Skip-Gram_word2vec.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "# Skip-gram word2vec\n",
    "\n",
    "In this notebook, I'll lead you through using TensorFlow to implement the word2vec algorithm using the skip-gram architecture. By implementing this, you'll learn about embedding words for use in natural language processing. This will come in handy when dealing with things like machine translation.\n",
    "\n",
    "## Readings\n",
    "\n",
    "Here are the resources I used to build this notebook. I suggest reading these either beforehand or while you're working on this material.\n",
    "\n",
    "* A really good [conceptual overview](http://mccormickml.com/2016/04/19/word2vec-tutorial-the-skip-gram-model/) of word2vec from Chris McCormick \n",
    "* [First word2vec paper](https://arxiv.org/pdf/1301.3781.pdf) from Mikolov et al.\n",
    "* [NIPS paper](http://papers.nips.cc/paper/5021-distributed-representations-of-words-and-phrases-and-their-compositionality.pdf) with improvements for word2vec also from Mikolov et al.\n",
    "* An [implementation of word2vec](http://www.thushv.com/natural_language_processing/word2vec-part-1-nlp-with-deep-learning-with-tensorflow-skip-gram/) from Thushan Ganegedara\n",
    "* TensorFlow [word2vec tutorial](https://www.tensorflow.org/tutorials/word2vec)\n",
    "\n",
    "## Word embeddings\n",
    "\n",
    "When you're dealing with words in text, you end up with tens of thousands of classes to predict, one for each word. Trying to one-hot encode these words is massively inefficient, you'll have one element set to 1 and the other 50,000 set to 0. The matrix multiplication going into the first hidden layer will have almost all of the resulting values be zero. This a huge waste of computation. \n",
    "\n",
    "![one-hot encodings](assets/one_hot_encoding.png)\n",
    "\n",
    "To solve this problem and greatly increase the efficiency of our networks, we use what are called embeddings. Embeddings are just a fully connected layer like you've seen before. We call this layer the embedding layer and the weights are embedding weights. We skip the multiplication into the embedding layer by instead directly grabbing the hidden layer values from the weight matrix. We can do this because the multiplication of a one-hot encoded vector with a matrix returns the row of the matrix corresponding the index of the \"on\" input unit.\n",
    "\n",
    "![lookup](assets/lookup_matrix.png)\n",
    "\n",
    "Instead of doing the matrix multiplication, we use the weight matrix as a lookup table. We encode the words as integers, for example \"heart\" is encoded as 958, \"mind\" as 18094. Then to get hidden layer values for \"heart\", you just take the 958th row of the embedding matrix. This process is called an **embedding lookup** and the number of hidden units is the **embedding dimension**.\n",
    "\n",
    "<img src='assets/tokenize_lookup.png' width=500>\n",
    " \n",
    "There is nothing magical going on here. The embedding lookup table is just a weight matrix. The embedding layer is just a hidden layer. The lookup is just a shortcut for the matrix multiplication. The lookup table is trained just like any weight matrix as well.\n",
    "\n",
    "Embeddings aren't only used for words of course. You can use them for any model where you have a massive number of classes. A particular type of model called **Word2Vec** uses the embedding layer to find vector representations of words that contain semantic meaning.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Word2Vec\n",
    "\n",
    "The word2vec algorithm finds much more efficient representations by finding vectors that represent the words. These vectors also contain semantic information about the words. Words that show up in similar contexts, such as \"black\", \"white\", and \"red\" will have vectors near each other. There are two architectures for implementing word2vec, CBOW (Continuous Bag-Of-Words) and Skip-gram.\n",
    "\n",
    "<img src=\"assets/word2vec_architectures.png\" width=\"500\">\n",
    "\n",
    "In this implementation, we'll be using the skip-gram architecture because it performs better than CBOW. Here, we pass in a word and try to predict the words surrounding it in the text. In this way, we can train the network to learn representations for words that show up in similar contexts.\n",
    "\n",
    "First up, importing packages."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "import time\n",
    "\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "\n",
    "import utils"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Load the [text8 dataset](http://mattmahoney.net/dc/textdata.html), a file of cleaned up Wikipedia articles from Matt Mahoney. The next cell will download the data set to the `data` folder. Then you can extract it and delete the archive file to save storage space."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Text8 Dataset: 31.4MB [00:15, 1.97MB/s]                            \n"
     ]
    }
   ],
   "source": [
    "from urllib.request import urlretrieve\n",
    "from os.path import isfile, isdir\n",
    "from tqdm import tqdm\n",
    "import zipfile\n",
    "\n",
    "dataset_folder_path = 'data'\n",
    "dataset_filename = 'text8.zip'\n",
    "dataset_name = 'Text8 Dataset'\n",
    "\n",
    "class DLProgress(tqdm):\n",
    "    last_block = 0\n",
    "\n",
    "    def hook(self, block_num=1, block_size=1, total_size=None):\n",
    "        self.total = total_size\n",
    "        self.update((block_num - self.last_block) * block_size)\n",
    "        self.last_block = block_num\n",
    "\n",
    "if not isfile(dataset_filename):\n",
    "    with DLProgress(unit='B', unit_scale=True, miniters=1, desc=dataset_name) as pbar:\n",
    "        urlretrieve(\n",
    "            'http://mattmahoney.net/dc/text8.zip',\n",
    "            dataset_filename,\n",
    "            pbar.hook)\n",
    "\n",
    "if not isdir(dataset_folder_path):\n",
    "    with zipfile.ZipFile(dataset_filename) as zip_ref:\n",
    "        zip_ref.extractall(dataset_folder_path)\n",
    "        \n",
    "with open('data/text8') as f:\n",
    "    text = f.read()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Preprocessing\n",
    "\n",
    "Here I'm fixing up the text to make training easier. This comes from the `utils` module I wrote. The `preprocess` function coverts any punctuation into tokens, so a period is changed to ` <PERIOD> `. In this data set, there aren't any periods, but it will help in other NLP problems. I'm also removing all words that show up five or fewer times in the dataset. This will greatly reduce issues due to noise in the data and improve the quality of the vector representations. If you want to write your own functions for this stuff, go for it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['anarchism', 'originated', 'as', 'a', 'term', 'of', 'abuse', 'first', 'used', 'against', 'early', 'working', 'class', 'radicals', 'including', 'the', 'diggers', 'of', 'the', 'english', 'revolution', 'and', 'the', 'sans', 'culottes', 'of', 'the', 'french', 'revolution', 'whilst']\n"
     ]
    }
   ],
   "source": [
    "words = utils.preprocess(text)\n",
    "print(words[:30])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total words: 16680599\n",
      "Unique words: 63641\n"
     ]
    }
   ],
   "source": [
    "print(\"Total words: {}\".format(len(words)))\n",
    "print(\"Unique words: {}\".format(len(set(words))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "And here I'm creating dictionaries to covert words to integers and backwards, integers to words. The integers are assigned in descending frequency order, so the most frequent word (\"the\") is given the integer 0 and the next most frequent is 1 and so on. The words are converted to integers and stored in the list `int_words`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "vocab_to_int, int_to_vocab = utils.create_lookup_tables(words)\n",
    "int_words = [vocab_to_int[word] for word in words]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Subsampling\n",
    "\n",
    "Words that show up often such as \"the\", \"of\", and \"for\" don't provide much context to the nearby words. If we discard some of them, we can remove some of the noise from our data and in return get faster training and better representations. This process is called subsampling by Mikolov. For each word $w_i$ in the training set, we'll discard it with probability given by \n",
    "\n",
    "$$ P(w_i) = 1 - \\sqrt{\\frac{t}{f(w_i)}} $$\n",
    "\n",
    "where $t$ is a threshold parameter and $f(w_i)$ is the frequency of word $w_i$ in the total dataset.\n",
    "\n",
    "I'm going to leave this up to you as an exercise. This is more of a programming challenge, than about deep learning specifically. But, being able to prepare your data for your network is an important skill to have. Check out my solution to see how I did it.\n",
    "\n",
    "> **Exercise:** Implement subsampling for the words in `int_words`. That is, go through `int_words` and discard each word given the probablility $P(w_i)$ shown above. Note that $P(w_i)$ is the probability that a word is discarded. Assign the subsampled data to `train_words`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "from collections import Counter\n",
    "import random\n",
    "\n",
    "# Set the threshold parameter, t\n",
    "threshold = 0.01\n",
    "\n",
    "# Get total word counts\n",
    "word_counts = Counter(int_words)\n",
    "total_count = len(int_words)\n",
    "\n",
    "# Calculate the word frequencies, f(w_i)\n",
    "word_frequencies = {word: count/total_count for word, count in word_counts.items()}\n",
    "\n",
    "# Calculate the drop probabilities\n",
    "p_drop = {word: 1 - np.sqrt(threshold / word_frequencies[word]) for word in word_counts}\n",
    "\n",
    "train_words = [word for word in int_words if random.random() < (1 - p_drop[word])]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Making batches"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Now that our data is in good shape, we need to get it into the proper form to pass it into our network. With the skip-gram architecture, for each word in the text, we want to grab all the words in a window around that word, with size $C$. \n",
    "\n",
    "From [Mikolov et al.](https://arxiv.org/pdf/1301.3781.pdf): \n",
    "\n",
    "\"Since the more distant words are usually less related to the current word than those close to it, we give less weight to the distant words by sampling less from those words in our training examples... If we choose $C = 5$, for each training word we will select randomly a number $R$ in range $< 1; C >$, and then use $R$ words from history and $R$ words from the future of the current word as correct labels.\"\n",
    "\n",
    "> **Exercise:** Implement a function `get_target` that receives a list of words, an index, and a window size, then returns a list of words in the window around the index. Make sure to use the algorithm described above, where you choose a random number of words from the window."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "def get_target(words, idx, window_size=5):\n",
    "    ''' Get a list of words in a window around an index. '''\n",
    "    \n",
    "    range = random.randint(1, window_size)\n",
    "    \n",
    "    # Get begin and end indices\n",
    "    start = idx - range\n",
    "    end = idx + range + 1\n",
    "    \n",
    "    # Make sure start and stop indices are within range\n",
    "    if start < 0:\n",
    "        start = 0\n",
    "    if end > len(words):\n",
    "        end = len(words)\n",
    "    \n",
    "    return words[start:end]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Here's a function that returns batches for our network. The idea is that it grabs `batch_size` words from a words list. Then for each of those words, it gets the target words in the window. I haven't found a way to pass in a random number of target words and get it to work with the architecture, so I make one row per input-target pair. This is a generator function by the way, helps save memory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "def get_batches(words, batch_size, window_size=5):\n",
    "    ''' Create a generator of word batches as a tuple (inputs, targets) '''\n",
    "    \n",
    "    n_batches = len(words)//batch_size\n",
    "    \n",
    "    # only full batches\n",
    "    words = words[:n_batches*batch_size]\n",
    "    \n",
    "    for idx in range(0, len(words), batch_size):\n",
    "        x, y = [], []\n",
    "        batch = words[idx:idx+batch_size]\n",
    "        for ii in range(len(batch)):\n",
    "            batch_x = batch[ii]\n",
    "            batch_y = get_target(batch, ii, window_size)\n",
    "            y.extend(batch_y)\n",
    "            x.extend([batch_x]*len(batch_y))\n",
    "        yield x, y\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Building the graph\n",
    "\n",
    "From [Chris McCormick's blog](http://mccormickml.com/2016/04/19/word2vec-tutorial-the-skip-gram-model/), we can see the general structure of our network.\n",
    "![embedding_network](./assets/skip_gram_net_arch.png)\n",
    "\n",
    "The input words are passed in as integers. This will go into a hidden layer of linear units, then into a softmax layer. We'll use the softmax layer to make a prediction like normal.\n",
    "\n",
    "The idea here is to train the hidden layer weight matrix to find efficient representations for our words. We can discard the softmax layer becuase we don't really care about making predictions with this network. We just want the embedding matrix so we can use it in other networks we build from the dataset.\n",
    "\n",
    "I'm going to have you build the graph in stages now. First off, creating the `inputs` and `labels` placeholders like normal.\n",
    "\n",
    "> **Exercise:** Assign `inputs` and `labels` using `tf.placeholder`. We're going to be passing in integers, so set the data types to `tf.int32`. The batches we're passing in will have varying sizes, so set the batch sizes to [`None`]. To make things work later, you'll need to set the second dimension of `labels` to `None` or `1`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "train_graph = tf.Graph()\n",
    "with train_graph.as_default():\n",
    "    inputs = tf.placeholder(tf.int32, shape=[None], name='inputs')\n",
    "    labels = tf.placeholder(tf.int32, shape=[None, None], name='labels')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Embedding\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "source": [
    "The embedding matrix has a size of the number of words by the number of units in the hidden layer. So, if you have 10,000 words and 300 hidden units, the matrix will have size $10,000 \\times 300$. Remember that we're using tokenized data for our inputs, usually as integers, where the number of tokens is the number of words in our vocabulary.\n",
    "\n",
    "\n",
    "> **Exercise:** Tensorflow provides a convenient function [`tf.nn.embedding_lookup`](https://www.tensorflow.org/api_docs/python/tf/nn/embedding_lookup) that does this lookup for us. You pass in the embedding matrix and a tensor of integers, then it returns rows in the matrix corresponding to those integers. Below, set the number of embedding features you'll use (200 is a good start), create the embedding matrix variable, and use `tf.nn.embedding_lookup` to get the embedding tensors. For the embedding matrix, I suggest you initialize it with a uniform random numbers between -1 and 1 using [tf.random_uniform](https://www.tensorflow.org/api_docs/python/tf/random_uniform)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "n_vocab = len(int_to_vocab)\n",
    "n_embedding = 200 # Number of embedding features \n",
    "with train_graph.as_default():\n",
    "    embedding = tf.Variable(tf.random_uniform((n_vocab, n_embedding), -1, 1))\n",
    "    embed = tf.nn.embedding_lookup(embedding, inputs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Negative sampling\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "For every example we give the network, we train it using the output from the softmax layer. That means for each input, we're making very small changes to millions of weights even though we only have one true example. This makes training the network very inefficient. We can approximate the loss from the softmax layer by only updating a small subset of all the weights at once. We'll update the weights for the correct label, but only a small number of incorrect labels. This is called [\"negative sampling\"](http://papers.nips.cc/paper/5021-distributed-representations-of-words-and-phrases-and-their-compositionality.pdf). Tensorflow has a convenient function to do this, [`tf.nn.sampled_softmax_loss`](https://www.tensorflow.org/api_docs/python/tf/nn/sampled_softmax_loss).\n",
    "\n",
    "> **Exercise:** Below, create weights and biases for the softmax layer. Then, use [`tf.nn.sampled_softmax_loss`](https://www.tensorflow.org/api_docs/python/tf/nn/sampled_softmax_loss) to calculate the loss. Be sure to read the documentation to figure out how it works."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "# Number of negative labels to sample\n",
    "n_sampled = 100\n",
    "with train_graph.as_default():\n",
    "    softmax_w = tf.Variable(tf.truncated_normal((n_vocab, n_embedding), stddev=0.1))\n",
    "    softmax_b = tf.Variable(tf.zeros(n_vocab))\n",
    "    \n",
    "    # Calculate the loss using negative sampling\n",
    "    loss = tf.nn.sampled_softmax_loss(softmax_w, softmax_b, labels, embed, n_sampled, n_vocab)\n",
    "    \n",
    "    cost = tf.reduce_mean(loss)\n",
    "    optimizer = tf.train.AdamOptimizer().minimize(cost)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Validation\n",
