",
      "[130/200, 1/7]: loss = 0.2837928831577301\n",
      "[130/200, 2/7]: loss = 0.2917129397392273\n",
      "[130/200, 3/7]: loss = 0.2829599678516388\n",
      "[130/200, 4/7]: loss = 0.2871916890144348\n",
      "[130/200, 5/7]: loss = 0.28450772166252136\n",
      "[130/200, 6/7]: loss = 0.2898712456226349\n",
      "[130/200, 7/7]: loss = 0.30262887477874756\n",
      "[131/200, 1/7]: loss = 0.27858445048332214\n",
      "[131/200, 2/7]: loss = 0.2802223861217499\n",
      "[131/200, 3/7]: loss = 0.28717127442359924\n",
      "[131/200, 4/7]: loss = 0.2875373661518097\n",
      "[131/200, 5/7]: loss = 0.286725252866745\n",
      "[131/200, 6/7]: loss = 0.29031044244766235\n",
      "[131/200, 7/7]: loss = 0.27334651350975037\n",
      "[132/200, 1/7]: loss = 0.29036927223205566\n",
      "[132/200, 2/7]: loss = 0.2817700505256653\n",
      "[132/200, 3/7]: loss = 0.2767319977283478\n",
      "[132/200, 4/7]: loss = 0.28478243947029114\n",
      "[132/200, 5/7]: loss = 0.2814062237739563\n",
      "[132/200, 6/7]: loss = 0.2846520245075226\n",
      "[132/200, 7/7]: loss = 0.3030461370944977\n",
      "[133/200, 1/7]: loss = 0.2801949679851532\n",
      "[133/200, 2/7]: loss = 0.28581127524375916\n",
      "[133/200, 3/7]: loss = 0.2852981686592102\n",
      "[133/200, 4/7]: loss = 0.2888339161872864\n",
      "[133/200, 5/7]: loss = 0.2837207615375519\n",
      "[133/200, 6/7]: loss = 0.28344327211380005\n",
      "[133/200, 7/7]: loss = 0.25035548210144043\n",
      "[134/200, 1/7]: loss = 0.28771910071372986\n",
      "[134/200, 2/7]: loss = 0.29385289549827576\n",
      "[134/200, 3/7]: loss = 0.28034719824790955\n",
      "[134/200, 4/7]: loss = 0.2887735664844513\n",
      "[134/200, 5/7]: loss = 0.27592554688453674\n",
      "[134/200, 6/7]: loss = 0.28732532262802124\n",
      "[134/200, 7/7]: loss = 0.31250065565109253\n",
      "[135/200, 1/7]: loss = 0.2928129732608795\n",
      "[135/200, 2/7]: loss = 0.2767925262451172\n",
      "[135/200, 3/7]: loss = 0.2838003635406494\n",
      "[135/200, 4/7]: loss = 0.2725484371185303\n",
      "[135/200, 5/7]: loss = 0.28253406286239624\n",
      "[135/200, 6/7]: loss = 0.2743833065032959\n",
      "[135/200, 7/7]: loss = 0.2767019271850586\n",
      "[136/200, 1/7]: loss = 0.28731584548950195\n",
      "[136/200, 2/7]: loss = 0.2861137390136719\n",
      "[136/200, 3/7]: loss = 0.2816736102104187\n",
      "[136/200, 4/7]: loss = 0.280965119600296\n",
      "[136/200, 5/7]: loss = 0.27563410997390747\n",
      "[136/200, 6/7]: loss = 0.28006449341773987\n",
      "[136/200, 7/7]: loss = 0.2858607769012451\n",
      "[137/200, 1/7]: loss = 0.2805195152759552\n",
      "[137/200, 2/7]: loss = 0.27883464097976685\n",
      "[137/200, 3/7]: loss = 0.2847580909729004\n",
      "[137/200, 4/7]: loss = 0.2824443578720093\n",
      "[137/200, 5/7]: loss = 0.2776145040988922\n",
      "[137/200, 6/7]: loss = 0.29040053486824036\n",
      "[137/200, 7/7]: loss = 0.26792141795158386\n",
      "[138/200, 1/7]: loss = 0.28968489170074463\n",
      "[138/200, 2/7]: loss = 0.28228306770324707\n",
      "[138/200, 3/7]: loss = 0.286668598651886\n",
      "[138/200, 4/7]: loss = 0.2858105003833771\n",
      "[138/200, 5/7]: loss = 0.28199905157089233\n",
      "[138/200, 6/7]: loss = 0.28134235739707947\n",
      "[138/200, 7/7]: loss = 0.27689021825790405\n",
      "[139/200, 1/7]: loss = 0.2913990616798401\n",
      "[139/200, 2/7]: loss = 0.2852807343006134\n",
      "[139/200, 3/7]: loss = 0.2842276096343994\n",
      "[139/200, 4/7]: loss = 0.2841120660305023\n",
      "[139/200, 5/7]: loss = 0.27902859449386597\n",
      "[139/200, 6/7]: loss = 0.26982828974723816\n",
      "[139/200, 7/7]: loss = 0.2980978190898895\n",
      "[140/200, 1/7]: loss = 0.28456249833106995\n",
      "[140/200, 2/7]: loss = 0.2755332291126251\n",
      "[140/200, 3/7]: loss = 0.28518733382225037\n",
      "[140/200, 4/7]: loss = 0.290522962808609\n",
      "[140/200, 5/7]: loss = 0.2818317413330078\n",