    "\n",
    "This code is from Thushan Ganegedara's implementation. Here we're going to choose a few common words and few uncommon words. Then, we'll print out the closest words to them. It's a nice way to check that our embedding table is grouping together words with similar semantic meanings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "with train_graph.as_default():\n",
    "    ## From Thushan Ganegedara's implementation\n",
    "    valid_size = 16 # Random set of words to evaluate similarity on.\n",
    "    valid_window = 100\n",
    "    # pick 8 samples from (0,100) and (1000,1100) each ranges. lower id implies more frequent \n",
    "    valid_examples = np.array(random.sample(range(valid_window), valid_size//2))\n",
    "    valid_examples = np.append(valid_examples, \n",
    "                               random.sample(range(1000,1000+valid_window), valid_size//2))\n",
    "\n",
    "    valid_dataset = tf.constant(valid_examples, dtype=tf.int32)\n",
    "    \n",
    "    # We use the cosine distance:\n",
    "    norm = tf.sqrt(tf.reduce_sum(tf.square(embedding), 1, keep_dims=True))\n",
    "    normalized_embedding = embedding / norm\n",
    "    valid_embedding = tf.nn.embedding_lookup(normalized_embedding, valid_dataset)\n",
    "    similarity = tf.matmul(valid_embedding, tf.transpose(normalized_embedding))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "# If the checkpoints directory doesn't exist:\n",
    "!mkdir checkpoints"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Training\n",
    "\n",
    "Below is the code to train the network. Every 100 batches it reports the training loss. Every 1000 batches, it'll print out the validation words."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10 Iteration: 100 Avg. Training loss: 7.8091 0.2632 sec/batch\n",
      "Epoch 1/10 Iteration: 200 Avg. Training loss: 7.2983 0.2575 sec/batch\n",
      "Epoch 1/10 Iteration: 300 Avg. Training loss: 6.7337 0.2590 sec/batch\n",
      "Epoch 1/10 Iteration: 400 Avg. Training loss: 6.1914 0.2622 sec/batch\n",
      "Epoch 1/10 Iteration: 500 Avg. Training loss: 5.7819 0.2644 sec/batch\n",
      "Epoch 1/10 Iteration: 600 Avg. Training loss: 5.5634 0.2603 sec/batch\n",
      "Epoch 1/10 Iteration: 700 Avg. Training loss: 5.3380 0.2589 sec/batch\n",
      "Epoch 1/10 Iteration: 800 Avg. Training loss: 5.2047 0.2524 sec/batch\n",
      "Epoch 1/10 Iteration: 900 Avg. Training loss: 4.9899 0.2866 sec/batch\n",
      "Epoch 1/10 Iteration: 1000 Avg. Training loss: 4.8526 0.2748 sec/batch\n",
      "Nearest to that: too, promise, celebrity, wives, vagabond, includes, up, without,\n",
      "Nearest to not: perhaps, bros, think, parameters, finnish, necessary, been, non,\n",
      "Nearest to also: el, maintainer, club, eclipse, control, rouge, defeated, let,\n",
      "Nearest to he: responses, holds, sold, caller, marriage, february, legnica, those,\n",
      "Nearest to all: materials, talk, functions, powers, attack, world, pastoral, developed,\n",
      "Nearest to used: british, yards, implications, abd, posts, mixed, screening, plans,\n",
      "Nearest to i: forces, india, furniture, spartan, prejudice, reliefs, alliance, exact,\n",
      "Nearest to when: parity, remaining, methods, attest, these, poking, muddle, odense,\n",
      "Nearest to woman: idea, minority, odrade, casbah, jordan, creators, hatred, reputed,\n",
      "Nearest to dr: citation, pietro, hexafluoride, active, requirements, bratislava, australian, him,\n",
      "Nearest to mainly: iceberg, lose, promise, wins, without, laws, leaving, volterra,\n",
      "Nearest to scale: nehemiah, blockades, fingerprinting, findlay, neutralist, denotation, governorates, carney,\n",
      "Nearest to test: berkeley, iliad, prejudice, helens, rajputs, el, stolpe, romero,\n",
      "Nearest to lived: course, eunuch, cinematographers, given, edge, countries, triumph, wishful,\n",
      "Nearest to road: uncredited, tosses, reznor, rich, dia, goldsmiths, aided, theodosian,\n",
      "Nearest to ocean: schalk, reconciled, tongues, inalienable, length, ties, gauntlets, posed,\n",
      "Epoch 1/10 Iteration: 1100 Avg. Training loss: 4.9481 0.2542 sec/batch\n",
      "Epoch 1/10 Iteration: 1200 Avg. Training loss: 4.8852 0.2488 sec/batch\n",
      "Epoch 1/10 Iteration: 1300 Avg. Training loss: 4.8370 0.2500 sec/batch\n",
      "Epoch 1/10 Iteration: 1400 Avg. Training loss: 4.7910 0.2818 sec/batch\n",
      "Epoch 1/10 Iteration: 1500 Avg. Training loss: 4.7503 0.2656 sec/batch\n",
      "Epoch 1/10 Iteration: 1600 Avg. Training loss: 4.7422 0.2661 sec/batch\n",
      "Epoch 1/10 Iteration: 1700 Avg. Training loss: 4.6872 0.2667 sec/batch\n",
      "Epoch 1/10 Iteration: 1800 Avg. Training loss: 4.7107 0.2769 sec/batch\n",
      "Epoch 1/10 Iteration: 1900 Avg. Training loss: 4.6493 0.2660 sec/batch\n",
      "Epoch 1/10 Iteration: 2000 Avg. Training loss: 4.6902 0.2641 sec/batch\n",
      "Nearest to that: too, promise, pps, mib, without, cremated, wives, huron,\n",
      "Nearest to not: perhaps, parameters, bros, necessary, think, onward, balto, thinly,\n",
      "Nearest to also: maintainer, eclipse, el, rouge, chile, resign, club, mathers,\n",
      "Nearest to he: responses, caller, academic, holds, desserts, sold, maclean, legnica,\n",
      "Nearest to all: materials, talk, adjustment, functions, serpent, monitoring, snake, duplicated,\n",
      "Nearest to used: implications, yards, sony, screening, british, posts, integers, proxies,\n",
      "Nearest to i: forces, furniture, reliefs, annual, spartan, india, prejudice, bankrupt,\n",
      "Nearest to when: parity, poking, makeshift, arabs, bough, governors, mitra, attest,\n",
      "Nearest to woman: minority, idea, creators, prevailing, corsican, odrade, reputed, regions,\n",
      "Nearest to dr: citation, pietro, thoroughbreds, bratislava, requirements, timothy, hexafluoride, modest,\n",
      "Nearest to mainly: iceberg, spaces, promise, wins, criminal, activate, volterra, without,\n",
      "Nearest to scale: nehemiah, blockades, fingerprinting, denotation, hunter, copyrightable, midst, carney,\n",
      "Nearest to test: rajputs, berkeley, iliad, harold, prejudice, mediterranean, romero, el,\n",
      "Nearest to lived: course, occasional, surprise, eunuch, cinematographers, wishful, buying, edge,\n",
      "Nearest to road: uncredited, reznor, unclear, goldsmiths, theodosian, aided, tosses, dia,\n",
      "Nearest to ocean: schalk, reconciled, tongues, posed, gauntlets, ties, inalienable, length,\n",
      "Epoch 1/10 Iteration: 2100 Avg. Training loss: 4.6631 0.2653 sec/batch\n",
      "Epoch 1/10 Iteration: 2200 Avg. Training loss: 4.4682 0.2600 sec/batch\n",
      "Epoch 1/10 Iteration: 2300 Avg. Training loss: 4.6410 0.2687 sec/batch\n",
      "Epoch 1/10 Iteration: 2400 Avg. Training loss: 4.5866 0.2645 sec/batch\n",
      "Epoch 1/10 Iteration: 2500 Avg. Training loss: 4.5615 0.2638 sec/batch\n",
      "Epoch 1/10 Iteration: 2600 Avg. Training loss: 4.5765 0.2661 sec/batch\n",
      "Epoch 1/10 Iteration: 2700 Avg. Training loss: 4.5616 0.2656 sec/batch\n",
      "Epoch 1/10 Iteration: 2800 Avg. Training loss: 4.5002 0.2646 sec/batch\n",
      "Epoch 1/10 Iteration: 2900 Avg. Training loss: 4.4737 0.2644 sec/batch\n",
      "Epoch 1/10 Iteration: 3000 Avg. Training loss: 4.5002 0.2645 sec/batch\n",
      "Nearest to that: too, mib, sudden, cremated, pps, toughest, vagabond, huron,\n",
      "Nearest to not: perhaps, bros, parameters, does, think, balto, onward, thinly,\n",
      "Nearest to also: maintainer, eclipse, rouge, unicameral, el, yielding, chile, workmen,\n",
      "Nearest to he: caller, responses, desserts, maclean, him, unarmed, beloved, legnica,\n",
      "Nearest to all: materials, adjustment, serpent, functions, talk, duplicated, monitoring, purdue,\n",
      "Nearest to used: implications, yards, integers, sony, tomahawk, ewing, british, screening,\n",
      "Nearest to i: forces, furniture, annual, intensive, spartan, reliefs, relevance, handful,\n",
      "Nearest to when: parity, makeshift, bough, poking, dubai, mcgraw, governors, attest,\n",
      "Nearest to woman: idea, odrade, telecast, corsican, minority, prevailing, casbah, quiescent,\n",
      "Nearest to dr: citation, pietro, bratislava, timothy, thoroughbreds, modest, hexafluoride, positioned,\n",
      "Nearest to mainly: iceberg, spaces, activate, wins, criminal, meat, volterra, promise,\n",
      "Nearest to scale: nehemiah, blockades, denotation, primates, fingerprinting, afb, serfdom, carney,\n",
      "Nearest to test: berkeley, rajputs, iliad, donors, drops, esoteric, harold, balcony,\n",
      "Nearest to lived: cinematographers, eunuch, occasional, idaho, surprise, interactivity, course, sexually,\n",
      "Nearest to road: uncredited, goldsmiths, reznor, theodosian, naturalist, unclear, aided, peder,\n",
      "Nearest to ocean: posed, schalk, tongues, reconciled, gauntlets, inalienable, hel, slumped,\n",
      "Epoch 1/10 Iteration: 3100 Avg. Training loss: 4.5396 0.2676 sec/batch\n",
      "Epoch 1/10 Iteration: 3200 Avg. Training loss: 4.5145 0.2666 sec/batch\n",
      "Epoch 1/10 Iteration: 3300 Avg. Training loss: 4.4862 0.2645 sec/batch\n",
      "Epoch 1/10 Iteration: 3400 Avg. Training loss: 4.4551 0.2643 sec/batch\n",
      "Epoch 1/10 Iteration: 3500 Avg. Training loss: 4.4676 0.2686 sec/batch\n",
      "Epoch 1/10 Iteration: 3600 Avg. Training loss: 4.4560 0.2677 sec/batch\n",
      "Epoch 1/10 Iteration: 3700 Avg. Training loss: 4.3146 0.2798 sec/batch\n",
      "Epoch 1/10 Iteration: 3800 Avg. Training loss: 4.4373 0.2876 sec/batch\n",
      "Epoch 1/10 Iteration: 3900 Avg. Training loss: 4.4675 0.2758 sec/batch\n",
      "Epoch 1/10 Iteration: 4000 Avg. Training loss: 4.3966 0.2538 sec/batch\n",
      "Nearest to that: too, mib, pps, sudden, cremated, vagabond, toughest, inviting,\n",
      "Nearest to not: does, perhaps, bros, parameters, accommodates, think, coll, emirs,\n",
      "Nearest to also: maintainer, unicameral, eclipse, chile, rouge, el, workmen, yielding,\n",
      "Nearest to he: caller, him, desserts, maclean, responses, unarmed, obey, beloved,\n",
      "Nearest to all: materials, adjustment, monitoring, functions, purdue, duplicated, serpent, adder,\n",
      "Nearest to used: integers, yards, implications, sony, tomahawk, effigies, screening, prabang,\n",
      "Nearest to i: furniture, intensive, annual, forces, reliefs, relevance, spartan, graves,\n",
      "Nearest to when: poking, makeshift, dubai, parity, bough, mcgraw, contemplated, arabs,\n",
      "Nearest to woman: corsican, casbah, sonnets, odrade, telecast, quiescent, idea, minority,\n",
      "Nearest to dr: citation, pietro, thoroughbreds, timothy, bratislava, hexafluoride, modest, insurgent,\n",
      "Nearest to mainly: iceberg, activate, spaces, wins, onslaught, peg, meat, criminal,\n",
      "Nearest to scale: nehemiah, blockades, denotation, primates, afb, serfdom, ley, fingerprinting,\n",
      "Nearest to test: berkeley, donors, rajputs, romero, stolpe, unfair, iliad, esoteric,\n",
      "Nearest to lived: eunuch, interactivity, idaho, cinematographers, surprise, inadvertently, buying, wishful,\n",
      "Nearest to road: uncredited, theodosian, reznor, goldsmiths, reflective, naturalist, hildebrand, dia,\n",
      "Nearest to ocean: schalk, posed, inalienable, reconciled, gauntlets, tongues, slumped, leeds,\n",
      "Epoch 1/10 Iteration: 4100 Avg. Training loss: 4.4265 0.2537 sec/batch\n",
      "Epoch 1/10 Iteration: 4200 Avg. Training loss: 4.4162 0.2497 sec/batch\n",
      "Epoch 1/10 Iteration: 4300 Avg. Training loss: 4.3996 0.2504 sec/batch\n",
      "Epoch 1/10 Iteration: 4400 Avg. Training loss: 4.4387 0.2516 sec/batch\n",
      "Epoch 1/10 Iteration: 4500 Avg. Training loss: 4.3795 0.2505 sec/batch\n",
      "Epoch 1/10 Iteration: 4600 Avg. Training loss: 4.4185 0.2509 sec/batch\n",
      "Epoch 1/10 Iteration: 4700 Avg. Training loss: 4.4096 0.2503 sec/batch\n",
      "Epoch 1/10 Iteration: 4800 Avg. Training loss: 4.4030 0.2496 sec/batch\n",
      "Epoch 1/10 Iteration: 4900 Avg. Training loss: 4.4068 0.2508 sec/batch\n",
      "Epoch 1/10 Iteration: 5000 Avg. Training loss: 4.4114 0.2511 sec/batch\n",
      "Nearest to that: mib, too, toughest, pps, cremated, vagabond, revisited, sudden,\n",
      "Nearest to not: does, bros, perhaps, parameters, accommodates, emirs, coll, detoxification,\n",
      "Nearest to also: maintainer, unicameral, duns, irvine, el, piedmont, mathers, workmen,\n",
      "Nearest to he: caller, him, desserts, maclean, legnica, responses, unarmed, beloved,\n",
      "Nearest to all: materials, serpent, adjustment, monitoring, purdue, duplicated, insult, functions,\n",
      "Nearest to used: integers, yards, sony, screening, frontispiece, effigies, proxies, implications,\n",
      "Nearest to i: furniture, t, sensor, annual, forces, intensive, reliefs, spartan,\n",
      "Nearest to when: makeshift, poking, dubai, bough, contemplated, affable, parity, kidman,\n",
      "Nearest to woman: corsican, odrade, sonnets, casbah, quiescent, tarmac, idea, telecast,\n",
      "Nearest to dr: citation, pietro, thoroughbreds, bratislava, insurgent, timothy, dissenters, hexafluoride,\n",
      "Nearest to mainly: iceberg, activate, qwerty, onslaught, evangelism, wins, spaces, escorted,\n",
      "Nearest to scale: blockades, nehemiah, serfdom, primates, afb, denotation, carney, findlay,\n",
      "Nearest to test: rajputs, donors, berkeley, romero, unfair, helens, stolpe, drops,\n",
      "Nearest to lived: eunuch, idaho, interactivity, cinematographers, slips, surprise, inadvertently, buying,\n",
      "Nearest to road: uncredited, theodosian, reznor, ounces, goldsmiths, naturalist, reflective, peder,\n",
      "Nearest to ocean: schalk, posed, inalienable, slumped, reconciled, tongues, gauntlets, leeds,\n",
      "Epoch 1/10 Iteration: 5100 Avg. Training loss: 4.3711 0.2524 sec/batch\n",
      "Epoch 1/10 Iteration: 5200 Avg. Training loss: 4.3516 0.2494 sec/batch\n",
      "Epoch 1/10 Iteration: 5300 Avg. Training loss: 4.4097 0.2496 sec/batch\n",
      "Epoch 1/10 Iteration: 5400 Avg. Training loss: 4.4075 0.2501 sec/batch\n",
      "Epoch 1/10 Iteration: 5500 Avg. Training loss: 4.3365 0.2507 sec/batch\n",
      "Epoch 1/10 Iteration: 5600 Avg. Training loss: 4.3352 0.2499 sec/batch\n",
      "Epoch 1/10 Iteration: 5700 Avg. Training loss: 4.3345 0.2502 sec/batch\n",
      "Epoch 1/10 Iteration: 5800 Avg. Training loss: 4.2983 0.2499 sec/batch\n",
      "Epoch 1/10 Iteration: 5900 Avg. Training loss: 4.3687 0.2504 sec/batch\n",
      "Epoch 1/10 Iteration: 6000 Avg. Training loss: 4.3577 0.2500 sec/batch\n",
      "Nearest to that: mib, toughest, cremated, pps, vagabond, revisited, remind, too,\n",
      "Nearest to not: does, bros, perhaps, accommodates, nightbreed, parameters, emirs, coll,\n",
      "Nearest to also: maintainer, mathers, irvine, see, outweigh, flattery, piedmont, dentition,\n",
      "Nearest to he: caller, him, legnica, desserts, beloved, maclean, tribesmen, ceau,\n",
      "Nearest to all: monitoring, serpent, adjustment, purdue, materials, lydia, functions, insult,\n",
      "Nearest to used: integers, commonly, yards, sony, frontispiece, screening, reference, effigies,\n",
      "Nearest to i: t, intensive, furniture, sensor, read, spartan, escobar, forces,\n",
      "Nearest to when: makeshift, poking, contemplated, bough, ionian, dubai, affable, shelley,\n",
      "Nearest to woman: casbah, sonnets, corsican, odrade, tarmac, quiescent, telecast, idea,\n",
      "Nearest to dr: citation, pietro, timothy, thoroughbreds, insurgent, hexafluoride, bratislava, rightful,\n",
      "Nearest to mainly: iceberg, activate, evangelism, qwerty, wins, lingala, tradeoffs, surreal,\n",
      "Nearest to scale: blockades, nehemiah, serfdom, primates, ley, carney, denotation, afb,\n",
      "Nearest to test: rajputs, donors, berkeley, helens, romero, stolpe, unfair, esoteric,\n",
      "Nearest to lived: interactivity, eunuch, cinematographers, idaho, slips, wishful, buying, surprise,\n",
      "Nearest to road: uncredited, theodosian, reznor, ounces, goldsmiths, naturalist, reflective, dia,\n",
      "Nearest to ocean: schalk, inalienable, posed, slumped, reconciled, leeds, gauntlets, tongues,\n",
      "Epoch 1/10 Iteration: 6100 Avg. Training loss: 4.3732 0.2521 sec/batch\n",
      "Epoch 1/10 Iteration: 6200 Avg. Training loss: 4.3495 0.2497 sec/batch\n",
      "Epoch 1/10 Iteration: 6300 Avg. Training loss: 4.3708 0.2514 sec/batch\n",
      "Epoch 1/10 Iteration: 6400 Avg. Training loss: 4.3300 0.2502 sec/batch\n",
      "Epoch 1/10 Iteration: 6500 Avg. Training loss: 4.2995 0.2502 sec/batch\n",
      "Epoch 1/10 Iteration: 6600 Avg. Training loss: 4.2885 0.2493 sec/batch\n",
      "Epoch 1/10 Iteration: 6700 Avg. Training loss: 4.3268 0.2495 sec/batch\n",
      "Epoch 1/10 Iteration: 6800 Avg. Training loss: 4.2623 0.2492 sec/batch\n",
      "Epoch 1/10 Iteration: 6900 Avg. Training loss: 4.3320 0.2515 sec/batch\n",
      "Epoch 1/10 Iteration: 7000 Avg. Training loss: 4.3196 0.2498 sec/batch\n",
      "Nearest to that: mib, vagabond, toughest, cremated, inviting, energies, revisited, postulated,\n",
      "Nearest to not: does, bros, did, nightbreed, accommodates, perhaps, swinging, but,\n",
      "Nearest to also: see, maintainer, el, pocketing, flattery, mathers, outweigh, irvine,\n",
      "Nearest to he: him, caller, beloved, legnica, desserts, marriage, tribesmen, sweating,\n",
      "Nearest to all: purdue, monitoring, pollinator, lydia, insult, serpent, adjustment, materials,\n",
      "Nearest to used: integers, commonly, screening, yards, sony, frontispiece, effigies, resisters,\n",
      "Nearest to i: t, furniture, am, intensive, escobar, sensor, grad, read,\n",
      "Nearest to when: makeshift, contemplated, ionian, kidman, poking, dubai, cdot, insufficiently,\n",
      "Nearest to woman: sonnets, odrade, casbah, tarmac, corsican, quiescent, idea, telecast,\n",
      "Nearest to dr: citation, timothy, hexafluoride, pietro, thoroughbreds, insurgent, bratislava, constantinian,\n",
      "Nearest to mainly: iceberg, activate, qwerty, evangelism, lingala, wins, metroid, gulfstream,\n",
      "Nearest to scale: blockades, nehemiah, serfdom, primates, carney, ley, denotation, afb,\n",
      "Nearest to test: rajputs, helens, donors, romero, stolpe, berkeley, unfair, esoteric,\n",
      "Nearest to lived: interactivity, parton, slips, idaho, cinematographers, eunuch, wishful, asmara,\n",
      "Nearest to road: uncredited, theodosian, reznor, ounces, goldsmiths, naturalist, reflective, trotskyists,\n",
      "Nearest to ocean: schalk, inalienable, posed, leeds, slumped, tongues, dungeons, west,\n",
      "Epoch 1/10 Iteration: 7100 Avg. Training loss: 4.3228 0.2519 sec/batch\n",
      "Epoch 1/10 Iteration: 7200 Avg. Training loss: 4.2819 0.2491 sec/batch\n",
      "Epoch 1/10 Iteration: 7300 Avg. Training loss: 4.3239 0.2499 sec/batch\n",
      "Epoch 1/10 Iteration: 7400 Avg. Training loss: 4.3205 0.2497 sec/batch\n",
      "Epoch 1/10 Iteration: 7500 Avg. Training loss: 4.2468 0.2502 sec/batch\n",
      "Epoch 1/10 Iteration: 7600 Avg. Training loss: 4.3134 0.2494 sec/batch\n",
      "Epoch 1/10 Iteration: 7700 Avg. Training loss: 4.2184 0.2517 sec/batch\n",
      "Epoch 1/10 Iteration: 7800 Avg. Training loss: 4.2742 0.2491 sec/batch\n",
      "Epoch 1/10 Iteration: 7900 Avg. Training loss: 4.2784 0.2494 sec/batch\n",
      "Epoch 1/10 Iteration: 8000 Avg. Training loss: 4.1837 0.2495 sec/batch\n",
      "Nearest to that: mib, vagabond, toughest, ensuring, inviting, cremated, energies, pps,\n",
      "Nearest to not: does, did, accommodates, bros, nightbreed, swinging, but, perhaps,\n",
      "Nearest to also: see, mathers, pocketing, el, flattery, maintainer, dentition, outweigh,\n",
      "Nearest to he: him, caller, marriage, beloved, nihonshoki, tribesmen, ceau, legnica,\n",
      "Nearest to all: serpent, pollinator, monitoring, lydia, insult, purdue, duplicated, monorail,\n",
      "Nearest to used: commonly, integers, sony, resisters, effigies, screening, frontispiece, yards,\n",
      "Nearest to i: t, am, escobar, sensor, grad, furniture, intensive, barkley,\n",
      "Nearest to when: makeshift, contemplated, kidman, ionian, insufficiently, dubai, cdot, poking,\n",
      "Nearest to woman: sonnets, corsican, casbah, odrade, quiescent, tarmac, telecast, tchad,\n",
      "Nearest to dr: citation, hexafluoride, timothy, thoroughbreds, insurgent, pietro, constantinian, bratislava,\n",
      "Nearest to mainly: iceberg, activate, lingala, evangelism, metroid, qwerty, spymaster, gulfstream,\n",
      "Nearest to scale: blockades, nehemiah, carney, primates, ley, afb, denotation, serfdom,\n",
      "Nearest to test: rajputs, helens, donors, romero, aberdeen, unfair, qurra, drops,\n",
      "Nearest to lived: parton, idaho, interactivity, eunuch, cinematographers, slips, wishful, asmara,\n",
      "Nearest to road: theodosian, uncredited, reznor, ounces, goldsmiths, trotskyists, naturalist, reflective,\n",
      "Nearest to ocean: schalk, inalienable, posed, leeds, slumped, west, dungeons, landing,\n",
      "Epoch 1/10 Iteration: 8100 Avg. Training loss: 4.3307 0.2530 sec/batch\n",
      "Epoch 1/10 Iteration: 8200 Avg. Training loss: 4.3037 0.2497 sec/batch\n",
      "Epoch 1/10 Iteration: 8300 Avg. Training loss: 4.2808 0.2496 sec/batch\n",
      "Epoch 1/10 Iteration: 8400 Avg. Training loss: 4.2707 0.2499 sec/batch\n",
      "Epoch 1/10 Iteration: 8500 Avg. Training loss: 4.2674 0.2500 sec/batch\n",
      "Epoch 1/10 Iteration: 8600 Avg. Training loss: 4.1904 0.2508 sec/batch\n",
      "Epoch 1/10 Iteration: 8700 Avg. Training loss: 4.2675 0.2498 sec/batch\n",
      "Epoch 1/10 Iteration: 8800 Avg. Training loss: 4.2854 0.2497 sec/batch\n",
      "Epoch 1/10 Iteration: 8900 Avg. Training loss: 4.2948 0.2498 sec/batch\n",
      "Epoch 1/10 Iteration: 9000 Avg. Training loss: 4.2903 0.2492 sec/batch\n",
      "Nearest to that: mib, vagabond, cremated, toughest, suggested, retailer, pps, postulated,\n",
      "Nearest to not: does, did, bros, accommodates, nightbreed, swinging, but, do,\n",
      "Nearest to also: see, pocketing, list, flattery, outweigh, piedmont, el, dentition,\n",
      "Nearest to he: him, caller, beloved, nihonshoki, marriage, legnica, tribesmen, celeste,\n",
      "Nearest to all: pollinator, purdue, lydia, monitoring, serpent, functions, insult, pissarro,\n",
      "Nearest to used: commonly, integers, sony, resisters, effigies, frontispiece, screening, serviced,\n",
      "Nearest to i: am, t, sensor, think, you, grad, escobar, hippopotamus,\n",
      "Nearest to when: makeshift, contemplated, kidman, ionian, insufficiently, mitra, dubai, poking,\n",
      "Nearest to woman: sonnets, corsican, casbah, odrade, quiescent, telecast, tarmac, tchad,\n",
      "Nearest to dr: citation, hexafluoride, thoroughbreds, timothy, constantinian, insurgent, pietro, rightful,\n",
      "Nearest to mainly: iceberg, activate, evangelism, metroid, lingala, qwerty, spymaster, lg,\n",
      "Nearest to scale: blockades, nehemiah, primates, carney, ley, serfdom, auctions, rotors,\n",
      "Nearest to test: helens, rajputs, donors, romero, qurra, aberdeen, drops, stolpe,\n",
      "Nearest to lived: parton, interactivity, marshland, eunuch, slips, idaho, cinematographers, wishful,\n",
      "Nearest to road: theodosian, uncredited, reznor, trotskyists, ounces, goldsmiths, reflective, offbeat,\n",
      "Nearest to ocean: schalk, inalienable, posed, slumped, atlantic, leeds, west, landing,\n",
      "Epoch 1/10 Iteration: 9100 Avg. Training loss: 4.2797 0.2529 sec/batch\n",
      "Epoch 1/10 Iteration: 9200 Avg. Training loss: 4.2889 0.2504 sec/batch\n",
      "Epoch 1/10 Iteration: 9300 Avg. Training loss: 4.2665 0.2502 sec/batch\n",
      "Epoch 1/10 Iteration: 9400 Avg. Training loss: 4.2259 0.2493 sec/batch\n",
      "Epoch 1/10 Iteration: 9500 Avg. Training loss: 4.2727 0.2499 sec/batch\n",
      "Epoch 1/10 Iteration: 9600 Avg. Training loss: 4.2943 0.2500 sec/batch\n",
      "Epoch 1/10 Iteration: 9700 Avg. Training loss: 4.2746 0.2524 sec/batch\n",
      "Epoch 1/10 Iteration: 9800 Avg. Training loss: 4.2440 0.2688 sec/batch\n",
      "Epoch 1/10 Iteration: 9900 Avg. Training loss: 4.2934 0.2695 sec/batch\n",
      "Epoch 1/10 Iteration: 10000 Avg. Training loss: 4.2747 0.2679 sec/batch\n",
      "Nearest to that: mib, cremated, toughest, suggested, vagabond, gravimeters, retailer, postulated,\n",
      "Nearest to not: does, did, accommodates, nightbreed, but, bros, swinging, do,\n",
      "Nearest to also: see, flattery, hipparchus, pocketing, list, blacklisting, piedmont, irvine,\n",
      "Nearest to he: caller, him, beloved, nihonshoki, legnica, marriage, celeste, maclean,\n",
      "Nearest to all: pollinator, purdue, lydia, serpent, pissarro, monitoring, functions, reappraisal,\n",
      "Nearest to used: commonly, effigies, integers, sony, resisters, screening, frontispiece, andreotti,\n",
      "Nearest to i: am, t, ii, think, grad, sensor, read, barkley,\n",
      "Nearest to when: makeshift, kidman, contemplated, ionian, insufficiently, mitra, dubai, cdot,\n",
      "Nearest to woman: sonnets, tarmac, casbah, corsican, quiescent, telecast, tchad, odrade,\n",
      "Nearest to dr: citation, hexafluoride, thoroughbreds, insurgent, timothy, constantinian, pietro, rightful,\n",
      "Nearest to mainly: iceberg, activate, evangelism, metroid, lingala, spymaster, qwerty, lg,\n",
      "Nearest to scale: blockades, nehemiah, primates, carney, ley, large, rotors, neu,\n",
      "Nearest to test: helens, donors, rajputs, qurra, drops, aberdeen, romero, stolpe,\n",
      "Nearest to lived: parton, marshland, interactivity, slips, eunuch, idaho, informed, haughty,\n",
      "Nearest to road: theodosian, uncredited, reznor, ounces, trotskyists, goldsmiths, reflective, offbeat,\n",
      "Nearest to ocean: schalk, inalienable, posed, atlantic, slumped, landing, leeds, pacific,\n",
      "Epoch 1/10 Iteration: 10100 Avg. Training loss: 4.2529 0.2717 sec/batch\n",
      "Epoch 1/10 Iteration: 10200 Avg. Training loss: 4.2922 0.2690 sec/batch\n",
      "Epoch 1/10 Iteration: 10300 Avg. Training loss: 4.2879 0.2675 sec/batch\n",
      "Epoch 1/10 Iteration: 10400 Avg. Training loss: 4.2366 0.2686 sec/batch\n",
      "Epoch 1/10 Iteration: 10500 Avg. Training loss: 4.1461 0.2695 sec/batch\n",
      "Epoch 1/10 Iteration: 10600 Avg. Training loss: 4.2351 0.2684 sec/batch\n",
      "Epoch 1/10 Iteration: 10700 Avg. Training loss: 4.2254 0.2696 sec/batch\n",
      "Epoch 1/10 Iteration: 10800 Avg. Training loss: 4.1561 0.2682 sec/batch\n",
      "Epoch 1/10 Iteration: 10900 Avg. Training loss: 4.2405 0.2699 sec/batch\n",
      "Epoch 1/10 Iteration: 11000 Avg. Training loss: 4.2264 0.2689 sec/batch\n",
      "Nearest to that: mib, suggested, proposition, postulated, cremated, forgets, gravimeters, ensuring,\n",
      "Nearest to not: does, did, do, accommodates, but, nightbreed, swinging, bros,\n",
      "Nearest to also: see, hipparchus, list, pocketing, dentition, flattery, blacklisting, outweigh,\n",
      "Nearest to he: him, caller, vengeance, nihonshoki, legnica, beloved, marriage, celeste,\n",
      "Nearest to all: pollinator, monitoring, lydia, purdue, reappraisal, crankshaft, functions, serpent,\n",
      "Nearest to used: commonly, effigies, integers, sony, widely, serviced, extensively, resisters,\n",
      "Nearest to i: am, t, ii, grad, escobar, barkley, think, sensor,\n",
      "Nearest to when: makeshift, kidman, contemplated, insufficiently, mitra, ionian, dubai, cdot,\n",
      "Nearest to woman: sonnets, tarmac, corsican, casbah, quiescent, tchad, telecast, revivalist,\n",
      "Nearest to dr: citation, hexafluoride, thoroughbreds, insurgent, timothy, constantinian, mummy, rightful,\n",
      "Nearest to mainly: activate, iceberg, lingala, metroid, evangelism, spymaster, twelver, qwerty,\n",
      "Nearest to scale: blockades, nehemiah, large, primates, carney, irradiation, rotors, ley,\n",
      "Nearest to test: helens, donors, drops, aberdeen, rajputs, qurra, stolpe, romero,\n",
      "Nearest to lived: parton, marshland, idaho, interactivity, informed, wishful, cinematographers, slips,\n",
      "Nearest to road: theodosian, uncredited, reznor, ounces, reflective, trotskyists, goldsmiths, naturalist,\n",
      "Nearest to ocean: schalk, atlantic, inalienable, posed, pacific, landing, slumped, leeds,\n",
      "Epoch 1/10 Iteration: 11100 Avg. Training loss: 4.1807 0.2705 sec/batch\n",
      "Epoch 1/10 Iteration: 11200 Avg. Training loss: 4.3043 0.2694 sec/batch\n",
      "Epoch 1/10 Iteration: 11300 Avg. Training loss: 4.0066 0.2853 sec/batch\n",
      "Epoch 1/10 Iteration: 11400 Avg. Training loss: 3.9001 0.2831 sec/batch\n",
      "Epoch 1/10 Iteration: 11500 Avg. Training loss: 4.2085 0.2746 sec/batch\n",
      "Epoch 1/10 Iteration: 11600 Avg. Training loss: 4.0214 0.2907 sec/batch\n",
      "Epoch 1/10 Iteration: 11700 Avg. Training loss: 4.1272 0.2975 sec/batch\n",
      "Epoch 1/10 Iteration: 11800 Avg. Training loss: 4.1991 0.2904 sec/batch\n",
      "Epoch 1/10 Iteration: 11900 Avg. Training loss: 4.2030 0.2816 sec/batch\n",
      "Epoch 1/10 Iteration: 12000 Avg. Training loss: 4.2609 0.2860 sec/batch\n",
      "Nearest to that: mib, suggested, cremated, solanas, toughest, vagabond, gravimeters, proposition,\n",
      "Nearest to not: does, did, but, accommodates, do, nightbreed, bros, nor,\n",
      "Nearest to also: see, hipparchus, pocketing, list, irvine, blacklisting, flattery, el,\n",
      "Nearest to he: caller, where, nihonshoki, ceau, him, received, sweating, vengeance,\n",
      "Nearest to all: pollinator, lydia, purdue, serpent, pissarro, monorail, heeded, insult,\n",
      "Nearest to used: commonly, effigies, sony, serviced, resisters, grozny, integers, term,\n",
      "Nearest to i: am, t, ii, think, grad, escobar, tell, read,\n",
      "Nearest to when: makeshift, kidman, contemplated, asked, mitra, insufficiently, dubai, ionian,\n",
      "Nearest to woman: sonnets, quiescent, tarmac, corsican, casbah, tchad, revivalist, telecast,\n",
      "Nearest to dr: citation, thoroughbreds, hexafluoride, constantinian, insurgent, mummy, pietro, kovacs,\n",
      "Nearest to mainly: activate, iceberg, spymaster, evangelism, lingala, metroid, qwerty, twelver,\n",
      "Nearest to scale: blockades, large, primates, nehemiah, irradiation, rotors, auctions, carney,\n",
      "Nearest to test: helens, donors, aberdeen, drops, qurra, stolpe, romero, endosperm,\n",
      "Nearest to lived: parton, marshland, interactivity, eunuch, idaho, informed, canonized, cinematographers,\n",
      "Nearest to road: theodosian, uncredited, ounces, reznor, naturalist, browning, reflective, gorky,\n",
      "Nearest to ocean: schalk, inalienable, atlantic, posed, landing, slumped, leeds, southeast,\n",
      "Epoch 1/10 Iteration: 12100 Avg. Training loss: 4.1983 0.2974 sec/batch\n",
      "Epoch 1/10 Iteration: 12200 Avg. Training loss: 4.2858 0.2808 sec/batch\n",
      "Epoch 1/10 Iteration: 12300 Avg. Training loss: 4.2810 0.2815 sec/batch\n",
      "Epoch 1/10 Iteration: 12400 Avg. Training loss: 4.2777 0.3073 sec/batch\n",
      "Epoch 1/10 Iteration: 12500 Avg. Training loss: 4.2535 0.2848 sec/batch\n",
      "Epoch 1/10 Iteration: 12600 Avg. Training loss: 4.2326 0.2800 sec/batch\n",
      "Epoch 1/10 Iteration: 12700 Avg. Training loss: 4.1162 0.2849 sec/batch\n",
      "Epoch 1/10 Iteration: 12800 Avg. Training loss: 4.1787 0.2827 sec/batch\n",
      "Epoch 1/10 Iteration: 12900 Avg. Training loss: 4.2305 0.2844 sec/batch\n",
      "Epoch 1/10 Iteration: 13000 Avg. Training loss: 4.2230 0.2993 sec/batch\n",
      "Nearest to that: mib, proposition, suggested, presuppose, cremated, solanas, promotes, vagabond,\n",
      "Nearest to not: does, did, but, do, accommodates, nightbreed, necessarily, bros,\n",
      "Nearest to also: see, list, hipparchus, piedmont, pocketing, blacklisting, osu, ofc,\n",
      "Nearest to he: caller, ceau, him, vengeance, received, where, nihonshoki, maclean,\n",
      "Nearest to all: pollinator, purdue, crankshaft, invisibility, lydia, reappraisal, pissarro, ketones,\n",
      "Nearest to used: commonly, effigies, extensively, term, sony, integers, serviced, proxies,\n",
      "Nearest to i: am, t, ii, think, grad, escobar, you, hippopotamus,\n",
      "Nearest to when: makeshift, kidman, contemplated, insufficiently, mitra, ionian, dubai, prioritization,\n",
      "Nearest to woman: sonnets, quiescent, tarmac, corsican, casbah, tchad, resident, butyrate,\n",
      "Nearest to dr: citation, thoroughbreds, hexafluoride, constantinian, insurgent, rightful, pietro, mummy,\n",
      "Nearest to mainly: activate, iceberg, twelver, spymaster, evangelism, lingala, metroid, emulsions,\n",
      "Nearest to scale: blockades, large, primates, nehemiah, irradiation, neu, auctions, carney,\n",
      "Nearest to test: helens, aberdeen, drops, donors, stolpe, qurra, romero, endosperm,\n",
      "Nearest to lived: parton, marshland, interactivity, eunuch, idaho, domini, informed, cinematographers,\n",
      "Nearest to road: theodosian, uncredited, ounces, reznor, maneaba, naturalist, browning, reflective,\n",
      "Nearest to ocean: schalk, inalienable, posed, landing, atlantic, slumped, leeds, pacific,\n",
      "Epoch 1/10 Iteration: 13100 Avg. Training loss: 4.2069 0.2841 sec/batch\n",
      "Epoch 1/10 Iteration: 13200 Avg. Training loss: 4.2154 0.2816 sec/batch\n",
      "Epoch 1/10 Iteration: 13300 Avg. Training loss: 4.2455 0.2821 sec/batch\n",
      "Epoch 1/10 Iteration: 13400 Avg. Training loss: 4.2698 0.2826 sec/batch\n",
      "Epoch 1/10 Iteration: 13500 Avg. Training loss: 4.1942 0.2840 sec/batch\n",
      "Epoch 1/10 Iteration: 13600 Avg. Training loss: 4.2323 0.2838 sec/batch\n",
      "Epoch 1/10 Iteration: 13700 Avg. Training loss: 4.0371 0.2842 sec/batch\n",
      "Epoch 1/10 Iteration: 13800 Avg. Training loss: 4.0537 0.2826 sec/batch\n",
      "Epoch 1/10 Iteration: 13900 Avg. Training loss: 4.2033 0.2838 sec/batch\n",
      "Epoch 1/10 Iteration: 14000 Avg. Training loss: 4.1933 0.2832 sec/batch\n",
      "Nearest to that: proposition, mib, suggested, cremated, presuppose, say, argue, retailer,\n",
      "Nearest to not: does, did, but, do, necessarily, nightbreed, accommodates, bros,\n",
      "Nearest to also: see, list, hipparchus, ofc, blacklisting, pocketing, osu, piedmont,\n",
      "Nearest to he: caller, where, him, ceau, tribesmen, immorality, nihonshoki, himself,\n",
      "Nearest to all: pollinator, purdue, lydia, insult, invisibility, almost, pissarro, reappraisal,\n",
      "Nearest to used: commonly, extensively, effigies, serviced, term, integers, sony, widely,\n",
      "Nearest to i: am, ii, think, t, grad, escobar, tell, barkley,\n",
      "Nearest to when: makeshift, kidman, contemplated, insufficiently, mitra, sangre, dubai, asked,\n",
      "Nearest to woman: sonnets, quiescent, tarmac, tchad, corsican, casbah, butyrate, resident,\n",
      "Nearest to dr: citation, insurgent, thoroughbreds, hexafluoride, constantinian, pietro, timothy, mummy,\n",
      "Nearest to mainly: activate, iceberg, twelver, lingala, metroid, spymaster, evangelism, emulsions,\n",
      "Nearest to scale: blockades, large, primates, nehemiah, auctions, irradiation, neu, carney,\n",
      "Nearest to test: helens, drops, endosperm, donors, romero, stolpe, qurra, aberdeen,\n",
      "Nearest to lived: parton, marshland, interactivity, eunuch, domini, idaho, slips, scheming,\n",