      "[140/200, 6/7]: loss = 0.28659388422966003\n",
      "[140/200, 7/7]: loss = 0.2708896994590759\n",
      "[141/200, 1/7]: loss = 0.2879008650779724\n",
      "[141/200, 2/7]: loss = 0.28225985169410706\n",
      "[141/200, 3/7]: loss = 0.2790200710296631\n",
      "[141/200, 4/7]: loss = 0.27998852729797363\n",
      "[141/200, 5/7]: loss = 0.28262093663215637\n",
      "[141/200, 6/7]: loss = 0.2822328507900238\n",
      "[141/200, 7/7]: loss = 0.2942374646663666\n",
      "[142/200, 1/7]: loss = 0.2794301509857178\n",
      "[142/200, 2/7]: loss = 0.2822101414203644\n",
      "[142/200, 3/7]: loss = 0.2886134088039398\n",
      "[142/200, 4/7]: loss = 0.2815582752227783\n",
      "[142/200, 5/7]: loss = 0.28295910358428955\n",
      "[142/200, 6/7]: loss = 0.2830064296722412\n",
      "[142/200, 7/7]: loss = 0.3199167251586914\n",
      "[143/200, 1/7]: loss = 0.2837771475315094\n",
      "[143/200, 2/7]: loss = 0.2888251543045044\n",
      "[143/200, 3/7]: loss = 0.28046301007270813\n",
      "[143/200, 4/7]: loss = 0.2773968577384949\n",
      "[143/200, 5/7]: loss = 0.2817908525466919\n",
      "[143/200, 6/7]: loss = 0.29221785068511963\n",
      "[143/200, 7/7]: loss = 0.2971407175064087\n",
      "[144/200, 1/7]: loss = 0.2762271761894226\n",
      "[144/200, 2/7]: loss = 0.2789852023124695\n",
      "[144/200, 3/7]: loss = 0.282654345035553\n",
      "[144/200, 4/7]: loss = 0.28402531147003174\n",
      "[144/200, 5/7]: loss = 0.2916422486305237\n",
      "[144/200, 6/7]: loss = 0.2895849347114563\n",
      "[144/200, 7/7]: loss = 0.304186075925827\n",
      "[145/200, 1/7]: loss = 0.28008630871772766\n",
      "[145/200, 2/7]: loss = 0.2790219783782959\n",
      "[145/200, 3/7]: loss = 0.2751234173774719\n",
      "[145/200, 4/7]: loss = 0.3013370633125305\n",
      "[145/200, 5/7]: loss = 0.28033649921417236\n",
      "[145/200, 6/7]: loss = 0.27712327241897583\n",
      "[145/200, 7/7]: loss = 0.2799445390701294\n",
      "[146/200, 1/7]: loss = 0.27451464533805847\n",
      "[146/200, 2/7]: loss = 0.29054588079452515\n",
      "[146/200, 3/7]: loss = 0.27632564306259155\n",
      "[146/200, 4/7]: loss = 0.2819638252258301\n",
      "[146/200, 5/7]: loss = 0.27860328555107117\n",
      "[146/200, 6/7]: loss = 0.28024914860725403\n",
      "[146/200, 7/7]: loss = 0.2824822962284088\n",
      "[147/200, 1/7]: loss = 0.2765101492404938\n",
      "[147/200, 2/7]: loss = 0.27715399861335754\n",
      "[147/200, 3/7]: loss = 0.28710508346557617\n",
      "[147/200, 4/7]: loss = 0.2797321081161499\n",
      "[147/200, 5/7]: loss = 0.28269869089126587\n",
      "[147/200, 6/7]: loss = 0.280892014503479\n",
      "[147/200, 7/7]: loss = 0.29491499066352844\n",
      "[148/200, 1/7]: loss = 0.28189799189567566\n",
      "[148/200, 2/7]: loss = 0.2869418263435364\n",
      "[148/200, 3/7]: loss = 0.29303908348083496\n",
      "[148/200, 4/7]: loss = 0.2742573022842407\n",
      "[148/200, 5/7]: loss = 0.2851833403110504\n",
      "[148/200, 6/7]: loss = 0.285849392414093\n",
      "[148/200, 7/7]: loss = 0.3096117079257965\n",
      "[149/200, 1/7]: loss = 0.28258076310157776\n",
      "[149/200, 2/7]: loss = 0.27665185928344727\n",
      "[149/200, 3/7]: loss = 0.28505077958106995\n",
      "[149/200, 4/7]: loss = 0.28213807940483093\n",
      "[149/200, 5/7]: loss = 0.28629961609840393\n",
      "[149/200, 6/7]: loss = 0.2787556052207947\n",
      "[149/200, 7/7]: loss = 0.3233048915863037\n",
      "[150/200, 1/7]: loss = 0.2866493761539459\n",
      "[150/200, 2/7]: loss = 0.2789551019668579\n",
      "[150/200, 3/7]: loss = 0.28808093070983887\n",
      "[150/200, 4/7]: loss = 0.29463881254196167\n",
      "[150/200, 5/7]: loss = 0.2811011075973511\n",
      "[150/200, 6/7]: loss = 0.28254011273384094\n",
      "[150/200, 7/7]: loss = 0.24429240822792053\n",
      "[151/200, 1/7]: loss = 0.28188392519950867\n",
      "[151/200, 2/7]: loss = 0.2832511365413666\n",
      "[151/200, 3/7]: loss = 0.2846131920814514\n",
      "[151/200, 4/7]: loss = 0.2826339900493622\n",
      "[151/200, 5/7]: loss = 0.28909072279930115\n",
      "[151/200, 6/7]: loss = 0.28468653559684753\n",
      "[151/200, 7/7]: loss = 0.3505628705024719\n",
      "[152/200, 1/7]: loss = 0.2807798385620117\n",
      "[152/200, 2/7]: loss = 0.28401055932044983\n",
      "[152/200, 3/7]: loss = 0.2879849970340729\n",
      "[152/200, 4/7]: loss = 0.2824101746082306\n",
      "[152/200, 5/7]: loss = 0.2920604348182678\n",
      "[152/200, 6/7]: loss = 0.2896701693534851\n",
      "[152/200, 7/7]: loss = 0.2834649085998535\n",
      "[153/200, 1/7]: loss = 0.28293442726135254\n",
      "[153/200, 2/7]: loss = 0.2842908203601837\n",
      "[153/200, 3/7]: loss = 0.2832252085208893\n",
      "[153/200, 4/7]: loss = 0.28452053666114807\n",
      "[153/200, 5/7]: loss = 0.29006457328796387\n",
      "[153/200, 6/7]: loss = 0.28110605478286743\n",
      "[153/200, 7/7]: loss = 0.39899271726608276\n",
      "[154/200, 1/7]: loss = 0.29241272807121277\n",
      "[154/200, 2/7]: loss = 0.2948780655860901\n",
      "[154/200, 3/7]: loss = 0.27923744916915894\n",
      "[154/200, 4/7]: loss = 0.2941492199897766\n",
      "[154/200, 5/7]: loss = 0.3188497722148895\n",
      "[154/200, 6/7]: loss = 0.3175000846385956\n",
      "[154/200, 7/7]: loss = 0.2989211082458496\n",
      "[155/200, 1/7]: loss = 0.30731838941574097\n",
      "[155/200, 2/7]: loss = 0.3061521351337433\n",
      "[155/200, 3/7]: loss = 0.30718210339546204\n",
      "[155/200, 4/7]: loss = 0.3239234387874603\n",
      "[155/200, 5/7]: loss = 0.3061457872390747\n",
      "[155/200, 6/7]: loss = 0.3196708858013153\n",
      "[155/200, 7/7]: loss = 0.36554667353630066\n",
      "[156/200, 1/7]: loss = 0.3174577057361603\n",
      "[156/200, 2/7]: loss = 0.3253805637359619\n",
      "[156/200, 3/7]: loss = 0.30822592973709106\n",
      "[156/200, 4/7]: loss = 0.3134216368198395\n",
      "[156/200, 5/7]: loss = 0.314081609249115\n",
      "[156/200, 6/7]: loss = 0.3336896598339081\n",
      "[156/200, 7/7]: loss = 0.378193736076355\n",
      "[157/200, 1/7]: loss = 0.3073621690273285\n",
      "[157/200, 2/7]: loss = 0.3162604868412018\n",
      "[157/200, 3/7]: loss = 0.33136987686157227\n",
      "[157/200, 4/7]: loss = 0.31315234303474426\n",
      "[157/200, 5/7]: loss = 0.31406712532043457\n",
      "[157/200, 6/7]: loss = 0.31743475794792175\n",
      "[157/200, 7/7]: loss = 0.30954575538635254\n",
      "[158/200, 1/7]: loss = 0.2993601858615875\n",
      "[158/200, 2/7]: loss = 0.30444759130477905\n",
      "[158/200, 3/7]: loss = 0.30297791957855225\n",
      "[158/200, 4/7]: loss = 0.299183189868927\n",
      "[158/200, 5/7]: loss = 0.3105289936065674\n",
      "[158/200, 6/7]: loss = 0.3075394630432129\n",
      "[158/200, 7/7]: loss = 0.25949379801750183\n",
      "[159/200, 1/7]: loss = 0.3064097464084625\n",
      "[159/200, 2/7]: loss = 0.30539852380752563\n",
      "[159/200, 3/7]: loss = 0.3011750280857086\n",
      "[159/200, 4/7]: loss = 0.3000037670135498\n",
      "[159/200, 5/7]: loss = 0.30442020297050476\n",
      "[159/200, 6/7]: loss = 0.2931259274482727\n",
      "[159/200, 7/7]: loss = 0.34536099433898926\n",
      "[160/200, 1/7]: loss = 0.28374505043029785\n",
      "[160/200, 2/7]: loss = 0.297118216753006\n",
      "[160/200, 3/7]: loss = 0.30097711086273193\n",
      "[160/200, 4/7]: loss = 0.29799380898475647\n",
      "[160/200, 5/7]: loss = 0.2888948619365692\n",
      "[160/200, 6/7]: loss = 0.2980521619319916\n",
      "[160/200, 7/7]: loss = 0.3301486670970917\n",
      "[161/200, 1/7]: loss = 0.28990471363067627\n",
      "[161/200, 2/7]: loss = 0.29264527559280396\n",
      "[161/200, 3/7]: loss = 0.298015296459198\n",
      "[161/200, 4/7]: loss = 0.30250492691993713\n",
      "[161/200, 5/7]: loss = 0.2959332764148712\n",
      "[161/200, 6/7]: loss = 0.2890093922615051\n",
      "[161/200, 7/7]: loss = 0.2599598169326782\n",