      "Nearest to road: uncredited, theodosian, ounces, reznor, kinetochores, naturalist, reflective, browning,\n",
      "Nearest to ocean: schalk, inalienable, atlantic, landing, posed, pacific, slumped, southeast,\n",
      "Epoch 1/10 Iteration: 14100 Avg. Training loss: 4.0244 0.2866 sec/batch\n",
      "Epoch 1/10 Iteration: 14200 Avg. Training loss: 4.2177 0.3129 sec/batch\n",
      "Epoch 1/10 Iteration: 14300 Avg. Training loss: 4.2219 0.2932 sec/batch\n",
      "Epoch 1/10 Iteration: 14400 Avg. Training loss: 4.1863 0.2953 sec/batch\n",
      "Epoch 1/10 Iteration: 14500 Avg. Training loss: 4.1801 0.2834 sec/batch\n",
      "Epoch 1/10 Iteration: 14600 Avg. Training loss: 4.1314 0.2820 sec/batch\n",
      "Epoch 1/10 Iteration: 14700 Avg. Training loss: 4.1638 0.2819 sec/batch\n",
      "Epoch 1/10 Iteration: 14800 Avg. Training loss: 4.2078 0.2836 sec/batch\n",
      "Epoch 1/10 Iteration: 14900 Avg. Training loss: 4.2179 0.2825 sec/batch\n",
      "Epoch 1/10 Iteration: 15000 Avg. Training loss: 4.2702 0.2840 sec/batch\n",
      "Nearest to that: mib, suggested, proposition, cremated, toughest, retailer, gravimeters, say,\n",
      "Nearest to not: does, did, do, but, necessarily, bros, nightbreed, accommodates,\n",
      "Nearest to also: see, list, blacklisting, osu, hipparchus, ofc, irvine, pocketing,\n",
      "Nearest to he: caller, ceau, where, vengeance, celeste, tribesmen, nihonshoki, himself,\n",
      "Nearest to all: pollinator, almost, purdue, crankshaft, lydia, functions, monitoring, invisibility,\n",
      "Nearest to used: commonly, extensively, effigies, term, sony, widely, integers, serviced,\n",
      "Nearest to i: ii, am, t, think, myself, escobar, grad, tell,\n",
      "Nearest to when: makeshift, kidman, contemplated, mitra, insufficiently, dubai, sangre, asked,\n",
      "Nearest to woman: sonnets, quiescent, tarmac, casbah, corsican, butyrate, resident, defensemen,\n",
      "Nearest to dr: citation, hexafluoride, insurgent, thoroughbreds, constantinian, mummy, ehr, kovacs,\n",
      "Nearest to mainly: activate, iceberg, twelver, metroid, lingala, evangelism, emulsions, siais,\n",
      "Nearest to scale: large, blockades, primates, nehemiah, auctions, irradiation, rotors, neu,\n",
      "Nearest to test: helens, drops, endosperm, donors, aberdeen, romero, qurra, stolpe,\n",
      "Nearest to lived: parton, marshland, interactivity, eunuch, shameless, slips, domini, idaho,\n",
      "Nearest to road: theodosian, uncredited, ounces, kinetochores, reflective, reznor, browning, naturalist,\n",
      "Nearest to ocean: schalk, inalienable, landing, posed, atlantic, pacific, slumped, brochures,\n",
      "Epoch 2/10 Iteration: 15100 Avg. Training loss: 4.2163 0.1907 sec/batch\n",
      "Epoch 2/10 Iteration: 15200 Avg. Training loss: 4.1936 0.2834 sec/batch\n",
      "Epoch 2/10 Iteration: 15300 Avg. Training loss: 4.1639 0.2822 sec/batch\n",
      "Epoch 2/10 Iteration: 15400 Avg. Training loss: 4.1891 0.2830 sec/batch\n",
      "Epoch 2/10 Iteration: 15500 Avg. Training loss: 4.2017 0.2827 sec/batch\n",
      "Epoch 2/10 Iteration: 15600 Avg. Training loss: 4.1076 0.2828 sec/batch\n",
      "Epoch 2/10 Iteration: 15700 Avg. Training loss: 4.1362 0.2981 sec/batch\n",
      "Epoch 2/10 Iteration: 15800 Avg. Training loss: 4.2252 0.2820 sec/batch\n",
      "Epoch 2/10 Iteration: 15900 Avg. Training loss: 4.0524 0.2829 sec/batch\n",
      "Epoch 2/10 Iteration: 16000 Avg. Training loss: 4.0802 0.2828 sec/batch\n",
      "Nearest to that: proposition, mib, suggested, cremated, say, solanas, gravimeters, presuppose,\n",
      "Nearest to not: does, did, do, but, necessarily, nightbreed, bros, accommodates,\n",
      "Nearest to also: see, list, pocketing, irvine, piedmont, creamy, blacklisting, osu,\n",
      "Nearest to he: celeste, where, himself, caller, received, vengeance, legnica, ceau,\n",
      "Nearest to all: pollinator, almost, lydia, ketones, invisibility, pissarro, shenanigans, purdue,\n",
      "Nearest to used: commonly, term, extensively, effigies, widely, serviced, grozny, sony,\n",
      "Nearest to i: ii, am, myself, think, escobar, grad, t, tell,\n",
      "Nearest to when: makeshift, kidman, mitra, contemplated, insufficiently, asked, dubai, ionian,\n",
      "Nearest to woman: sonnets, quiescent, tarmac, casbah, corsican, resident, tchad, defensemen,\n",
      "Nearest to dr: citation, hexafluoride, constantinian, thoroughbreds, insurgent, mummy, kovacs, sierpinski,\n",
      "Nearest to mainly: activate, iceberg, twelver, evangelism, metroid, emulsions, spymaster, boasts,\n",
      "Nearest to scale: large, blockades, primates, auctions, rotors, nehemiah, neu, irradiation,\n",
      "Nearest to test: helens, drops, endosperm, aberdeen, donors, romero, stolpe, jinn,\n",
      "Nearest to lived: parton, marshland, shameless, domini, interactivity, eunuch, slips, canonized,\n",
      "Nearest to road: theodosian, uncredited, ounces, kinetochores, browning, reflective, reznor, naturalist,\n",
      "Nearest to ocean: atlantic, schalk, inalienable, landing, posed, pacific, brochures, slumped,\n",
      "Epoch 2/10 Iteration: 16100 Avg. Training loss: 4.0959 0.3112 sec/batch\n",
      "Epoch 2/10 Iteration: 16200 Avg. Training loss: 4.1745 0.2860 sec/batch\n",
      "Epoch 2/10 Iteration: 16300 Avg. Training loss: 4.1608 0.2735 sec/batch\n",
      "Epoch 2/10 Iteration: 16400 Avg. Training loss: 4.1498 0.2779 sec/batch\n",
      "Epoch 2/10 Iteration: 16500 Avg. Training loss: 4.1477 0.2891 sec/batch\n",
      "Epoch 2/10 Iteration: 16600 Avg. Training loss: 4.1500 0.2872 sec/batch\n",
      "Epoch 2/10 Iteration: 16700 Avg. Training loss: 4.1627 0.2841 sec/batch\n",
      "Epoch 2/10 Iteration: 16800 Avg. Training loss: 4.1768 0.3006 sec/batch\n",
      "Epoch 2/10 Iteration: 16900 Avg. Training loss: 4.1725 0.2765 sec/batch\n",
      "Epoch 2/10 Iteration: 17000 Avg. Training loss: 4.1629 0.2685 sec/batch\n",
      "Nearest to that: suggested, proposition, mib, cremated, postulated, argue, promotes, presuppose,\n",
      "Nearest to not: does, did, do, necessarily, but, nightbreed, accommodates, bros,\n",
      "Nearest to also: see, list, pocketing, blacklisting, hipparchus, osu, piedmont, creamy,\n",
      "Nearest to he: where, himself, vengeance, celeste, caller, received, legnica, ceau,\n",
      "Nearest to all: pollinator, almost, powers, lydia, ketones, invisibility, purdue, pissarro,\n",
      "Nearest to used: commonly, extensively, term, effigies, widely, serviced, grozny, remediation,\n",
      "Nearest to i: ii, am, myself, think, escobar, t, tell, nimzowitsch,\n",
      "Nearest to when: makeshift, kidman, contemplated, mitra, insufficiently, asked, sangre, prioritization,\n",
      "Nearest to woman: sonnets, quiescent, tarmac, casbah, corsican, resident, defensemen, butyrate,\n",
      "Nearest to dr: citation, hexafluoride, insurgent, constantinian, kovacs, sierpinski, thoroughbreds, ehr,\n",
      "Nearest to mainly: activate, twelver, iceberg, evangelism, emulsions, boasts, spymaster, metroid,\n",
      "Nearest to scale: large, blockades, primates, neu, auctions, rotors, nehemiah, irradiation,\n",
      "Nearest to test: helens, drops, endosperm, donors, aberdeen, stolpe, computational, jinn,\n",
      "Nearest to lived: parton, marshland, interactivity, eunuch, shameless, domini, slips, scheming,\n",
      "Nearest to road: theodosian, uncredited, kinetochores, ounces, browning, naturalist, reznor, reflective,\n",
      "Nearest to ocean: atlantic, landing, inalienable, schalk, posed, pacific, castilla, slumped,\n",
      "Epoch 2/10 Iteration: 17100 Avg. Training loss: 4.1790 0.2704 sec/batch\n",
      "Epoch 2/10 Iteration: 17200 Avg. Training loss: 4.0789 0.2674 sec/batch\n",
      "Epoch 2/10 Iteration: 17300 Avg. Training loss: 4.0919 0.2674 sec/batch\n",
      "Epoch 2/10 Iteration: 17400 Avg. Training loss: 4.1874 0.2665 sec/batch\n",
      "Epoch 2/10 Iteration: 17500 Avg. Training loss: 4.1831 0.2968 sec/batch\n",
      "Epoch 2/10 Iteration: 17600 Avg. Training loss: 4.1985 0.2824 sec/batch\n",
      "Epoch 2/10 Iteration: 17700 Avg. Training loss: 4.1674 0.2931 sec/batch\n",
      "Epoch 2/10 Iteration: 17800 Avg. Training loss: 4.1843 0.2877 sec/batch\n",
      "Epoch 2/10 Iteration: 17900 Avg. Training loss: 4.1565 0.2812 sec/batch\n",
      "Epoch 2/10 Iteration: 18000 Avg. Training loss: 4.1466 0.2673 sec/batch\n",
      "Nearest to that: suggested, proposition, mib, cremated, postulated, solanas, claimed, say,\n",
      "Nearest to not: does, did, do, necessarily, but, nightbreed, bros, accommodates,\n",
      "Nearest to also: see, list, blacklisting, hipparchus, creamy, muscovite, osu, flattery,\n",
      "Nearest to he: where, himself, ceau, vengeance, caller, celeste, received, legnica,\n",
      "Nearest to all: pollinator, almost, pissarro, lydia, purdue, ketones, powers, monorail,\n",
      "Nearest to used: commonly, extensively, term, effigies, widely, serviced, remediation, grozny,\n",
      "Nearest to i: ii, am, myself, think, t, escobar, tell, you,\n",
      "Nearest to when: makeshift, kidman, contemplated, asked, mitra, insufficiently, sangre, prioritization,\n",
      "Nearest to woman: sonnets, quiescent, tarmac, casbah, butyrate, defensemen, corsican, resident,\n",
      "Nearest to dr: citation, hexafluoride, insurgent, constantinian, kovacs, timothy, thoroughbreds, ehr,\n",
      "Nearest to mainly: activate, twelver, iceberg, emulsions, evangelism, boasts, siais, metroid,\n",
      "Nearest to scale: large, blockades, primates, auctions, irradiation, casualties, advertisements, neu,\n",
      "Nearest to test: helens, drops, endosperm, donors, aberdeen, reposition, hoare, sumo,\n",
      "Nearest to lived: parton, marshland, eunuch, shameless, interactivity, domini, rabindranath, slips,\n",
      "Nearest to road: theodosian, uncredited, ounces, browning, kinetochores, naturalist, reznor, reflective,\n",
      "Nearest to ocean: landing, inalienable, atlantic, schalk, posed, castilla, pacific, brochures,\n",
      "Epoch 2/10 Iteration: 18100 Avg. Training loss: 4.1832 0.2776 sec/batch\n",
      "Epoch 2/10 Iteration: 18200 Avg. Training loss: 4.1888 0.2745 sec/batch\n",
      "Epoch 2/10 Iteration: 18300 Avg. Training loss: 4.1522 0.2802 sec/batch\n",
      "Epoch 2/10 Iteration: 18400 Avg. Training loss: 4.1687 0.2717 sec/batch\n",
      "Epoch 2/10 Iteration: 18500 Avg. Training loss: 4.1366 0.2753 sec/batch\n",
      "Epoch 2/10 Iteration: 18600 Avg. Training loss: 4.1393 0.2751 sec/batch\n",
      "Epoch 2/10 Iteration: 18700 Avg. Training loss: 4.0339 0.2749 sec/batch\n",
      "Epoch 2/10 Iteration: 18800 Avg. Training loss: 4.0390 0.2761 sec/batch\n",
      "Epoch 2/10 Iteration: 18900 Avg. Training loss: 4.1839 0.2771 sec/batch\n",
      "Epoch 2/10 Iteration: 19000 Avg. Training loss: 4.1292 0.2746 sec/batch\n",
      "Nearest to that: proposition, suggested, mib, postulated, argue, promotes, solanas, say,\n",
      "Nearest to not: does, did, necessarily, do, but, nightbreed, accommodates, nor,\n",
      "Nearest to also: see, list, osu, blacklisting, disambiguation, ofc, muscovite, pocketing,\n",
      "Nearest to he: himself, vengeance, where, celeste, caller, ceau, nihonshoki, received,\n",
      "Nearest to all: pollinator, almost, pissarro, purdue, ketones, powers, lydia, srebrenica,\n",
      "Nearest to used: commonly, extensively, term, widely, effigies, refer, serviced, remediation,\n",
      "Nearest to i: ii, am, myself, think, t, hippopotamus, escobar, grad,\n",
      "Nearest to when: makeshift, kidman, contemplated, insufficiently, mitra, asked, dubai, prioritization,\n",
      "Nearest to woman: sonnets, quiescent, butyrate, casbah, tarmac, resident, defensemen, tchad,\n",
      "Nearest to dr: citation, hexafluoride, insurgent, leach, kovacs, constantinian, purported, mummy,\n",
      "Nearest to mainly: activate, twelver, iceberg, evangelism, emulsions, metroid, boasts, helier,\n",
      "Nearest to scale: large, blockades, primates, neu, advertisements, development, rotors, irradiation,\n",
      "Nearest to test: helens, drops, donors, endosperm, aberdeen, hoare, reposition, turntables,\n",
      "Nearest to lived: parton, marshland, shameless, eunuch, domini, slips, interactivity, rabindranath,\n",
      "Nearest to road: theodosian, uncredited, ounces, browning, kinetochores, reflective, offbeat, interscope,\n",
      "Nearest to ocean: atlantic, landing, inalienable, schalk, posed, pacific, castilla, winters,\n",
      "Epoch 2/10 Iteration: 19100 Avg. Training loss: 4.1309 0.2790 sec/batch\n",
      "Epoch 2/10 Iteration: 19200 Avg. Training loss: 4.1229 0.2720 sec/batch\n",
      "Epoch 2/10 Iteration: 19300 Avg. Training loss: 4.1569 0.2691 sec/batch\n",
      "Epoch 2/10 Iteration: 19400 Avg. Training loss: 4.1474 0.2974 sec/batch\n",
      "Epoch 2/10 Iteration: 19500 Avg. Training loss: 4.0875 0.2809 sec/batch\n",
      "Epoch 2/10 Iteration: 19600 Avg. Training loss: 4.1868 0.2680 sec/batch\n",
      "Epoch 2/10 Iteration: 19700 Avg. Training loss: 4.1476 0.2666 sec/batch\n",
      "Epoch 2/10 Iteration: 19800 Avg. Training loss: 4.1722 0.2760 sec/batch\n",
      "Epoch 2/10 Iteration: 19900 Avg. Training loss: 4.1674 0.2717 sec/batch\n",
      "Epoch 2/10 Iteration: 20000 Avg. Training loss: 4.1531 0.2682 sec/batch\n",
      "Nearest to that: proposition, suggested, mib, argue, postulated, say, claimed, solanas,\n",
      "Nearest to not: does, did, necessarily, do, but, nightbreed, bros, nor,\n",
      "Nearest to also: see, list, osu, blacklisting, hipparchus, disambiguation, flattery, muscovite,\n",
      "Nearest to he: himself, where, caller, celeste, ceau, received, nihonshoki, dimitrov,\n",
      "Nearest to all: pollinator, almost, pissarro, lydia, powers, srebrenica, purdue, ketones,\n",
      "Nearest to used: commonly, extensively, term, effigies, widely, refer, serviced, remediation,\n",
      "Nearest to i: ii, am, myself, t, think, tell, grad, hippopotamus,\n",
      "Nearest to when: makeshift, kidman, contemplated, asked, mitra, insufficiently, prioritization, dubai,\n",
      "Nearest to woman: sonnets, quiescent, resident, tarmac, casbah, butyrate, young, corsican,\n",
      "Nearest to dr: citation, insurgent, hexafluoride, leach, kovacs, pixar, emile, mummy,\n",
      "Nearest to mainly: activate, iceberg, evangelism, twelver, emulsions, metroid, boasts, helier,\n",
      "Nearest to scale: large, blockades, primates, neu, rotors, irradiation, auctions, development,\n",
      "Nearest to test: helens, drops, endosperm, donors, aberdeen, reposition, sumo, turntables,\n",
      "Nearest to lived: parton, marshland, domini, slips, shameless, interactivity, simcity, eunuch,\n",
      "Nearest to road: theodosian, uncredited, browning, ounces, kinetochores, interscope, naturalist, reflective,\n",
      "Nearest to ocean: landing, atlantic, inalienable, schalk, posed, castilla, pacific, winters,\n",
      "Epoch 2/10 Iteration: 20100 Avg. Training loss: 4.1533 0.2712 sec/batch\n",
      "Epoch 2/10 Iteration: 20200 Avg. Training loss: 4.1377 0.2672 sec/batch\n",
      "Epoch 2/10 Iteration: 20300 Avg. Training loss: 4.1598 0.2674 sec/batch\n",
      "Epoch 2/10 Iteration: 20400 Avg. Training loss: 4.1832 0.2742 sec/batch\n",
      "Epoch 2/10 Iteration: 20500 Avg. Training loss: 4.1799 0.2749 sec/batch\n",
      "Epoch 2/10 Iteration: 20600 Avg. Training loss: 4.0560 0.2682 sec/batch\n",
      "Epoch 2/10 Iteration: 20700 Avg. Training loss: 4.1562 0.2657 sec/batch\n",
      "Epoch 2/10 Iteration: 20800 Avg. Training loss: 4.0452 0.2510 sec/batch\n",
      "Epoch 2/10 Iteration: 20900 Avg. Training loss: 4.1304 0.2499 sec/batch\n",
      "Epoch 2/10 Iteration: 21000 Avg. Training loss: 4.1654 0.2592 sec/batch\n",
      "Nearest to that: suggested, proposition, argue, mib, claimed, postulated, promotes, say,\n",
      "Nearest to not: does, did, necessarily, do, but, bros, nor, nightbreed,\n",
      "Nearest to also: see, list, blacklisting, osu, hipparchus, flattery, disambiguation, muscovite,\n",
      "Nearest to he: himself, where, caller, nihonshoki, ceau, received, vengeance, celeste,\n",
      "Nearest to all: pollinator, almost, lydia, pissarro, powers, purdue, invisibility, insult,\n",
      "Nearest to used: commonly, extensively, term, refer, effigies, widely, remediation, serviced,\n",
      "Nearest to i: ii, am, myself, think, t, escobar, you, hippopotamus,\n",
      "Nearest to when: makeshift, kidman, contemplated, asked, mitra, insufficiently, prioritization, sangre,\n",
      "Nearest to woman: sonnets, tarmac, quiescent, resident, butyrate, casbah, defensemen, young,\n",
      "Nearest to dr: citation, hexafluoride, insurgent, leach, pixar, kovacs, mummy, fyodorovna,\n",
      "Nearest to mainly: activate, iceberg, evangelism, twelver, emulsions, boasts, analogical, lingala,\n",
      "Nearest to scale: large, blockades, irradiation, primates, neu, auctions, advertisements, development,\n",
      "Nearest to test: helens, donors, drops, endosperm, reposition, aberdeen, sumo, javelin,\n",
      "Nearest to lived: parton, marshland, slips, interactivity, shameless, eunuch, rabindranath, domini,\n",
      "Nearest to road: theodosian, uncredited, interscope, kinetochores, browning, offbeat, naturalist, ounces,\n",
      "Nearest to ocean: landing, atlantic, inalienable, schalk, posed, pacific, castilla, winters,\n",
      "Epoch 2/10 Iteration: 21100 Avg. Training loss: 4.1545 0.2919 sec/batch\n",
      "Epoch 2/10 Iteration: 21200 Avg. Training loss: 4.1488 0.2723 sec/batch\n",
      "Epoch 2/10 Iteration: 21300 Avg. Training loss: 4.1809 0.2590 sec/batch\n",
      "Epoch 2/10 Iteration: 21400 Avg. Training loss: 4.1688 0.2525 sec/batch\n",
      "Epoch 2/10 Iteration: 21500 Avg. Training loss: 4.1168 0.2737 sec/batch\n",