      "[162/200, 1/7]: loss = 0.2876355051994324\n",
      "[162/200, 2/7]: loss = 0.2927296459674835\n",
      "[162/200, 3/7]: loss = 0.30525046586990356\n",
      "[162/200, 4/7]: loss = 0.2815665900707245\n",
      "[162/200, 5/7]: loss = 0.2925427258014679\n",
      "[162/200, 6/7]: loss = 0.29196012020111084\n",
      "[162/200, 7/7]: loss = 0.28571105003356934\n",
      "[163/200, 1/7]: loss = 0.29615575075149536\n",
      "[163/200, 2/7]: loss = 0.2974858582019806\n",
      "[163/200, 3/7]: loss = 0.29102590680122375\n",
      "[163/200, 4/7]: loss = 0.2868931293487549\n",
      "[163/200, 5/7]: loss = 0.2876937687397003\n",
      "[163/200, 6/7]: loss = 0.2938101887702942\n",
      "[163/200, 7/7]: loss = 0.33459657430648804\n",
      "[164/200, 1/7]: loss = 0.2892738878726959\n",
      "[164/200, 2/7]: loss = 0.287206768989563\n",
      "[164/200, 3/7]: loss = 0.29001525044441223\n",
      "[164/200, 4/7]: loss = 0.2906855642795563\n",
      "[164/200, 5/7]: loss = 0.27575013041496277\n",
      "[164/200, 6/7]: loss = 0.301926851272583\n",
      "[164/200, 7/7]: loss = 0.3044503927230835\n",
      "[165/200, 1/7]: loss = 0.2872236669063568\n",
      "[165/200, 2/7]: loss = 0.28638824820518494\n",
      "[165/200, 3/7]: loss = 0.28523722290992737\n",
      "[165/200, 4/7]: loss = 0.2849854528903961\n",
      "[165/200, 5/7]: loss = 0.2892565429210663\n",
      "[165/200, 6/7]: loss = 0.28265637159347534\n",
      "[165/200, 7/7]: loss = 0.3143601715564728\n",
      "[166/200, 1/7]: loss = 0.29009392857551575\n",
      "[166/200, 2/7]: loss = 0.28803932666778564\n",
      "[166/200, 3/7]: loss = 0.28525862097740173\n",
      "[166/200, 4/7]: loss = 0.2852749526500702\n",
      "[166/200, 5/7]: loss = 0.28564053773880005\n",
      "[166/200, 6/7]: loss = 0.2899996042251587\n",
      "[166/200, 7/7]: loss = 0.3123525083065033\n",
      "[167/200, 1/7]: loss = 0.29252326488494873\n",
      "[167/200, 2/7]: loss = 0.2809391915798187\n",
      "[167/200, 3/7]: loss = 0.2876410186290741\n",
      "[167/200, 4/7]: loss = 0.29112550616264343\n",
      "[167/200, 5/7]: loss = 0.29124462604522705\n",
      "[167/200, 6/7]: loss = 0.29122963547706604\n",
      "[167/200, 7/7]: loss = 0.2490943819284439\n",
      "[168/200, 1/7]: loss = 0.2839636206626892\n",
      "[168/200, 2/7]: loss = 0.2907554507255554\n",
      "[168/200, 3/7]: loss = 0.2805769443511963\n",
      "[168/200, 4/7]: loss = 0.2982960045337677\n",
      "[168/200, 5/7]: loss = 0.28482741117477417\n",
      "[168/200, 6/7]: loss = 0.2903216481208801\n",
      "[168/200, 7/7]: loss = 0.3065668046474457\n",
      "[169/200, 1/7]: loss = 0.29089611768722534\n",
      "[169/200, 2/7]: loss = 0.28767144680023193\n",
      "[169/200, 3/7]: loss = 0.27888748049736023\n",
      "[169/200, 4/7]: loss = 0.29226213693618774\n",
      "[169/200, 5/7]: loss = 0.2759360074996948\n",
      "[169/200, 6/7]: loss = 0.2920279800891876\n",
      "[169/200, 7/7]: loss = 0.2786109447479248\n",
      "[170/200, 1/7]: loss = 0.2837619483470917\n",
      "[170/200, 2/7]: loss = 0.29030874371528625\n",
      "[170/200, 3/7]: loss = 0.28782641887664795\n",
      "[170/200, 4/7]: loss = 0.28154221177101135\n",
      "[170/200, 5/7]: loss = 0.27928346395492554\n",
      "[170/200, 6/7]: loss = 0.2902012765407562\n",
      "[170/200, 7/7]: loss = 0.2590485215187073\n",
      "[171/200, 1/7]: loss = 0.2838590145111084\n",
      "[171/200, 2/7]: loss = 0.2910735607147217\n",
      "[171/200, 3/7]: loss = 0.2803502380847931\n",
      "[171/200, 4/7]: loss = 0.2896418869495392\n",
      "[171/200, 5/7]: loss = 0.2852085530757904\n",
      "[171/200, 6/7]: loss = 0.2845487892627716\n",
      "[171/200, 7/7]: loss = 0.24864718317985535\n",
      "[172/200, 1/7]: loss = 0.28178009390830994\n",
      "[172/200, 2/7]: loss = 0.28390389680862427\n",
      "[172/200, 3/7]: loss = 0.27849313616752625\n",
      "[172/200, 4/7]: loss = 0.2860809862613678\n",
      "[172/200, 5/7]: loss = 0.29129284620285034\n",