      "Epoch 2/10 Iteration: 21600 Avg. Training loss: 4.0452 0.2657 sec/batch\n",
      "Epoch 2/10 Iteration: 21700 Avg. Training loss: 4.1475 0.2566 sec/batch\n",
      "Epoch 2/10 Iteration: 21800 Avg. Training loss: 4.0981 0.2563 sec/batch\n",
      "Epoch 2/10 Iteration: 21900 Avg. Training loss: 4.1359 0.2637 sec/batch\n",
      "Epoch 2/10 Iteration: 22000 Avg. Training loss: 4.1566 0.2547 sec/batch\n",
      "Nearest to that: suggested, proposition, argue, postulated, presuppose, claimed, fact, promotes,\n",
      "Nearest to not: does, did, necessarily, do, but, nor, nightbreed, bros,\n",
      "Nearest to also: see, list, osu, hipparchus, blacklisting, dentition, disambiguation, pocketing,\n",
      "Nearest to he: ceau, dimitrov, himself, where, nihonshoki, caller, vengeance, celeste,\n",
      "Nearest to all: almost, pollinator, lydia, pissarro, powers, ketones, srebrenica, purdue,\n",
      "Nearest to used: commonly, term, extensively, widely, effigies, refer, remediation, serviced,\n",
      "Nearest to i: ii, am, myself, think, t, escobar, hippopotamus, nimzowitsch,\n",
      "Nearest to when: makeshift, contemplated, kidman, mitra, asked, insufficiently, prioritization, dubai,\n",
      "Nearest to woman: sonnets, butyrate, defensemen, tarmac, resident, young, casbah, quiescent,\n",
      "Nearest to dr: citation, insurgent, hexafluoride, mummy, pixar, leach, kovacs, purported,\n",
      "Nearest to mainly: twelver, evangelism, activate, emulsions, iceberg, boasts, spymaster, lingala,\n",
      "Nearest to scale: large, blockades, neu, rotors, irradiation, casualties, primates, auctions,\n",
      "Nearest to test: helens, donors, endosperm, drops, reposition, aberdeen, hoare, computational,\n",
      "Nearest to lived: parton, marshland, slips, shameless, interactivity, domini, eunuch, milked,\n",
      "Nearest to road: theodosian, interscope, kinetochores, uncredited, browning, offbeat, naturalist, goldsmiths,\n",
      "Nearest to ocean: landing, atlantic, inalienable, schalk, posed, pacific, castilla, brochures,\n",
      "Epoch 2/10 Iteration: 22100 Avg. Training loss: 4.1459 0.2573 sec/batch\n",
      "Epoch 2/10 Iteration: 22200 Avg. Training loss: 4.1123 0.2546 sec/batch\n",
      "Epoch 2/10 Iteration: 22300 Avg. Training loss: 4.1612 0.2561 sec/batch\n",
      "Epoch 2/10 Iteration: 22400 Avg. Training loss: 4.1517 0.2650 sec/batch\n",
      "Epoch 2/10 Iteration: 22500 Avg. Training loss: 4.0852 0.2588 sec/batch\n",
      "Epoch 2/10 Iteration: 22600 Avg. Training loss: 4.1452 0.2556 sec/batch\n",
      "Epoch 2/10 Iteration: 22700 Avg. Training loss: 4.1134 0.2640 sec/batch\n",
      "Epoch 2/10 Iteration: 22800 Avg. Training loss: 4.0841 0.2702 sec/batch\n",
      "Epoch 2/10 Iteration: 22900 Avg. Training loss: 4.0989 0.2714 sec/batch\n",
      "Epoch 2/10 Iteration: 23000 Avg. Training loss: 4.0757 0.2743 sec/batch\n",
      "Nearest to that: suggested, proposition, argue, claimed, mib, postulated, toughest, say,\n",
      "Nearest to not: does, did, necessarily, do, but, bros, nightbreed, nor,\n",
      "Nearest to also: see, list, blacklisting, osu, hipparchus, dentition, muscovite, denslow,\n",
      "Nearest to he: himself, where, ceau, dimitrov, received, vengeance, caller, lecturer,\n",
      "Nearest to all: almost, pollinator, lydia, pissarro, insult, nearly, srebrenica, invisibility,\n",
      "Nearest to used: commonly, extensively, term, widely, refer, effigies, remediation, describe,\n",
      "Nearest to i: ii, am, myself, think, escobar, nimzowitsch, you, tell,\n",
      "Nearest to when: makeshift, kidman, contemplated, mitra, asked, insufficiently, prioritization, sangre,\n",
      "Nearest to woman: sonnets, butyrate, defensemen, resident, tarmac, quiescent, young, casbah,\n",
      "Nearest to dr: citation, insurgent, hexafluoride, pixar, mummy, purported, leach, usability,\n",
      "Nearest to mainly: emulsions, twelver, iceberg, activate, boasts, evangelism, spymaster, lingala,\n",
      "Nearest to scale: large, blockades, neu, auctions, primates, rotors, casualties, irradiation,\n",
      "Nearest to test: helens, donors, aberdeen, endosperm, hoare, reposition, drops, sumo,\n",
      "Nearest to lived: parton, marshland, shameless, milked, eunuch, slips, interactivity, rabindranath,\n",
      "Nearest to road: theodosian, interscope, uncredited, offbeat, browning, kinetochores, naturalist, goldsmiths,\n",
      "Nearest to ocean: atlantic, landing, pacific, schalk, inalienable, posed, castilla, winters,\n",
      "Epoch 2/10 Iteration: 23100 Avg. Training loss: 4.0736 0.2756 sec/batch\n",
      "Epoch 2/10 Iteration: 23200 Avg. Training loss: 4.1464 0.2640 sec/batch\n",
      "Epoch 2/10 Iteration: 23300 Avg. Training loss: 4.1445 0.2637 sec/batch\n",
      "Epoch 2/10 Iteration: 23400 Avg. Training loss: 4.0999 0.2695 sec/batch\n",
      "Epoch 2/10 Iteration: 23500 Avg. Training loss: 4.1380 0.2721 sec/batch\n",
      "Epoch 2/10 Iteration: 23600 Avg. Training loss: 4.0071 0.2682 sec/batch\n",
      "Epoch 2/10 Iteration: 23700 Avg. Training loss: 4.1550 0.2542 sec/batch\n",
      "Epoch 2/10 Iteration: 23800 Avg. Training loss: 4.1318 0.2671 sec/batch\n",
      "Epoch 2/10 Iteration: 23900 Avg. Training loss: 4.1487 0.2797 sec/batch\n",
      "Epoch 2/10 Iteration: 24000 Avg. Training loss: 4.1286 0.2597 sec/batch\n",
      "Nearest to that: suggested, proposition, claimed, postulated, argue, presuppose, fact, say,\n",
      "Nearest to not: does, did, do, necessarily, but, nightbreed, nor, bros,\n",
      "Nearest to also: see, list, blacklisting, osu, dentition, hipparchus, disambiguation, known,\n",
      "Nearest to he: where, himself, ceau, dimitrov, received, attended, caller, wrote,\n",
      "Nearest to all: almost, pollinator, lydia, pissarro, nearly, ketones, purdue, virtually,\n",
      "Nearest to used: commonly, term, extensively, widely, refer, effigies, remediation, describe,\n",
      "Nearest to i: am, ii, think, myself, tell, you, nimzowitsch, wept,\n",
      "Nearest to when: makeshift, mitra, contemplated, kidman, asked, insufficiently, prioritization, sangre,\n",
      "Nearest to woman: sonnets, butyrate, defensemen, tarmac, resident, young, tchad, casbah,\n",
      "Nearest to dr: citation, insurgent, hexafluoride, pixar, mummy, purported, kovacs, leach,\n",
      "Nearest to mainly: emulsions, twelver, iceberg, activate, boasts, evangelism, spymaster, siais,\n",
      "Nearest to scale: large, blockades, primates, neu, auctions, rotors, irradiation, casualties,\n",
      "Nearest to test: helens, donors, aberdeen, endosperm, reposition, hoare, computational, ozone,\n",
      "Nearest to lived: parton, marshland, milked, shameless, domini, eunuch, slips, interactivity,\n",
      "Nearest to road: theodosian, interscope, offbeat, kinetochores, uncredited, browning, ounces, maneaba,\n",
      "Nearest to ocean: atlantic, landing, inalienable, pacific, posed, schalk, castilla, climatic,\n",
      "Epoch 2/10 Iteration: 24100 Avg. Training loss: 4.1579 0.2672 sec/batch\n",
      "Epoch 2/10 Iteration: 24200 Avg. Training loss: 4.1863 0.2656 sec/batch\n",
      "Epoch 2/10 Iteration: 24300 Avg. Training loss: 4.1210 0.2500 sec/batch\n",
      "Epoch 2/10 Iteration: 24400 Avg. Training loss: 4.0632 0.2498 sec/batch\n",
      "Epoch 2/10 Iteration: 24500 Avg. Training loss: 4.1379 0.2499 sec/batch\n",
      "Epoch 2/10 Iteration: 24600 Avg. Training loss: 4.1769 0.2504 sec/batch\n",
      "Epoch 2/10 Iteration: 24700 Avg. Training loss: 4.1321 0.2491 sec/batch\n",
      "Epoch 2/10 Iteration: 24800 Avg. Training loss: 4.1193 0.2496 sec/batch\n",
      "Epoch 2/10 Iteration: 24900 Avg. Training loss: 4.1269 0.2488 sec/batch\n",
      "Epoch 2/10 Iteration: 25000 Avg. Training loss: 4.1575 0.2573 sec/batch\n",
      "Nearest to that: suggested, proposition, claimed, postulated, argue, presuppose, fact, say,\n",
      "Nearest to not: does, did, do, necessarily, but, nor, nightbreed, bros,\n",
      "Nearest to also: see, list, blacklisting, osu, sprague, flattery, hipparchus, disambiguation,\n",
      "Nearest to he: where, himself, dimitrov, ceau, received, nonconformists, lecturer, myrdal,\n",
      "Nearest to all: almost, pollinator, nearly, virtually, lydia, pissarro, purdue, pontryagin,\n",
      "Nearest to used: commonly, extensively, term, widely, refer, describe, effigies, remediation,\n",
      "Nearest to i: ii, am, think, myself, escobar, nimzowitsch, wept, barkley,\n",
      "Nearest to when: makeshift, mitra, kidman, asked, contemplated, prioritization, insufficiently, sangre,\n",
      "Nearest to woman: sonnets, butyrate, defensemen, resident, tarmac, young, tchad, quiescent,\n",
      "Nearest to dr: citation, insurgent, hexafluoride, mummy, purported, pixar, kovacs, leach,\n",
      "Nearest to mainly: emulsions, twelver, iceberg, boasts, activate, evangelism, spymaster, siais,\n",
      "Nearest to scale: large, blockades, neu, primates, irradiation, auctions, rotors, casualties,\n",
      "Nearest to test: helens, donors, reposition, ozone, aberdeen, hoare, endosperm, computational,\n",
      "Nearest to lived: parton, marshland, milked, shameless, slips, eunuch, who, domini,\n",
      "Nearest to road: theodosian, interscope, uncredited, offbeat, browning, naturalist, kinetochores, ounces,\n",
      "Nearest to ocean: atlantic, landing, pacific, posed, castilla, inalienable, schalk, climatic,\n",
      "Epoch 2/10 Iteration: 25100 Avg. Training loss: 4.1369 0.2793 sec/batch\n",
      "Epoch 2/10 Iteration: 25200 Avg. Training loss: 4.1512 0.2617 sec/batch\n",
      "Epoch 2/10 Iteration: 25300 Avg. Training loss: 4.1580 0.2657 sec/batch\n",
      "Epoch 2/10 Iteration: 25400 Avg. Training loss: 4.1333 0.2656 sec/batch\n",
      "Epoch 2/10 Iteration: 25500 Avg. Training loss: 4.0572 0.2531 sec/batch\n",
      "Epoch 2/10 Iteration: 25600 Avg. Training loss: 4.0834 0.2500 sec/batch\n",
      "Epoch 2/10 Iteration: 25700 Avg. Training loss: 4.0974 0.2503 sec/batch\n",
      "Epoch 2/10 Iteration: 25800 Avg. Training loss: 4.1265 0.2499 sec/batch\n",
      "Epoch 2/10 Iteration: 25900 Avg. Training loss: 4.0662 0.2499 sec/batch\n",
      "Epoch 2/10 Iteration: 26000 Avg. Training loss: 4.1029 0.2495 sec/batch\n",
      "Nearest to that: suggested, proposition, argue, claimed, postulated, note, presuppose, fact,\n",
      "Nearest to not: does, did, do, necessarily, but, nor, minkowski, nightbreed,\n",
      "Nearest to also: see, list, blacklisting, osu, disambiguation, dentition, muscovite, hipparchus,\n",
      "Nearest to he: himself, dimitrov, where, ceau, legnica, vengeance, received, myrdal,\n",
      "Nearest to all: almost, pollinator, virtually, nearly, ketones, pontryagin, srebrenica, powers,\n",
      "Nearest to used: commonly, extensively, widely, term, refer, describe, effigies, remediation,\n",
      "Nearest to i: am, ii, myself, think, escobar, nimzowitsch, sensor, remember,\n",
      "Nearest to when: makeshift, kidman, asked, mitra, contemplated, prioritization, insufficiently, sangre,\n",
      "Nearest to woman: sonnets, butyrate, defensemen, resident, tarmac, young, tchad, quiescent,\n",
      "Nearest to dr: citation, insurgent, hexafluoride, mummy, purported, usability, pixar, baronetcy,\n",
      "Nearest to mainly: emulsions, activate, iceberg, twelver, boasts, evangelism, spymaster, siais,\n",
      "Nearest to scale: large, blockades, auctions, primates, neu, rotors, irradiation, casualties,\n",
      "Nearest to test: donors, helens, ozone, hoare, computational, endosperm, aberdeen, sumo,\n",
      "Nearest to lived: parton, marshland, eunuch, milked, domini, rabindranath, slips, shameless,\n",
      "Nearest to road: theodosian, interscope, uncredited, offbeat, kinetochores, browning, reflective, follicle,\n",
      "Nearest to ocean: atlantic, landing, pacific, castilla, posed, inalienable, schalk, climatic,\n",
      "Epoch 2/10 Iteration: 26100 Avg. Training loss: 4.0912 0.2538 sec/batch\n",
      "Epoch 2/10 Iteration: 26200 Avg. Training loss: 4.1774 0.2660 sec/batch\n",
      "Epoch 2/10 Iteration: 26300 Avg. Training loss: 4.0301 0.2655 sec/batch\n",
      "Epoch 2/10 Iteration: 26400 Avg. Training loss: 3.6695 0.2867 sec/batch\n",
      "Epoch 2/10 Iteration: 26500 Avg. Training loss: 4.0456 0.2913 sec/batch\n",
      "Epoch 2/10 Iteration: 26600 Avg. Training loss: 3.9086 0.2650 sec/batch\n",
      "Epoch 2/10 Iteration: 26700 Avg. Training loss: 4.0348 0.2640 sec/batch\n",
      "Epoch 2/10 Iteration: 26800 Avg. Training loss: 4.0708 0.2660 sec/batch\n",
      "Epoch 2/10 Iteration: 26900 Avg. Training loss: 4.0997 0.2664 sec/batch\n",
      "Epoch 2/10 Iteration: 27000 Avg. Training loss: 4.1187 0.2655 sec/batch\n",
      "Nearest to that: claimed, suggested, argue, proposition, fact, argued, say, announced,\n",
      "Nearest to not: does, did, do, necessarily, but, nightbreed, nor, bros,\n",
      "Nearest to also: see, list, blacklisting, known, osu, muscovite, sprague, disambiguation,\n",
      "Nearest to he: where, received, attended, himself, lecturer, ceau, dimitrov, wrote,\n",
      "Nearest to all: almost, pollinator, virtually, lydia, nearly, srebrenica, pissarro, invisibility,\n",
      "Nearest to used: commonly, extensively, term, widely, refer, describe, effigies, remediation,\n",
      "Nearest to i: ii, am, myself, think, wept, remember, nimzowitsch, escobar,\n",
      "Nearest to when: asked, makeshift, kidman, contemplated, mitra, insufficiently, prioritization, quit,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, tarmac, young, tchad, quiescent,\n",
      "Nearest to dr: citation, insurgent, mummy, purported, hexafluoride, kovacs, baronetcy, pixar,\n",
      "Nearest to mainly: emulsions, iceberg, evangelism, activate, twelver, spymaster, boasts, siais,\n",
      "Nearest to scale: large, blockades, primates, neu, irradiation, auctions, rotors, casualties,\n",
      "Nearest to test: helens, donors, ozone, hoare, aberdeen, endosperm, monterey, sumo,\n",
      "Nearest to lived: parton, eunuch, marshland, rabindranath, milked, slips, rural, who,\n",
      "Nearest to road: theodosian, interscope, uncredited, haight, browning, kinetochores, follicle, offbeat,\n",
      "Nearest to ocean: atlantic, landing, castilla, pacific, posed, inalienable, schalk, intersputnik,\n",
      "Epoch 2/10 Iteration: 27100 Avg. Training loss: 4.0934 0.2676 sec/batch\n",
      "Epoch 2/10 Iteration: 27200 Avg. Training loss: 4.1818 0.2642 sec/batch\n",
      "Epoch 2/10 Iteration: 27300 Avg. Training loss: 4.1821 0.2668 sec/batch\n",
      "Epoch 2/10 Iteration: 27400 Avg. Training loss: 4.1840 0.2654 sec/batch\n",
      "Epoch 2/10 Iteration: 27500 Avg. Training loss: 4.1582 0.2803 sec/batch\n",
      "Epoch 2/10 Iteration: 27600 Avg. Training loss: 4.1444 0.2659 sec/batch\n",
      "Epoch 2/10 Iteration: 27700 Avg. Training loss: 4.0978 0.2656 sec/batch\n",
      "Epoch 2/10 Iteration: 27800 Avg. Training loss: 3.9936 0.2652 sec/batch\n",
      "Epoch 2/10 Iteration: 27900 Avg. Training loss: 4.1191 0.2645 sec/batch\n",
      "Epoch 2/10 Iteration: 28000 Avg. Training loss: 4.1440 0.2645 sec/batch\n",
      "Nearest to that: suggested, claimed, proposition, argue, fact, note, presuppose, argued,\n",
      "Nearest to not: does, did, necessarily, do, but, nightbreed, accommodates, minkowski,\n",
      "Nearest to also: see, list, blacklisting, muscovite, known, disambiguation, sprague, osu,\n",
      "Nearest to he: where, received, himself, ceau, attended, sevenfold, lecturer, wrote,\n",
      "Nearest to all: almost, pollinator, virtually, nearly, srebrenica, lydia, pontryagin, ketones,\n",
      "Nearest to used: commonly, extensively, term, describe, widely, refer, effigies, remediation,\n",
      "Nearest to i: am, ii, myself, think, remember, nimzowitsch, escobar, e,\n",
      "Nearest to when: makeshift, asked, kidman, mitra, contemplated, insufficiently, prioritization, jed,\n",
      "Nearest to woman: sonnets, butyrate, resident, tarmac, defensemen, tchad, quiescent, young,\n",
      "Nearest to dr: citation, insurgent, purported, usability, hexafluoride, mummy, baronetcy, rms,\n",
      "Nearest to mainly: emulsions, iceberg, twelver, activate, evangelism, boasts, spymaster, primarily,\n",
      "Nearest to scale: large, blockades, neu, primates, irradiation, auctions, rotors, culpa,\n",
      "Nearest to test: ozone, helens, donors, hoare, aberdeen, desertification, endosperm, ban,\n",
      "Nearest to lived: parton, eunuch, marshland, rabindranath, milked, rural, who, slips,\n",
      "Nearest to road: theodosian, interscope, haight, uncredited, kinetochores, follicle, maneaba, offbeat,\n",
      "Nearest to ocean: atlantic, landing, castilla, posed, pacific, inalienable, schalk, intersputnik,\n",
      "Epoch 2/10 Iteration: 28100 Avg. Training loss: 4.0687 0.2665 sec/batch\n",
      "Epoch 2/10 Iteration: 28200 Avg. Training loss: 4.1584 0.2580 sec/batch\n",
      "Epoch 2/10 Iteration: 28300 Avg. Training loss: 4.0972 0.2652 sec/batch\n",
      "Epoch 2/10 Iteration: 28400 Avg. Training loss: 4.1814 0.2628 sec/batch\n",
      "Epoch 2/10 Iteration: 28500 Avg. Training loss: 4.0813 0.2713 sec/batch\n",
      "Epoch 2/10 Iteration: 28600 Avg. Training loss: 4.1447 0.2643 sec/batch\n",
      "Epoch 2/10 Iteration: 28700 Avg. Training loss: 3.9849 0.2648 sec/batch\n",
      "Epoch 2/10 Iteration: 28800 Avg. Training loss: 3.9148 0.2642 sec/batch\n",
      "Epoch 2/10 Iteration: 28900 Avg. Training loss: 4.1104 0.2657 sec/batch\n",
      "Epoch 2/10 Iteration: 29000 Avg. Training loss: 4.1054 0.2651 sec/batch\n",
      "Nearest to that: claimed, suggested, argue, proposition, fact, say, note, presuppose,\n",
      "Nearest to not: does, did, necessarily, do, but, nightbreed, nor, bros,\n",
      "Nearest to also: see, list, blacklisting, known, muscovite, osu, sprague, ofc,\n",
      "Nearest to he: where, himself, ceau, wrote, dimitrov, received, sevenfold, attended,\n",
      "Nearest to all: almost, pollinator, virtually, nearly, srebrenica, ketones, invisibility, listing,\n",
      "Nearest to used: commonly, extensively, term, describe, widely, refer, effigies, remediation,\n",
      "Nearest to i: ii, am, think, myself, remember, nimzowitsch, escobar, wept,\n",
      "Nearest to when: makeshift, asked, kidman, mitra, contemplated, insufficiently, prioritization, sangre,\n",
      "Nearest to woman: sonnets, butyrate, resident, tarmac, young, defensemen, tchad, quiescent,\n",
      "Nearest to dr: citation, insurgent, purported, usability, baronetcy, rms, hexafluoride, mummy,\n",
      "Nearest to mainly: emulsions, evangelism, twelver, iceberg, activate, boasts, primarily, reinforced,\n",
      "Nearest to scale: large, neu, blockades, irradiation, auctions, primates, rotors, casualties,\n",
      "Nearest to test: ozone, helens, donors, hoare, endosperm, monterey, desertification, aberdeen,\n",
      "Nearest to lived: parton, eunuch, marshland, rabindranath, milked, slips, who, rural,\n",
      "Nearest to road: interscope, theodosian, uncredited, offbeat, haight, maneaba, interconnect, osmium,\n",
      "Nearest to ocean: atlantic, castilla, landing, pacific, intersputnik, inalienable, posed, schalk,\n",
      "Epoch 2/10 Iteration: 29100 Avg. Training loss: 4.0316 0.2805 sec/batch\n",
      "Epoch 2/10 Iteration: 29200 Avg. Training loss: 4.0213 0.2664 sec/batch\n",
      "Epoch 2/10 Iteration: 29300 Avg. Training loss: 4.0866 0.2650 sec/batch\n",
      "Epoch 2/10 Iteration: 29400 Avg. Training loss: 4.1419 0.2646 sec/batch\n",
      "Epoch 2/10 Iteration: 29500 Avg. Training loss: 4.1343 0.2641 sec/batch\n",
      "Epoch 2/10 Iteration: 29600 Avg. Training loss: 4.0361 0.2648 sec/batch\n",
      "Epoch 2/10 Iteration: 29700 Avg. Training loss: 3.9996 0.2646 sec/batch\n",
      "Epoch 2/10 Iteration: 29800 Avg. Training loss: 4.1502 0.2656 sec/batch\n",
      "Epoch 2/10 Iteration: 29900 Avg. Training loss: 4.1038 0.2671 sec/batch\n",
      "Epoch 2/10 Iteration: 30000 Avg. Training loss: 4.1965 0.2627 sec/batch\n",
      "Nearest to that: suggested, claimed, argue, proposition, fact, transinformation, note, say,\n",
      "Nearest to not: did, does, do, necessarily, but, nightbreed, bros, until,\n",
      "Nearest to also: see, list, blacklisting, known, sprague, denslow, muscovite, osu,\n",
      "Nearest to he: where, himself, ceau, attended, wrote, eton, aps, received,\n",
      "Nearest to all: almost, pollinator, virtually, nearly, listing, srebrenica, ketones, pontryagin,\n",
      "Nearest to used: commonly, extensively, term, describe, refer, widely, effigies, remediation,\n",
      "Nearest to i: ii, am, myself, think, escobar, remember, nimzowitsch, e,\n",
      "Nearest to when: makeshift, asked, kidman, mitra, contemplated, prioritization, insufficiently, sangre,\n",
      "Nearest to woman: sonnets, butyrate, resident, young, defensemen, tarmac, quiescent, tchad,\n",
      "Nearest to dr: citation, insurgent, purported, usability, hexafluoride, rms, baronetcy, leach,\n",
      "Nearest to mainly: emulsions, iceberg, evangelism, twelver, activate, boasts, primarily, reinforced,\n",
      "Nearest to scale: large, neu, irradiation, primates, culpa, rotors, auctions, small,\n",
      "Nearest to test: ozone, helens, donors, hoare, aberdeen, monterey, endosperm, sumo,\n",
      "Nearest to lived: parton, eunuch, rabindranath, marshland, slips, who, milked, rural,\n",
      "Nearest to road: interscope, theodosian, uncredited, offbeat, interconnect, haight, browning, reflective,\n",
      "Nearest to ocean: atlantic, castilla, landing, posed, pacific, intersputnik, inalienable, schalk,\n",
      "Epoch 3/10 Iteration: 30100 Avg. Training loss: 4.1345 0.0908 sec/batch\n",
      "Epoch 3/10 Iteration: 30200 Avg. Training loss: 4.1357 0.2639 sec/batch\n",
      "Epoch 3/10 Iteration: 30300 Avg. Training loss: 4.0592 0.2644 sec/batch\n",
      "Epoch 3/10 Iteration: 30400 Avg. Training loss: 4.1263 0.2653 sec/batch\n",
      "Epoch 3/10 Iteration: 30500 Avg. Training loss: 4.1236 0.2651 sec/batch\n",
      "Epoch 3/10 Iteration: 30600 Avg. Training loss: 4.0571 0.2687 sec/batch\n",
      "Epoch 3/10 Iteration: 30700 Avg. Training loss: 4.0812 0.2651 sec/batch\n",
      "Epoch 3/10 Iteration: 30800 Avg. Training loss: 4.0826 0.2640 sec/batch\n",
      "Epoch 3/10 Iteration: 30900 Avg. Training loss: 4.0268 0.2647 sec/batch\n",
      "Epoch 3/10 Iteration: 31000 Avg. Training loss: 4.0243 0.2646 sec/batch\n",
      "Nearest to that: claimed, suggested, proposition, argue, postulated, fact, transinformation, say,\n",
      "Nearest to not: did, does, do, necessarily, but, nightbreed, bros, until,\n",
      "Nearest to also: see, list, known, blacklisting, denslow, sprague, muscovite, hipparchus,\n",
      "Nearest to he: himself, where, wrote, ceau, attended, received, celeste, myrdal,\n",
      "Nearest to all: almost, virtually, pollinator, nearly, listing, srebrenica, lydia, invisibility,\n",
      "Nearest to used: commonly, extensively, term, describe, refer, widely, treat, effigies,\n",
      "Nearest to i: ii, am, myself, think, wept, escobar, remember, nimzowitsch,\n",
      "Nearest to when: makeshift, asked, mitra, kidman, prioritization, insufficiently, contemplated, peerless,\n",
      "Nearest to woman: sonnets, butyrate, resident, tarmac, young, defensemen, quiescent, tchad,\n",
      "Nearest to dr: citation, insurgent, usability, purported, hexafluoride, rms, baronetcy, mummy,\n",
      "Nearest to mainly: emulsions, evangelism, iceberg, activate, boasts, twelver, reinforced, primarily,\n",
      "Nearest to scale: large, neu, culpa, primates, auctions, irradiation, casualties, small,\n",
      "Nearest to test: helens, ozone, aberdeen, hoare, monterey, donors, abort, endosperm,\n",
      "Nearest to lived: parton, eunuch, marshland, who, rabindranath, slips, rural, scheming,\n",
      "Nearest to road: interscope, theodosian, uncredited, haight, offbeat, browning, interconnect, osmium,\n",
      "Nearest to ocean: atlantic, landing, pacific, castilla, posed, inalienable, intersputnik, schalk,\n",
      "Epoch 3/10 Iteration: 31100 Avg. Training loss: 4.0317 0.2692 sec/batch\n",
      "Epoch 3/10 Iteration: 31200 Avg. Training loss: 4.1061 0.2665 sec/batch\n",
      "Epoch 3/10 Iteration: 31300 Avg. Training loss: 4.0586 0.2814 sec/batch\n",
      "Epoch 3/10 Iteration: 31400 Avg. Training loss: 4.0770 0.2748 sec/batch\n",
      "Epoch 3/10 Iteration: 31500 Avg. Training loss: 4.0720 0.2641 sec/batch\n",
      "Epoch 3/10 Iteration: 31600 Avg. Training loss: 4.0823 0.2652 sec/batch\n",
      "Epoch 3/10 Iteration: 31700 Avg. Training loss: 4.1260 0.2646 sec/batch\n",
      "Epoch 3/10 Iteration: 31800 Avg. Training loss: 4.0692 0.2642 sec/batch\n",
      "Epoch 3/10 Iteration: 31900 Avg. Training loss: 4.1340 0.2652 sec/batch\n",
      "Epoch 3/10 Iteration: 32000 Avg. Training loss: 4.0959 0.2653 sec/batch\n",
      "Nearest to that: suggested, claimed, argue, note, postulated, proposition, fact, transinformation,\n",
      "Nearest to not: did, does, do, necessarily, but, nightbreed, nor, bros,\n",
      "Nearest to also: see, list, known, blacklisting, hipparchus, denslow, sprague, references,\n",
      "Nearest to he: where, himself, wrote, ceau, celeste, thoroughbred, disciplinarian, attended,\n",
      "Nearest to all: almost, pollinator, virtually, nearly, listing, pontryagin, invisibility, srebrenica,\n",
      "Nearest to used: commonly, extensively, term, widely, describe, refer, treat, remediation,\n",
      "Nearest to i: ii, am, myself, think, nimzowitsch, wept, escobar, e,\n",
      "Nearest to when: makeshift, asked, prioritization, mitra, kidman, insufficiently, contemplated, sangre,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, tarmac, quiescent, young, tchad,\n",
      "Nearest to dr: citation, insurgent, purported, hexafluoride, usability, sierpinski, rms, baronetcy,\n",
      "Nearest to mainly: emulsions, evangelism, iceberg, boasts, activate, twelver, reinforced, primarily,\n",
      "Nearest to scale: large, neu, culpa, casualties, auctions, toilets, primates, irradiation,\n",
      "Nearest to test: ozone, helens, hoare, donors, abort, aberdeen, endosperm, sumo,\n",
      "Nearest to lived: parton, eunuch, rabindranath, marshland, slips, interactivity, milked, shameless,\n",
      "Nearest to road: interscope, theodosian, uncredited, offbeat, afar, haight, maneaba, browning,\n",
      "Nearest to ocean: atlantic, landing, castilla, posed, pacific, inalienable, intersputnik, climatic,\n",
      "Epoch 3/10 Iteration: 32100 Avg. Training loss: 4.1100 0.2681 sec/batch\n",
      "Epoch 3/10 Iteration: 32200 Avg. Training loss: 4.0820 0.2663 sec/batch\n",
      "Epoch 3/10 Iteration: 32300 Avg. Training loss: 3.9538 0.2650 sec/batch\n",
      "Epoch 3/10 Iteration: 32400 Avg. Training loss: 4.1365 0.2651 sec/batch\n",
      "Epoch 3/10 Iteration: 32500 Avg. Training loss: 4.0921 0.2640 sec/batch\n",
      "Epoch 3/10 Iteration: 32600 Avg. Training loss: 4.1347 0.2640 sec/batch\n",
      "Epoch 3/10 Iteration: 32700 Avg. Training loss: 4.1098 0.2653 sec/batch\n",
      "Epoch 3/10 Iteration: 32800 Avg. Training loss: 4.1210 0.2640 sec/batch\n",
      "Epoch 3/10 Iteration: 32900 Avg. Training loss: 4.0927 0.2764 sec/batch\n",
      "Epoch 3/10 Iteration: 33000 Avg. Training loss: 4.0969 0.2831 sec/batch\n",
      "Nearest to that: suggested, claimed, note, transinformation, argue, proposition, postulated, fact,\n",
      "Nearest to not: did, does, do, necessarily, but, nightbreed, bros, until,\n",
      "Nearest to also: see, list, known, blacklisting, sprague, denslow, hipparchus, muscovite,\n",
      "Nearest to he: where, himself, ceau, aps, thoroughbred, wrote, attended, celeste,\n",
      "Nearest to all: almost, pollinator, virtually, nearly, listing, pontryagin, lydia, invisibility,\n",
      "Nearest to used: commonly, extensively, describe, term, widely, refer, treat, remediation,\n",
      "Nearest to i: ii, am, myself, think, wept, remember, nimzowitsch, my,\n",
      "Nearest to when: asked, makeshift, kidman, contemplated, mitra, prioritization, insufficiently, guessing,\n",
      "Nearest to woman: sonnets, butyrate, resident, young, defensemen, tarmac, quiescent, tchad,\n",
      "Nearest to dr: citation, insurgent, purported, hexafluoride, usability, sierpinski, baronetcy, kovacs,\n",
      "Nearest to mainly: emulsions, evangelism, activate, iceberg, primarily, twelver, reinforced, analogical,\n",
      "Nearest to scale: large, neu, casualties, auctions, culpa, irradiation, toilets, small,\n",
      "Nearest to test: ozone, helens, hoare, donors, aberdeen, ban, abort, monterey,\n",
      "Nearest to lived: parton, eunuch, rabindranath, marshland, slips, who, short, shameless,\n",
      "Nearest to road: interscope, theodosian, uncredited, offbeat, haight, browning, athletics, maneaba,\n",
      "Nearest to ocean: atlantic, castilla, landing, intersputnik, posed, pacific, inalienable, schalk,\n",
      "Epoch 3/10 Iteration: 33100 Avg. Training loss: 4.1034 0.2791 sec/batch\n",
      "Epoch 3/10 Iteration: 33200 Avg. Training loss: 4.1126 0.2659 sec/batch\n",
      "Epoch 3/10 Iteration: 33300 Avg. Training loss: 4.1199 0.2678 sec/batch\n",
      "Epoch 3/10 Iteration: 33400 Avg. Training loss: 4.1197 0.2669 sec/batch\n",
      "Epoch 3/10 Iteration: 33500 Avg. Training loss: 4.0711 0.2680 sec/batch\n",
      "Epoch 3/10 Iteration: 33600 Avg. Training loss: 4.0723 0.2678 sec/batch\n",
      "Epoch 3/10 Iteration: 33700 Avg. Training loss: 4.0159 0.2640 sec/batch\n",
      "Epoch 3/10 Iteration: 33800 Avg. Training loss: 3.9237 0.2655 sec/batch\n",
      "Epoch 3/10 Iteration: 33900 Avg. Training loss: 4.1028 0.2646 sec/batch\n",
      "Epoch 3/10 Iteration: 34000 Avg. Training loss: 4.1166 0.2655 sec/batch\n",
      "Nearest to that: suggested, claimed, argue, note, postulated, proposition, transinformation, fact,\n",
      "Nearest to not: did, does, do, necessarily, but, nightbreed, until, minkowski,\n",
      "Nearest to also: see, list, blacklisting, known, sprague, references, muscovite, denslow,\n",
      "Nearest to he: himself, where, thoroughbred, ceau, wrote, aps, attended, disciplinarian,\n",
      "Nearest to all: almost, virtually, pollinator, nearly, listing, pontryagin, powers, srebrenica,\n",
      "Nearest to used: commonly, extensively, refer, term, describe, widely, treat, remediation,\n",
      "Nearest to i: ii, am, myself, think, e, remember, wept, ferret,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, contemplated, prioritization, insufficiently, guessing,\n",
      "Nearest to woman: sonnets, butyrate, resident, young, defensemen, tarmac, est, gentle,\n",
      "Nearest to dr: citation, insurgent, purported, usability, hexafluoride, baronetcy, sierpinski, rms,\n",
      "Nearest to mainly: emulsions, evangelism, iceberg, primarily, activate, twelver, analogical, reinforced,\n",
      "Nearest to scale: large, neu, casualties, culpa, auctions, toilets, small, blockades,\n",
      "Nearest to test: ozone, helens, hoare, ban, donors, desertification, aberdeen, endosperm,\n",
      "Nearest to lived: parton, eunuch, rabindranath, marshland, slips, short, shameless, milked,\n",
      "Nearest to road: interscope, theodosian, haight, offbeat, uncredited, browning, interconnect, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, castilla, posed, landing, inalienable, indian,\n",
      "Epoch 3/10 Iteration: 34100 Avg. Training loss: 4.0539 0.2688 sec/batch\n",
      "Epoch 3/10 Iteration: 34200 Avg. Training loss: 4.0811 0.2669 sec/batch\n",
      "Epoch 3/10 Iteration: 34300 Avg. Training loss: 4.0852 0.2642 sec/batch\n",
      "Epoch 3/10 Iteration: 34400 Avg. Training loss: 4.0718 0.2653 sec/batch\n",
      "Epoch 3/10 Iteration: 34500 Avg. Training loss: 4.1079 0.2634 sec/batch\n",
      "Epoch 3/10 Iteration: 34600 Avg. Training loss: 3.9959 0.2637 sec/batch\n",
      "Epoch 3/10 Iteration: 34700 Avg. Training loss: 4.1283 0.2651 sec/batch\n",
      "Epoch 3/10 Iteration: 34800 Avg. Training loss: 4.0989 0.2669 sec/batch\n",
      "Epoch 3/10 Iteration: 34900 Avg. Training loss: 4.1173 0.2753 sec/batch\n",
      "Epoch 3/10 Iteration: 35000 Avg. Training loss: 4.1099 0.2847 sec/batch\n",
      "Nearest to that: suggested, claimed, note, argue, transinformation, proposition, fact, postulated,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, nightbreed,\n",
      "Nearest to also: see, list, known, blacklisting, hipparchus, denslow, osu, sprague,\n",
      "Nearest to he: himself, wrote, where, ceau, thoroughbred, celeste, dimitrov, aps,\n",
      "Nearest to all: almost, virtually, pollinator, nearly, listing, pontryagin, ketones, srebrenica,\n",
      "Nearest to used: commonly, extensively, describe, refer, widely, term, treat, remediation,\n",
      "Nearest to i: ii, am, myself, e, think, escobar, nimzowitsch, ferret,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, contemplated, guessing, prioritization, insufficiently,\n",
      "Nearest to woman: sonnets, butyrate, resident, young, defensemen, tarmac, gentle, quiescent,\n",
      "Nearest to dr: citation, insurgent, purported, usability, hexafluoride, kovacs, rms, leach,\n",
      "Nearest to mainly: emulsions, activate, iceberg, evangelism, primarily, analogical, twelver, reinforced,\n",
      "Nearest to scale: large, neu, casualties, culpa, irradiation, auctions, blockades, rotors,\n",
      "Nearest to test: ozone, helens, ban, hoare, donors, desertification, aberdeen, abort,\n",
      "Nearest to lived: parton, eunuch, short, rabindranath, marshland, slips, who, milked,\n",
      "Nearest to road: interscope, theodosian, uncredited, haight, athletics, farrar, browning, offbeat,\n",
      "Nearest to ocean: atlantic, intersputnik, landing, posed, pacific, castilla, inalienable, indian,\n",
      "Epoch 3/10 Iteration: 35100 Avg. Training loss: 4.1062 0.2703 sec/batch\n",
      "Epoch 3/10 Iteration: 35200 Avg. Training loss: 4.1032 0.2668 sec/batch\n",
      "Epoch 3/10 Iteration: 35300 Avg. Training loss: 4.0737 0.2614 sec/batch\n",
      "Epoch 3/10 Iteration: 35400 Avg. Training loss: 4.1340 0.2643 sec/batch\n",
      "Epoch 3/10 Iteration: 35500 Avg. Training loss: 4.1226 0.2666 sec/batch\n",
      "Epoch 3/10 Iteration: 35600 Avg. Training loss: 4.0356 0.2726 sec/batch\n",
      "Epoch 3/10 Iteration: 35700 Avg. Training loss: 4.0575 0.2737 sec/batch\n",
      "Epoch 3/10 Iteration: 35800 Avg. Training loss: 4.0831 0.2662 sec/batch\n",
      "Epoch 3/10 Iteration: 35900 Avg. Training loss: 4.0235 0.2667 sec/batch\n",
      "Epoch 3/10 Iteration: 36000 Avg. Training loss: 4.1126 0.2672 sec/batch\n",
      "Nearest to that: suggested, note, claimed, argue, proposition, transinformation, fact, postulated,\n",
      "Nearest to not: did, does, necessarily, do, but, until, bros, uncommon,\n",
      "Nearest to also: see, list, blacklisting, known, hipparchus, denslow, sprague, muscovite,\n",
      "Nearest to he: himself, wrote, where, aps, thoroughbred, tutti, myrdal, dimitrov,\n",
      "Nearest to all: almost, virtually, nearly, pollinator, listing, pontryagin, pissarro, lydia,\n",
      "Nearest to used: commonly, extensively, describe, refer, term, widely, treat, interchangeably,\n",
      "Nearest to i: ii, am, myself, e, think, wept, world, escobar,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, contemplated, prioritization, guessing, mmu,\n",
      "Nearest to woman: sonnets, butyrate, resident, young, defensemen, tarmac, man, est,\n",
      "Nearest to dr: citation, insurgent, usability, purported, hexafluoride, kovacs, baronetcy, zaman,\n",