      "[172/200, 6/7]: loss = 0.2838538587093353\n",
      "[172/200, 7/7]: loss = 0.31954243779182434\n",
      "[173/200, 1/7]: loss = 0.2814568281173706\n",
      "[173/200, 2/7]: loss = 0.2785528898239136\n",
      "[173/200, 3/7]: loss = 0.28039202094078064\n",
      "[173/200, 4/7]: loss = 0.2886272668838501\n",
      "[173/200, 5/7]: loss = 0.2907724678516388\n",
      "[173/200, 6/7]: loss = 0.29622411727905273\n",
      "[173/200, 7/7]: loss = 0.2803453505039215\n",
      "[174/200, 1/7]: loss = 0.2906941771507263\n",
      "[174/200, 2/7]: loss = 0.28398004174232483\n",
      "[174/200, 3/7]: loss = 0.2944405674934387\n",
      "[174/200, 4/7]: loss = 0.28702229261398315\n",
      "[174/200, 5/7]: loss = 0.2844060957431793\n",
      "[174/200, 6/7]: loss = 0.2947588264942169\n",
      "[174/200, 7/7]: loss = 0.25152233242988586\n",
      "[175/200, 1/7]: loss = 0.28725531697273254\n",
      "[175/200, 2/7]: loss = 0.28544312715530396\n",
      "[175/200, 3/7]: loss = 0.28686314821243286\n",
      "[175/200, 4/7]: loss = 0.2946174442768097\n",
      "[175/200, 5/7]: loss = 0.2846527397632599\n",
      "[175/200, 6/7]: loss = 0.2885937988758087\n",
      "[175/200, 7/7]: loss = 0.31198757886886597\n",
      "[176/200, 1/7]: loss = 0.2826121747493744\n",
      "[176/200, 2/7]: loss = 0.2741035521030426\n",
      "[176/200, 3/7]: loss = 0.28703489899635315\n",
      "[176/200, 4/7]: loss = 0.29205602407455444\n",
      "[176/200, 5/7]: loss = 0.29267939925193787\n",
      "[176/200, 6/7]: loss = 0.2908368408679962\n",
      "[176/200, 7/7]: loss = 0.30026063323020935\n",
      "[177/200, 1/7]: loss = 0.27942517399787903\n",
      "[177/200, 2/7]: loss = 0.2754231095314026\n",
      "[177/200, 3/7]: loss = 0.2855653464794159\n",
      "[177/200, 4/7]: loss = 0.2888182997703552\n",
      "[177/200, 5/7]: loss = 0.28482189774513245\n",
      "[177/200, 6/7]: loss = 0.29008644819259644\n",
      "[177/200, 7/7]: loss = 0.332956999540329\n",
      "[178/200, 1/7]: loss = 0.28695613145828247\n",
      "[178/200, 2/7]: loss = 0.29392004013061523\n",
      "[178/200, 3/7]: loss = 0.2758006751537323\n",
      "[178/200, 4/7]: loss = 0.28393760323524475\n",
      "[178/200, 5/7]: loss = 0.2862169146537781\n",
      "[178/200, 6/7]: loss = 0.28961485624313354\n",
      "[178/200, 7/7]: loss = 0.3004510998725891\n",
      "[179/200, 1/7]: loss = 0.2915865182876587\n",
      "[179/200, 2/7]: loss = 0.2855300307273865\n",
      "[179/200, 3/7]: loss = 0.2901557683944702\n",
      "[179/200, 4/7]: loss = 0.2900764048099518\n",
      "[179/200, 5/7]: loss = 0.29018598794937134\n",
      "[179/200, 6/7]: loss = 0.289180189371109\n",
      "[179/200, 7/7]: loss = 0.30176860094070435\n",
      "[180/200, 1/7]: loss = 0.28316059708595276\n",
      "[180/200, 2/7]: loss = 0.29350030422210693\n",
      "[180/200, 3/7]: loss = 0.2810850739479065\n",
      "[180/200, 4/7]: loss = 0.29428204894065857\n",
      "[180/200, 5/7]: loss = 0.2891766428947449\n",
      "[180/200, 6/7]: loss = 0.2862153947353363\n",
      "[180/200, 7/7]: loss = 0.2987235486507416\n",
      "[181/200, 1/7]: loss = 0.29053381085395813\n",
      "[181/200, 2/7]: loss = 0.2837922275066376\n",
      "[181/200, 3/7]: loss = 0.2796485126018524\n",
      "[181/200, 4/7]: loss = 0.28995010256767273\n",
      "[181/200, 5/7]: loss = 0.2876891791820526\n",
      "[181/200, 6/7]: loss = 0.2862201929092407\n",
      "[181/200, 7/7]: loss = 0.2662951648235321\n",
      "[182/200, 1/7]: loss = 0.28020042181015015\n",
      "[182/200, 2/7]: loss = 0.2975452244281769\n",
      "[182/200, 3/7]: loss = 0.2802676260471344\n",
      "[182/200, 4/7]: loss = 0.28793734312057495\n",
      "[182/200, 5/7]: loss = 0.292758971452713\n",
      "[182/200, 6/7]: loss = 0.28643494844436646\n",
      "[182/200, 7/7]: loss = 0.2988649904727936\n",
      "[183/200, 1/7]: loss = 0.27180159091949463\n",
      "[183/200, 2/7]: loss = 0.27304235100746155\n",
      "[183/200, 3/7]: loss = 0.2920997142791748\n",