      "Nearest to mainly: emulsions, iceberg, evangelism, activate, primarily, reinforced, analogical, twelver,\n",
      "Nearest to scale: large, neu, casualties, culpa, rotors, irradiation, auctions, small,\n",
      "Nearest to test: ozone, ban, helens, donors, desertification, hoare, aberdeen, reposition,\n",
      "Nearest to lived: parton, eunuch, short, rabindranath, marshland, slips, interactivity, milked,\n",
      "Nearest to road: interscope, theodosian, uncredited, athletics, haight, offbeat, farrar, browning,\n",
      "Nearest to ocean: atlantic, intersputnik, landing, posed, pacific, castilla, inalienable, indian,\n",
      "Epoch 3/10 Iteration: 36100 Avg. Training loss: 4.0833 0.2700 sec/batch\n",
      "Epoch 3/10 Iteration: 36200 Avg. Training loss: 4.1174 0.2662 sec/batch\n",
      "Epoch 3/10 Iteration: 36300 Avg. Training loss: 4.1092 0.2655 sec/batch\n",
      "Epoch 3/10 Iteration: 36400 Avg. Training loss: 4.1242 0.2692 sec/batch\n",
      "Epoch 3/10 Iteration: 36500 Avg. Training loss: 4.0647 0.2682 sec/batch\n",
      "Epoch 3/10 Iteration: 36600 Avg. Training loss: 3.9896 0.2771 sec/batch\n",
      "Epoch 3/10 Iteration: 36700 Avg. Training loss: 4.1032 0.2837 sec/batch\n",
      "Epoch 3/10 Iteration: 36800 Avg. Training loss: 4.0372 0.2708 sec/batch\n",
      "Epoch 3/10 Iteration: 36900 Avg. Training loss: 4.0627 0.2713 sec/batch\n",
      "Epoch 3/10 Iteration: 37000 Avg. Training loss: 4.1123 0.2670 sec/batch\n",
      "Nearest to that: suggested, argue, claimed, note, proposition, fact, presuppose, transinformation,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, nor,\n",
      "Nearest to also: see, list, known, blacklisting, hipparchus, references, denslow, sprague,\n",
      "Nearest to he: wrote, where, himself, dimitrov, disciplinarian, myrdal, attended, tutti,\n",
      "Nearest to all: almost, virtually, nearly, pollinator, listing, pontryagin, lydia, pissarro,\n",
      "Nearest to used: commonly, describe, extensively, term, refer, widely, treat, interchangeably,\n",
      "Nearest to i: ii, am, myself, e, think, wept, remember, escobar,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, contemplated, prioritization, guessing, insufficiently,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, tarmac, young, man, gentle,\n",
      "Nearest to dr: citation, usability, purported, insurgent, kovacs, hexafluoride, mummy, baronetcy,\n",
      "Nearest to mainly: emulsions, primarily, evangelism, twelver, boasts, iceberg, activate, reinforced,\n",
      "Nearest to scale: large, neu, casualties, irradiation, culpa, small, auctions, richter,\n",
      "Nearest to test: ozone, ban, desertification, donors, helens, hoare, abort, reposition,\n",
      "Nearest to lived: parton, eunuch, short, rabindranath, marshland, slips, who, milked,\n",
      "Nearest to road: interscope, theodosian, offbeat, haight, athletics, inner, farrar, kinetochores,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, landing, castilla, inalienable, indian,\n",
      "Epoch 3/10 Iteration: 37100 Avg. Training loss: 4.1146 0.2699 sec/batch\n",
      "Epoch 3/10 Iteration: 37200 Avg. Training loss: 4.0642 0.2668 sec/batch\n",
      "Epoch 3/10 Iteration: 37300 Avg. Training loss: 4.1183 0.2659 sec/batch\n",
      "Epoch 3/10 Iteration: 37400 Avg. Training loss: 4.0941 0.2674 sec/batch\n",
      "Epoch 3/10 Iteration: 37500 Avg. Training loss: 4.0137 0.2702 sec/batch\n",
      "Epoch 3/10 Iteration: 37600 Avg. Training loss: 4.0917 0.2820 sec/batch\n",
      "Epoch 3/10 Iteration: 37700 Avg. Training loss: 4.0920 0.2679 sec/batch\n",
      "Epoch 3/10 Iteration: 37800 Avg. Training loss: 3.9956 0.2637 sec/batch\n",
      "Epoch 3/10 Iteration: 37900 Avg. Training loss: 4.0527 0.2647 sec/batch\n",
      "Epoch 3/10 Iteration: 38000 Avg. Training loss: 4.1103 0.2653 sec/batch\n",
      "Nearest to that: claimed, suggested, argue, note, proposition, transinformation, fact, argues,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, nightbreed,\n",
      "Nearest to also: see, list, blacklisting, known, sprague, denslow, hipparchus, references,\n",
      "Nearest to he: himself, where, wrote, dimitrov, thoroughbred, aps, sevenfold, celeste,\n",
      "Nearest to all: almost, virtually, nearly, pollinator, listing, pontryagin, pissarro, lydia,\n",
      "Nearest to used: commonly, extensively, describe, refer, term, widely, treat, interchangeably,\n",
      "Nearest to i: ii, am, myself, think, e, wept, remember, world,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, prioritization, contemplated, guessing, insufficiently,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, young, tarmac, gentle, man,\n",
      "Nearest to dr: citation, usability, purported, insurgent, kovacs, mummy, hexafluoride, leach,\n",
      "Nearest to mainly: emulsions, boasts, primarily, evangelism, iceberg, reinforced, twelver, analogical,\n",
      "Nearest to scale: large, neu, casualties, richter, small, irradiation, culpa, auctions,\n",
      "Nearest to test: ozone, ban, hoare, helens, donors, aberdeen, abort, desertification,\n",
      "Nearest to lived: parton, eunuch, short, rabindranath, slips, marshland, milked, who,\n",
      "Nearest to road: interscope, theodosian, farrar, athletics, offbeat, uncredited, haight, interconnect,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, castilla, landing, inalienable, indian,\n",
      "Epoch 3/10 Iteration: 38100 Avg. Training loss: 3.9393 0.2687 sec/batch\n",
      "Epoch 3/10 Iteration: 38200 Avg. Training loss: 4.1193 0.2644 sec/batch\n",
      "Epoch 3/10 Iteration: 38300 Avg. Training loss: 4.0985 0.2650 sec/batch\n",
      "Epoch 3/10 Iteration: 38400 Avg. Training loss: 4.0755 0.2656 sec/batch\n",
      "Epoch 3/10 Iteration: 38500 Avg. Training loss: 4.1043 0.2669 sec/batch\n",
      "Epoch 3/10 Iteration: 38600 Avg. Training loss: 3.9487 0.2673 sec/batch\n",
      "Epoch 3/10 Iteration: 38700 Avg. Training loss: 4.0792 0.2660 sec/batch\n",
      "Epoch 3/10 Iteration: 38800 Avg. Training loss: 4.0601 0.2635 sec/batch\n",
      "Epoch 3/10 Iteration: 38900 Avg. Training loss: 4.1155 0.2651 sec/batch\n",
      "Epoch 3/10 Iteration: 39000 Avg. Training loss: 4.1012 0.2816 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, fact, transinformation, proposition, presuppose,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, nor,\n",
      "Nearest to also: see, list, known, blacklisting, sprague, references, hipparchus, denslow,\n",
      "Nearest to he: himself, where, wrote, dimitrov, attended, thoroughbred, sevenfold, aps,\n",
      "Nearest to all: almost, virtually, nearly, pollinator, listing, pontryagin, srebrenica, pissarro,\n",
      "Nearest to used: commonly, extensively, describe, refer, widely, term, treat, interchangeably,\n",
      "Nearest to i: ii, am, myself, think, e, wept, remember, world,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, contemplated, prioritization, regularly, guessing,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, est, young, tarmac, gentle,\n",
      "Nearest to dr: citation, usability, insurgent, purported, kovacs, hexafluoride, mummy, zaman,\n",
      "Nearest to mainly: emulsions, primarily, boasts, evangelism, iceberg, analogical, twelver, chronological,\n",
      "Nearest to scale: large, neu, casualties, small, richter, irradiation, culpa, primates,\n",
      "Nearest to test: ozone, ban, helens, hoare, donors, aberdeen, desertification, abort,\n",
      "Nearest to lived: parton, eunuch, short, rabindranath, marshland, milked, who, slips,\n",
      "Nearest to road: interscope, theodosian, offbeat, interconnect, farrar, athletics, traffic, reflective,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, landing, castilla, earth, indian,\n",
      "Epoch 3/10 Iteration: 39100 Avg. Training loss: 4.0896 0.2838 sec/batch\n",
      "Epoch 3/10 Iteration: 39200 Avg. Training loss: 4.1306 0.2629 sec/batch\n",
      "Epoch 3/10 Iteration: 39300 Avg. Training loss: 4.1034 0.2591 sec/batch\n",
      "Epoch 3/10 Iteration: 39400 Avg. Training loss: 4.0657 0.2560 sec/batch\n",
      "Epoch 3/10 Iteration: 39500 Avg. Training loss: 4.0411 0.2539 sec/batch\n",
      "Epoch 3/10 Iteration: 39600 Avg. Training loss: 4.1179 0.2640 sec/batch\n",
      "Epoch 3/10 Iteration: 39700 Avg. Training loss: 4.0959 0.2720 sec/batch\n",
      "Epoch 3/10 Iteration: 39800 Avg. Training loss: 4.1058 0.2638 sec/batch\n",
      "Epoch 3/10 Iteration: 39900 Avg. Training loss: 4.0649 0.2652 sec/batch\n",
      "Epoch 3/10 Iteration: 40000 Avg. Training loss: 4.0955 0.2546 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, fact, transinformation, announced, proposition,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, nor,\n",
      "Nearest to also: see, list, blacklisting, known, sprague, denslow, hipparchus, references,\n",
      "Nearest to he: himself, where, dimitrov, wrote, attended, sevenfold, ceau, thoroughbred,\n",
      "Nearest to all: almost, virtually, nearly, pollinator, listing, pontryagin, srebrenica, invisibility,\n",
      "Nearest to used: commonly, extensively, describe, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, wept, world, escobar,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, prioritization, contemplated, regularly, guessing,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, tarmac, young, gentle, est,\n",
      "Nearest to dr: insurgent, citation, usability, purported, hexafluoride, kovacs, mummy, baronetcy,\n",
      "Nearest to mainly: emulsions, primarily, evangelism, boasts, iceberg, twelver, reinforced, analogical,\n",
      "Nearest to scale: large, neu, casualties, small, irradiation, culpa, richter, primates,\n",
      "Nearest to test: ozone, ban, donors, helens, hoare, desertification, aberdeen, abort,\n",
      "Nearest to lived: parton, short, eunuch, who, rabindranath, slips, marshland, milked,\n",
      "Nearest to road: interscope, theodosian, offbeat, farrar, interconnect, haight, athletics, traffic,\n",
      "Nearest to ocean: atlantic, pacific, intersputnik, posed, castilla, landing, indian, inalienable,\n",
      "Epoch 3/10 Iteration: 40100 Avg. Training loss: 4.1031 0.2521 sec/batch\n",
      "Epoch 3/10 Iteration: 40200 Avg. Training loss: 4.0959 0.2488 sec/batch\n",
      "Epoch 3/10 Iteration: 40300 Avg. Training loss: 4.1132 0.2498 sec/batch\n",
      "Epoch 3/10 Iteration: 40400 Avg. Training loss: 4.1116 0.2689 sec/batch\n",
      "Epoch 3/10 Iteration: 40500 Avg. Training loss: 4.0513 0.2488 sec/batch\n",
      "Epoch 3/10 Iteration: 40600 Avg. Training loss: 3.9931 0.2474 sec/batch\n",
      "Epoch 3/10 Iteration: 40700 Avg. Training loss: 4.0480 0.2480 sec/batch\n",
      "Epoch 3/10 Iteration: 40800 Avg. Training loss: 4.0822 0.2476 sec/batch\n",
      "Epoch 3/10 Iteration: 40900 Avg. Training loss: 4.0152 0.2482 sec/batch\n",
      "Epoch 3/10 Iteration: 41000 Avg. Training loss: 4.0879 0.2486 sec/batch\n",
      "Nearest to that: claimed, note, suggested, argue, fact, proposition, announced, argued,\n",
      "Nearest to not: does, did, do, necessarily, but, until, ratified, uncommon,\n",
      "Nearest to also: see, list, known, blacklisting, muscovite, references, denslow, hipparchus,\n",
      "Nearest to he: himself, dimitrov, wrote, where, attended, thoroughbred, sevenfold, ceau,\n",
      "Nearest to all: almost, virtually, nearly, pollinator, listing, pontryagin, srebrenica, ketones,\n",
      "Nearest to used: commonly, describe, extensively, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, world, ferret, wept,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, contemplated, prioritization, guessing, mistaking,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, est, tarmac, gentle, young,\n",
      "Nearest to dr: citation, insurgent, usability, purported, hexafluoride, kovacs, zaman, mummy,\n",
      "Nearest to mainly: emulsions, primarily, boasts, iceberg, reinforced, evangelism, twelver, activate,\n",
      "Nearest to scale: large, neu, casualties, small, richter, culpa, irradiation, primates,\n",
      "Nearest to test: ozone, ban, helens, donors, hoare, desertification, aberdeen, endosperm,\n",
      "Nearest to lived: parton, short, eunuch, rabindranath, who, marshland, slips, epochs,\n",
      "Nearest to road: interscope, theodosian, traffic, offbeat, athletics, interconnect, reflective, haight,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, castilla, landing, indian, inalienable,\n",
      "Epoch 3/10 Iteration: 41100 Avg. Training loss: 4.0540 0.2516 sec/batch\n",
      "Epoch 3/10 Iteration: 41200 Avg. Training loss: 4.0946 0.2481 sec/batch\n",
      "Epoch 3/10 Iteration: 41300 Avg. Training loss: 4.0434 0.2486 sec/batch\n",
      "Epoch 3/10 Iteration: 41400 Avg. Training loss: 3.7458 0.2480 sec/batch\n",
      "Epoch 3/10 Iteration: 41500 Avg. Training loss: 3.8503 0.2476 sec/batch\n",
      "Epoch 3/10 Iteration: 41600 Avg. Training loss: 3.8983 0.2475 sec/batch\n",
      "Epoch 3/10 Iteration: 41700 Avg. Training loss: 3.9490 0.2484 sec/batch\n",
      "Epoch 3/10 Iteration: 41800 Avg. Training loss: 4.0175 0.2479 sec/batch\n",
      "Epoch 3/10 Iteration: 41900 Avg. Training loss: 4.0833 0.2488 sec/batch\n",
      "Epoch 3/10 Iteration: 42000 Avg. Training loss: 4.1020 0.2489 sec/batch\n",
      "Nearest to that: claimed, suggested, argue, note, announced, fact, asserts, argued,\n",
      "Nearest to not: did, does, do, necessarily, but, until, nor, uncommon,\n",
      "Nearest to also: see, list, blacklisting, known, denslow, muscovite, references, sprague,\n",
      "Nearest to he: where, himself, attended, wrote, thoroughbred, ceau, sevenfold, disciplinarian,\n",
      "Nearest to all: virtually, almost, listing, nearly, pollinator, srebrenica, pontryagin, things,\n",
      "Nearest to used: commonly, describe, extensively, term, refer, widely, interchangeably, treat,\n",
      "Nearest to i: ii, am, myself, think, e, wept, world, ve,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, contemplated, insufficiently, prioritization, regularly,\n",
      "Nearest to woman: sonnets, butyrate, resident, defensemen, tarmac, est, man, young,\n",
      "Nearest to dr: citation, insurgent, purported, usability, kovacs, zaman, hexafluoride, baronetcy,\n",
      "Nearest to mainly: emulsions, evangelism, primarily, reinforced, iceberg, boasts, chronological, analogical,\n",
      "Nearest to scale: large, neu, casualties, small, richter, irradiation, culpa, primates,\n",
      "Nearest to test: ozone, ban, helens, hoare, donors, aberdeen, desertification, monterey,\n",
      "Nearest to lived: short, parton, eunuch, rabindranath, who, slips, milked, marshland,\n",
      "Nearest to road: interscope, theodosian, haight, athletics, offbeat, traffic, farrar, hypocrisy,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, indian, inalienable,\n",
      "Epoch 3/10 Iteration: 42100 Avg. Training loss: 4.0890 0.2516 sec/batch\n",
      "Epoch 3/10 Iteration: 42200 Avg. Training loss: 4.0852 0.2494 sec/batch\n",
      "Epoch 3/10 Iteration: 42300 Avg. Training loss: 4.1375 0.2485 sec/batch\n",
      "Epoch 3/10 Iteration: 42400 Avg. Training loss: 4.1417 0.2490 sec/batch\n",
      "Epoch 3/10 Iteration: 42500 Avg. Training loss: 4.1259 0.2493 sec/batch\n",
      "Epoch 3/10 Iteration: 42600 Avg. Training loss: 4.1092 0.2478 sec/batch\n",
      "Epoch 3/10 Iteration: 42700 Avg. Training loss: 4.0864 0.2490 sec/batch\n",
      "Epoch 3/10 Iteration: 42800 Avg. Training loss: 3.8945 0.2500 sec/batch\n",
      "Epoch 3/10 Iteration: 42900 Avg. Training loss: 4.0952 0.2488 sec/batch\n",
      "Epoch 3/10 Iteration: 43000 Avg. Training loss: 4.1039 0.2477 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, announced, fact, asserts, argues,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, ratified,\n",
      "Nearest to also: see, list, blacklisting, known, muscovite, references, sprague, hipparchus,\n",
      "Nearest to he: himself, where, attended, ceau, wrote, thoroughbred, sevenfold, eton,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, pontryagin, srebrenica, invisibility,\n",
      "Nearest to used: commonly, extensively, describe, refer, term, widely, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, wept, ve, remember,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, prioritization, contemplated, regularly, guessing,\n",
      "Nearest to woman: sonnets, butyrate, resident, est, defensemen, tarmac, nationality, gentle,\n",
      "Nearest to dr: insurgent, citation, purported, usability, hexafluoride, zaman, baronetcy, kovacs,\n",
      "Nearest to mainly: emulsions, primarily, evangelism, reinforced, iceberg, boasts, twelver, analogical,\n",
      "Nearest to scale: large, neu, casualties, small, richter, irradiation, culpa, lv,\n",
      "Nearest to test: ozone, ban, helens, hoare, donors, aberdeen, desertification, abort,\n",
      "Nearest to lived: parton, short, eunuch, rabindranath, who, milked, slips, graveyard,\n",
      "Nearest to road: interscope, haight, theodosian, traffic, athletics, offbeat, farrar, interconnect,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, inalienable, indian,\n",
      "Epoch 3/10 Iteration: 43100 Avg. Training loss: 4.0384 0.2510 sec/batch\n",
      "Epoch 3/10 Iteration: 43200 Avg. Training loss: 4.0887 0.2486 sec/batch\n",
      "Epoch 3/10 Iteration: 43300 Avg. Training loss: 4.0367 0.2481 sec/batch\n",
      "Epoch 3/10 Iteration: 43400 Avg. Training loss: 4.1419 0.2495 sec/batch\n",
      "Epoch 3/10 Iteration: 43500 Avg. Training loss: 4.1029 0.2492 sec/batch\n",
      "Epoch 3/10 Iteration: 43600 Avg. Training loss: 4.0631 0.2485 sec/batch\n",
      "Epoch 3/10 Iteration: 43700 Avg. Training loss: 4.0253 0.2482 sec/batch\n",
      "Epoch 3/10 Iteration: 43800 Avg. Training loss: 3.8903 0.2483 sec/batch\n",