      "[183/200, 4/7]: loss = 0.28160983324050903\n",
      "[183/200, 5/7]: loss = 0.2800716459751129\n",
      "[183/200, 6/7]: loss = 0.2949668765068054\n",
      "[183/200, 7/7]: loss = 0.24325110018253326\n",
      "[184/200, 1/7]: loss = 0.28849196434020996\n",
      "[184/200, 2/7]: loss = 0.28748074173927307\n",
      "[184/200, 3/7]: loss = 0.2805558741092682\n",
      "[184/200, 4/7]: loss = 0.28877151012420654\n",
      "[184/200, 5/7]: loss = 0.2748959958553314\n",
      "[184/200, 6/7]: loss = 0.2885635197162628\n",
      "[184/200, 7/7]: loss = 0.30580779910087585\n",
      "[185/200, 1/7]: loss = 0.28759536147117615\n",
      "[185/200, 2/7]: loss = 0.28364136815071106\n",
      "[185/200, 3/7]: loss = 0.2955780327320099\n",
      "[185/200, 4/7]: loss = 0.2848075032234192\n",
      "[185/200, 5/7]: loss = 0.2889135479927063\n",
      "[185/200, 6/7]: loss = 0.29302701354026794\n",
      "[185/200, 7/7]: loss = 0.2958330512046814\n",
      "[186/200, 1/7]: loss = 0.29107388854026794\n",
      "[186/200, 2/7]: loss = 0.27768585085868835\n",
      "[186/200, 3/7]: loss = 0.28620603680610657\n",
      "[186/200, 4/7]: loss = 0.29367563128471375\n",
      "[186/200, 5/7]: loss = 0.28417056798934937\n",
      "[186/200, 6/7]: loss = 0.28723040223121643\n",
      "[186/200, 7/7]: loss = 0.2917860448360443\n",
      "[187/200, 1/7]: loss = 0.28093767166137695\n",
      "[187/200, 2/7]: loss = 0.28138336539268494\n",
      "[187/200, 3/7]: loss = 0.2830304205417633\n",
      "[187/200, 4/7]: loss = 0.28194019198417664\n",
      "[187/200, 5/7]: loss = 0.2840192914009094\n",
      "[187/200, 6/7]: loss = 0.2827359735965729\n",
      "[187/200, 7/7]: loss = 0.2572265565395355\n",
      "[188/200, 1/7]: loss = 0.27950766682624817\n",
      "[188/200, 2/7]: loss = 0.28733524680137634\n",
      "[188/200, 3/7]: loss = 0.2816430628299713\n",
      "[188/200, 4/7]: loss = 0.28510230779647827\n",
      "[188/200, 5/7]: loss = 0.2774231433868408\n",
      "[188/200, 6/7]: loss = 0.2779785990715027\n",
      "[188/200, 7/7]: loss = 0.2631497383117676\n",
      "[189/200, 1/7]: loss = 0.27800148725509644\n",
      "[189/200, 2/7]: loss = 0.2849808633327484\n",
      "[189/200, 3/7]: loss = 0.27732935547828674\n",
      "[189/200, 4/7]: loss = 0.27942973375320435\n",
      "[189/200, 5/7]: loss = 0.29197177290916443\n",
      "[189/200, 6/7]: loss = 0.28309088945388794\n",
      "[189/200, 7/7]: loss = 0.26352304220199585\n",
      "[190/200, 1/7]: loss = 0.2826440930366516\n",
      "[190/200, 2/7]: loss = 0.2794111371040344\n",
      "[190/200, 3/7]: loss = 0.27829304337501526\n",
      "[190/200, 4/7]: loss = 0.2924233675003052\n",
      "[190/200, 5/7]: loss = 0.29517796635627747\n",
      "[190/200, 6/7]: loss = 0.28096917271614075\n",
      "[190/200, 7/7]: loss = 0.2888679802417755\n",
      "[191/200, 1/7]: loss = 0.2819347679615021\n",
      "[191/200, 2/7]: loss = 0.2845418155193329\n",
      "[191/200, 3/7]: loss = 0.2940773665904999\n",
      "[191/200, 4/7]: loss = 0.27608776092529297\n",
      "[191/200, 5/7]: loss = 0.28733155131340027\n",
      "[191/200, 6/7]: loss = 0.2804543673992157\n",
      "[191/200, 7/7]: loss = 0.2756441533565521\n",
      "[192/200, 1/7]: loss = 0.2808727025985718\n",
      "[192/200, 2/7]: loss = 0.2793291211128235\n",
      "[192/200, 3/7]: loss = 0.28728237748146057\n",
      "[192/200, 4/7]: loss = 0.2877114713191986\n",
      "[192/200, 5/7]: loss = 0.28602519631385803\n",
      "[192/200, 6/7]: loss = 0.2879793047904968\n",
      "[192/200, 7/7]: loss = 0.2990308403968811\n",
      "[193/200, 1/7]: loss = 0.27744078636169434\n",
      "[193/200, 2/7]: loss = 0.28048765659332275\n",
      "[193/200, 3/7]: loss = 0.28472426533699036\n",
      "[193/200, 4/7]: loss = 0.2802107334136963\n",
      "[193/200, 5/7]: loss = 0.2795605957508087\n",
      "[193/200, 6/7]: loss = 0.2837926745414734\n",
      "[193/200, 7/7]: loss = 0.3250967860221863\n",
      "[194/200, 1/7]: loss = 0.27599698305130005\n",
      "[194/200, 2/7]: loss = 0.28270378708839417\n",
      "[194/200, 3/7]: loss = 0.28154098987579346\n",
      "[194/200, 4/7]: loss = 0.296318918466568\n",
      "[194/200, 5/7]: loss = 0.2804073393344879\n",
      "[194/200, 6/7]: loss = 0.287908136844635\n",
      "[194/200, 7/7]: loss = 0.3117397427558899\n",
      "[195/200, 1/7]: loss = 0.2871403992176056\n",
      "[195/200, 2/7]: loss = 0.28470614552497864\n",
      "[195/200, 3/7]: loss = 0.28569620847702026\n",
      "[195/200, 4/7]: loss = 0.28369036316871643\n",
      "[195/200, 5/7]: loss = 0.2828100621700287\n",
      "[195/200, 6/7]: loss = 0.28116488456726074\n",
      "[195/200, 7/7]: loss = 0.24229422211647034\n",
      "[196/200, 1/7]: loss = 0.2764851450920105\n",
      "[196/200, 2/7]: loss = 0.2902935743331909\n",
      "[196/200, 3/7]: loss = 0.28125932812690735\n",
      "[196/200, 4/7]: loss = 0.2738141417503357\n",
      "[196/200, 5/7]: loss = 0.2815571129322052\n",
      "[196/200, 6/7]: loss = 0.2873632311820984\n",
      "[196/200, 7/7]: loss = 0.35191482305526733\n",
      "[197/200, 1/7]: loss = 0.2857449948787689\n",
      "[197/200, 2/7]: loss = 0.2732866406440735\n",
      "[197/200, 3/7]: loss = 0.2829159200191498\n",
      "[197/200, 4/7]: loss = 0.28223758935928345\n",
      "[197/200, 5/7]: loss = 0.287765234708786\n",
      "[197/200, 6/7]: loss = 0.2864774763584137\n",
      "[197/200, 7/7]: loss = 0.2971329689025879\n",
      "[198/200, 1/7]: loss = 0.2750469148159027\n",
      "[198/200, 2/7]: loss = 0.2769608795642853\n",
      "[198/200, 3/7]: loss = 0.2874253988265991\n",
      "[198/200, 4/7]: loss = 0.2803911566734314\n",
      "[198/200, 5/7]: loss = 0.27371716499328613\n",
      "[198/200, 6/7]: loss = 0.28646788001060486\n",
      "[198/200, 7/7]: loss = 0.2716742157936096\n",
      "[199/200, 1/7]: loss = 0.2823635935783386\n",
      "[199/200, 2/7]: loss = 0.2817482054233551\n",
      "[199/200, 3/7]: loss = 0.27842646837234497\n",
      "[199/200, 4/7]: loss = 0.29113098978996277\n",
      "[199/200, 5/7]: loss = 0.28031453490257263\n",
      "[199/200, 6/7]: loss = 0.28086134791374207\n",
      "[199/200, 7/7]: loss = 0.30052492022514343\n",
      "[200/200, 1/7]: loss = 0.2847726345062256\n",
      "[200/200, 2/7]: loss = 0.27838757634162903\n",
      "[200/200, 3/7]: loss = 0.27659639716148376\n",
      "[200/200, 4/7]: loss = 0.28025349974632263\n",
      "[200/200, 5/7]: loss = 0.27972832322120667\n",
      "[200/200, 6/7]: loss = 0.2830356955528259\n",
      "[200/200, 7/7]: loss = 0.25204992294311523\n"
     ]
    }
   ],
   "source": [
    "model.train()\n",
    "\n",
    "losses = []\n",
    "\n",
    "for epoch in range(n_epochs):\n",
    "    iteration_loss = 0\n",
    "    for batch, (X, y) in enumerate(dataloader):\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        # forward propagation\n",
    "        h_previous = torch.zeros(n_layers, X.size(0), hidden_size)\n",
    "        c_previous = torch.zeros(n_layers, X.size(0), hidden_size)\n",
    "\n",
    "        y_hat, state_current = model(X, (h_previous, c_previous))\n",
    "\n",
    "        # loss calculation\n",
    "        loss = criterion(y_hat.transpose(1, 2), y)\n",
    "        iteration_loss += loss.item()\n",
    "\n",
    "        # backward propagation\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        print(f\"[{epoch + 1}/{n_epochs}, {batch + 1}/{len(dataloader)}]: loss = {loss.item()}\")\n",
    "        \n",
    "    iteration_loss /= len(dataloader)\n",
    "    losses.append(iteration_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(len(losses)), losses)\n",
    "plt.xlabel(\"iteration\")\n",
    "plt.ylabel(\"loss\")\n",
    "plt.title(\"loss wrt iteration\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Saving the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(model.state_dict(), \"v2.pt\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Model(vocabulary_size=len(char2idx), embedding_dimensions=embedding_dimensions, hidden_size=hidden_size, n_layers=n_layers, keep_prob=keep_prob)\n",
    "model.load_state_dict(torch.load(\"main.pt\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sampling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sampling using a random initial character."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "THE MODEL HAVE GENERATED THE NAME \"walkersaurus\", WHICH IS A **real dinosaur** NAME\n"
     ]
    }
   ],
   "source": [
    "# model.eval()\n",
    "characters = []\n",
    "character_idx = torch.randint(low=1, high=len(vocabulary), size=(1, 1))\n",
    "characters.append(character_idx.squeeze().item())\n",
    "h_previous = torch.zeros(n_layers, character_idx.size(0), hidden_size)\n",
    "c_previous = torch.zeros(n_layers, character_idx.size(0), hidden_size)\n",
    "\n",
    "with torch.no_grad():\n",
    "    while character_idx.squeeze().item() != char2idx[\"EOS\"]:\n",
    "        y_hat, state_current = model(character_idx, (h_previous, c_previous))\n",
    "        softmax_y_hat = F.softmax(y_hat, dim=-1)\n",
    "        character_idx = torch.argmax(softmax_y_hat, dim=-1)\n",
    "        characters.append(character_idx.squeeze().item())\n",
    "\n",
    "        h_previous, c_previous = state_current\n",
    "\n",
    "\n",
    "name = \"\".join([idx2char[character] for character in characters if character != 0])\n",
    "autenticity = \"real dinosaur\" if name in names else \"made up\"\n",
    "print(f\"THE MODEL HAVE GENERATED THE NAME \\\"{name}\\\", WHICH IS A **{autenticity}** NAME\")\n",
    "if name not in names:\n",
    "    with open(\"gen_names.txt\", \"a+\") as file:\n",
    "        file.seek(0)\n",
    "        file_names = file.read().splitlines()\n",
    "        if name not in file_names:\n",
    "            file.seek(0, 2)\n",
    "            file.write(f\"{name}\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sampling using a pre-determined initial character."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "THE MODEL HAVE GENERATED THE NAME \"quillesaurus\", WHICH IS A **made up** NAME\n"
     ]
    }
   ],
   "source": [
    "# model.eval()\n",
    "characters = []\n",
    "initial_character = input(\">> input an initial dinosaur name character: \")\n",
    "character_idx = torch.tensor([[char2idx[initial_character]]])\n",
    "characters.append(character_idx.squeeze().item())\n",
    "h_previous = torch.zeros(n_layers, character_idx.size(0), hidden_size)\n",
    "c_previous = torch.zeros(n_layers, character_idx.size(0), hidden_size)\n",
    "\n",
    "with torch.no_grad():\n",
    "    while character_idx.squeeze().item() != char2idx[\"EOS\"]:\n",
    "        y_hat, state_current = model(character_idx, (h_previous, c_previous))\n",
    "        softmax_y_hat = F.softmax(y_hat, dim=-1)\n",
    "        character_idx = torch.argmax(softmax_y_hat, dim=-1)\n",
    "        characters.append(character_idx.squeeze().item())\n",
    "\n",
    "        h_previous, c_previous = state_current\n",
    "\n",
    "name = \"\".join([idx2char[character] for character in characters if character != 0])\n",
    "autenticity = \"real dinosaur\" if name in names else \"made up\"\n",
    "print(f\"THE MODEL HAVE GENERATED THE NAME \\\"{name}\\\", WHICH IS A **{autenticity}** NAME\")\n",
    "if name not in names:\n",
    "    with open(\"gen_names.txt\", \"a+\") as file:\n",
    "        file.seek(0)\n",
    "        file_names = file.read().splitlines()\n",
    "        if name not in file_names:\n",
    "            file.seek(0, 2)\n",
    "            file.write(f\"{name}\\n\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.4 ('deep-learning')",
   "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": "3.10.4"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "4bc787278428e062441c10305e68d335b87a7095797e2280490ed32d74dbb366"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}