      "Epoch 3/10 Iteration: 43900 Avg. Training loss: 3.9723 0.2475 sec/batch\n",
      "Epoch 3/10 Iteration: 44000 Avg. Training loss: 4.0779 0.2485 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, fact, announced, argues, say,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, nor,\n",
      "Nearest to also: see, list, known, blacklisting, muscovite, references, sprague, osu,\n",
      "Nearest to he: himself, where, wrote, attended, ceau, thoroughbred, aps, sevenfold,\n",
      "Nearest to all: virtually, almost, listing, nearly, pollinator, pontryagin, srebrenica, ketones,\n",
      "Nearest to used: commonly, describe, extensively, refer, widely, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, think, e, myself, world, wept, escobar,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, prioritization, guessing, contemplated, regularly,\n",
      "Nearest to woman: sonnets, butyrate, est, resident, nationality, defensemen, man, tarmac,\n",
      "Nearest to dr: insurgent, purported, citation, usability, hexafluoride, baronetcy, zaman, rms,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, iceberg, evangelism, boasts, twelver, analogical,\n",
      "Nearest to scale: large, neu, casualties, small, richter, irradiation, culpa, lv,\n",
      "Nearest to test: ozone, ban, helens, hoare, donors, desertification, aberdeen, monterey,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, who, slips, marshland, milked,\n",
      "Nearest to road: interscope, theodosian, haight, traffic, offbeat, farrar, interconnect, hypocrisy,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, castilla, posed, landing, indian, inalienable,\n",
      "Epoch 3/10 Iteration: 44100 Avg. Training loss: 4.0809 0.2515 sec/batch\n",
      "Epoch 3/10 Iteration: 44200 Avg. Training loss: 3.9335 0.2465 sec/batch\n",
      "Epoch 3/10 Iteration: 44300 Avg. Training loss: 4.0753 0.2496 sec/batch\n",
      "Epoch 3/10 Iteration: 44400 Avg. Training loss: 4.1063 0.2484 sec/batch\n",
      "Epoch 3/10 Iteration: 44500 Avg. Training loss: 4.0686 0.2485 sec/batch\n",
      "Epoch 3/10 Iteration: 44600 Avg. Training loss: 4.0308 0.2487 sec/batch\n",
      "Epoch 3/10 Iteration: 44700 Avg. Training loss: 4.0176 0.2484 sec/batch\n",
      "Epoch 3/10 Iteration: 44800 Avg. Training loss: 4.0435 0.2488 sec/batch\n",
      "Epoch 3/10 Iteration: 44900 Avg. Training loss: 4.0534 0.2487 sec/batch\n",
      "Epoch 3/10 Iteration: 45000 Avg. Training loss: 4.1337 0.2495 sec/batch\n",
      "Nearest to that: claimed, suggested, note, announced, argue, fact, transinformation, argued,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, nor,\n",
      "Nearest to also: see, list, known, blacklisting, muscovite, sprague, hipparchus, references,\n",
      "Nearest to he: himself, where, attended, ceau, thoroughbred, wrote, aps, eton,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, pontryagin, srebrenica, ketones,\n",
      "Nearest to used: commonly, describe, extensively, term, refer, widely, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, world, escobar, wept,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, guessing, prioritization, regularly, contemplated,\n",
      "Nearest to woman: sonnets, butyrate, resident, est, defensemen, nationality, gentle, man,\n",
      "Nearest to dr: insurgent, purported, citation, usability, hexafluoride, baronetcy, zaman, sierpinski,\n",
      "Nearest to mainly: emulsions, reinforced, primarily, iceberg, evangelism, boasts, twelver, analogical,\n",
      "Nearest to scale: large, neu, richter, casualties, small, irradiation, culpa, tuning,\n",
      "Nearest to test: ozone, ban, helens, hoare, desertification, aberdeen, donors, monterey,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, who, slips, marshland, milked,\n",
      "Nearest to road: interscope, haight, offbeat, interconnect, athletics, hypocrisy, traffic, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, castilla, pacific, posed, landing, inalienable, indian,\n",
      "Epoch 4/10 Iteration: 45100 Avg. Training loss: 4.1173 0.0028 sec/batch\n",
      "Epoch 4/10 Iteration: 45200 Avg. Training loss: 4.1134 0.2488 sec/batch\n",
      "Epoch 4/10 Iteration: 45300 Avg. Training loss: 4.0425 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 45400 Avg. Training loss: 4.0824 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 45500 Avg. Training loss: 4.1029 0.2497 sec/batch\n",
      "Epoch 4/10 Iteration: 45600 Avg. Training loss: 4.0056 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 45700 Avg. Training loss: 4.0511 0.2496 sec/batch\n",
      "Epoch 4/10 Iteration: 45800 Avg. Training loss: 4.0617 0.2483 sec/batch\n",
      "Epoch 4/10 Iteration: 45900 Avg. Training loss: 4.0433 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 46000 Avg. Training loss: 3.9643 0.2488 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, fact, argues, asserts, argued,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, nor,\n",
      "Nearest to also: see, known, list, blacklisting, muscovite, sprague, denslow, hipparchus,\n",
      "Nearest to he: himself, where, wrote, attended, ceau, thoroughbred, nihonshoki, eton,\n",
      "Nearest to all: virtually, almost, listing, nearly, pollinator, pontryagin, srebrenica, invisibility,\n",
      "Nearest to used: describe, commonly, extensively, term, widely, refer, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, think, myself, world, wept, escobar,\n",
      "Nearest to when: asked, makeshift, mitra, kidman, guessing, contemplated, prioritization, arrived,\n",
      "Nearest to woman: sonnets, butyrate, est, resident, defensemen, man, nationality, gentle,\n",
      "Nearest to dr: insurgent, purported, citation, usability, hexafluoride, baronetcy, kovacs, zaman,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, iceberg, boasts, twelver, analogical,\n",
      "Nearest to scale: large, neu, casualties, richter, small, culpa, irradiation, auctions,\n",
      "Nearest to test: ozone, ban, helens, hoare, aberdeen, desertification, obsoletes, monterey,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, who, slips, epochs, marshland,\n",
      "Nearest to road: interscope, traffic, athletics, offbeat, hypocrisy, haight, farrar, interconnect,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, castilla, landing, inalienable, indian,\n",
      "Epoch 4/10 Iteration: 46100 Avg. Training loss: 3.9854 0.2521 sec/batch\n",
      "Epoch 4/10 Iteration: 46200 Avg. Training loss: 4.0417 0.2487 sec/batch\n",
      "Epoch 4/10 Iteration: 46300 Avg. Training loss: 4.0646 0.2483 sec/batch\n",
      "Epoch 4/10 Iteration: 46400 Avg. Training loss: 4.0522 0.2483 sec/batch\n",
      "Epoch 4/10 Iteration: 46500 Avg. Training loss: 4.0434 0.2494 sec/batch\n",
      "Epoch 4/10 Iteration: 46600 Avg. Training loss: 4.0283 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 46700 Avg. Training loss: 4.0846 0.2498 sec/batch\n",
      "Epoch 4/10 Iteration: 46800 Avg. Training loss: 4.0690 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 46900 Avg. Training loss: 4.0717 0.2479 sec/batch\n",
      "Epoch 4/10 Iteration: 47000 Avg. Training loss: 4.0772 0.2492 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, announced, fact, argued, noted,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, nor,\n",
      "Nearest to also: see, list, known, blacklisting, hipparchus, denslow, references, muscovite,\n",
      "Nearest to he: himself, wrote, where, eton, attended, ceau, disciplinarian, aps,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, pontryagin, invisibility, srebrenica,\n",
      "Nearest to used: describe, commonly, extensively, widely, term, refer, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, myself, think, wept, world, escobar,\n",
      "Nearest to when: asked, makeshift, mitra, kidman, prioritization, guessing, contemplated, regularly,\n",
      "Nearest to woman: sonnets, butyrate, est, resident, man, defensemen, gentle, nationality,\n",
      "Nearest to dr: insurgent, citation, usability, purported, hexafluoride, zaman, baronetcy, sierpinski,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, boasts, iceberg, twelver, activate,\n",
      "Nearest to scale: large, casualties, neu, richter, small, culpa, irradiation, auctions,\n",
      "Nearest to test: ozone, ban, hoare, helens, desertification, aberdeen, abort, donors,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, who, epochs, kingdoms, slips,\n",
      "Nearest to road: interscope, hypocrisy, athletics, theodosian, farrar, traffic, inner, haight,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, landing, inalienable, indian,\n",
      "Epoch 4/10 Iteration: 47100 Avg. Training loss: 4.0723 0.2521 sec/batch\n",
      "Epoch 4/10 Iteration: 47200 Avg. Training loss: 4.0800 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 47300 Avg. Training loss: 3.8897 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 47400 Avg. Training loss: 4.1199 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 47500 Avg. Training loss: 4.0678 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 47600 Avg. Training loss: 4.0857 0.2496 sec/batch\n",
      "Epoch 4/10 Iteration: 47700 Avg. Training loss: 4.0891 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 47800 Avg. Training loss: 4.0889 0.2481 sec/batch\n",
      "Epoch 4/10 Iteration: 47900 Avg. Training loss: 4.0870 0.2486 sec/batch\n",
      "Epoch 4/10 Iteration: 48000 Avg. Training loss: 4.0701 0.2495 sec/batch\n",
      "Nearest to that: claimed, note, suggested, argue, announced, transinformation, noted, argues,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, blacklisting, sprague, hipparchus, denslow, references,\n",
      "Nearest to he: himself, where, eton, wrote, attended, disciplinarian, thoroughbred, ceau,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, pontryagin, srebrenica, invisibility,\n",
      "Nearest to used: describe, commonly, extensively, widely, term, refer, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, myself, think, wept, my, world,\n",
      "Nearest to when: asked, makeshift, kidman, mitra, guessing, contemplated, regularly, prioritization,\n",
      "Nearest to woman: sonnets, butyrate, man, est, defensemen, resident, gentle, nationality,\n",
      "Nearest to dr: insurgent, purported, usability, citation, zaman, hexafluoride, baronetcy, kovacs,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, twelver, iceberg, analogical, activate,\n",
      "Nearest to scale: large, casualties, neu, richter, small, culpa, irradiation, auctions,\n",
      "Nearest to test: ozone, ban, hoare, desertification, helens, aberdeen, abort, donors,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, epochs, who, kingdoms, milked,\n",
      "Nearest to road: interscope, athletics, traffic, hypocrisy, inner, osmium, roads, haight,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, landing, inalienable, indian,\n",
      "Epoch 4/10 Iteration: 48100 Avg. Training loss: 4.0363 0.2511 sec/batch\n",
      "Epoch 4/10 Iteration: 48200 Avg. Training loss: 4.1161 0.2487 sec/batch\n",
      "Epoch 4/10 Iteration: 48300 Avg. Training loss: 4.0961 0.2499 sec/batch\n",
      "Epoch 4/10 Iteration: 48400 Avg. Training loss: 4.0800 0.2519 sec/batch\n",
      "Epoch 4/10 Iteration: 48500 Avg. Training loss: 4.0199 0.2478 sec/batch\n",
      "Epoch 4/10 Iteration: 48600 Avg. Training loss: 4.0755 0.2480 sec/batch\n",
      "Epoch 4/10 Iteration: 48700 Avg. Training loss: 4.0353 0.2486 sec/batch\n",
      "Epoch 4/10 Iteration: 48800 Avg. Training loss: 3.8731 0.2477 sec/batch\n",
      "Epoch 4/10 Iteration: 48900 Avg. Training loss: 4.0222 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 49000 Avg. Training loss: 4.0879 0.2484 sec/batch\n",
      "Nearest to that: suggested, claimed, note, argue, argues, fact, announced, transinformation,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, ratified,\n",
      "Nearest to also: see, list, known, blacklisting, references, sprague, muscovite, denslow,\n",
      "Nearest to he: himself, thoroughbred, attended, wrote, disciplinarian, where, ceau, aps,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, pontryagin, srebrenica, ketones,\n",
      "Nearest to used: describe, extensively, commonly, widely, term, refer, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, myself, think, world, wept, ferret,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, prioritization, contemplated, regularly, guessing,\n",
      "Nearest to woman: sonnets, est, butyrate, defensemen, gentle, man, resident, nationality,\n",
      "Nearest to dr: insurgent, purported, citation, hexafluoride, zaman, usability, rms, baronetcy,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, twelver, iceberg, evangelism, analogical, boasts,\n",
      "Nearest to scale: large, casualties, neu, small, richter, culpa, duddy, irradiation,\n",
      "Nearest to test: ozone, ban, desertification, hoare, aberdeen, hazardous, helens, wastes,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, who, kingdoms, epochs, slips,\n",
      "Nearest to road: interscope, traffic, hypocrisy, roads, inner, offbeat, athletics, haight,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, castilla, landing, indian, inalienable,\n",
      "Epoch 4/10 Iteration: 49100 Avg. Training loss: 4.0299 0.2514 sec/batch\n",
      "Epoch 4/10 Iteration: 49200 Avg. Training loss: 4.0771 0.2487 sec/batch\n",
      "Epoch 4/10 Iteration: 49300 Avg. Training loss: 4.0385 0.2480 sec/batch\n",
      "Epoch 4/10 Iteration: 49400 Avg. Training loss: 4.0446 0.2495 sec/batch\n",
      "Epoch 4/10 Iteration: 49500 Avg. Training loss: 4.0646 0.2470 sec/batch\n",
      "Epoch 4/10 Iteration: 49600 Avg. Training loss: 3.9741 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 49700 Avg. Training loss: 4.0624 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 49800 Avg. Training loss: 4.0755 0.2483 sec/batch\n",
      "Epoch 4/10 Iteration: 49900 Avg. Training loss: 4.0840 0.2497 sec/batch\n",
      "Epoch 4/10 Iteration: 50000 Avg. Training loss: 4.0745 0.2486 sec/batch\n",
      "Nearest to that: note, claimed, suggested, argue, announced, fact, argues, noted,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, nor,\n",
      "Nearest to also: see, list, known, blacklisting, sprague, references, denslow, hipparchus,\n",
      "Nearest to he: himself, wrote, thoroughbred, where, attended, ceau, eton, nihonshoki,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, pontryagin, ketones, srebrenica,\n",
      "Nearest to used: commonly, describe, extensively, widely, refer, term, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, myself, think, world, ve, wept,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, guessing, regularly, prioritization, contemplated,\n",
      "Nearest to woman: sonnets, butyrate, man, est, defensemen, resident, gentle, nationality,\n",
      "Nearest to dr: insurgent, purported, usability, citation, hexafluoride, zaman, baronetcy, rms,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, iceberg, twelver, analogical, activate,\n",
      "Nearest to scale: large, casualties, neu, richter, small, culpa, irradiation, duddy,\n",
      "Nearest to test: ozone, ban, desertification, hoare, hazardous, helens, wastes, aberdeen,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, who, slips, epochs,\n",
      "Nearest to road: interscope, hypocrisy, traffic, athletics, offbeat, waterbury, haight, inner,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, landing, indian, inalienable,\n",
      "Epoch 4/10 Iteration: 50100 Avg. Training loss: 4.0723 0.2516 sec/batch\n",
      "Epoch 4/10 Iteration: 50200 Avg. Training loss: 4.0812 0.2489 sec/batch\n",
      "Epoch 4/10 Iteration: 50300 Avg. Training loss: 4.0524 0.2479 sec/batch\n",
      "Epoch 4/10 Iteration: 50400 Avg. Training loss: 4.0968 0.2494 sec/batch\n",
      "Epoch 4/10 Iteration: 50500 Avg. Training loss: 4.1132 0.2488 sec/batch\n",
      "Epoch 4/10 Iteration: 50600 Avg. Training loss: 4.0073 0.2487 sec/batch\n",
      "Epoch 4/10 Iteration: 50700 Avg. Training loss: 4.0403 0.2511 sec/batch\n",
      "Epoch 4/10 Iteration: 50800 Avg. Training loss: 4.0524 0.2489 sec/batch\n",
      "Epoch 4/10 Iteration: 50900 Avg. Training loss: 3.9954 0.2495 sec/batch\n",
      "Epoch 4/10 Iteration: 51000 Avg. Training loss: 4.0712 0.2487 sec/batch\n",
      "Nearest to that: note, claimed, suggested, argue, fact, announced, argues, debasement,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, sprague, denslow, hipparchus,\n",
      "Nearest to he: himself, wrote, thoroughbred, ceau, eton, where, attended, myrdal,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, pontryagin, practically, things,\n",
      "Nearest to used: describe, commonly, extensively, refer, widely, term, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, myself, think, world, my, ve,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, guessing, mistaking, regularly, contemplated,\n",
      "Nearest to woman: sonnets, butyrate, est, man, defensemen, nationality, resident, gentle,\n",
      "Nearest to dr: insurgent, usability, citation, purported, zaman, kovacs, hexafluoride, baronetcy,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, iceberg, twelver, analogical, activate,\n",
      "Nearest to scale: large, casualties, neu, richter, small, culpa, duddy, irradiation,\n",
      "Nearest to test: ozone, ban, desertification, hoare, aberdeen, wastes, hazardous, globes,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, slips, kingdoms, worked, milked,\n",
      "Nearest to road: traffic, interscope, hypocrisy, athletics, offbeat, farrar, inner, waterbury,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, landing, indian, inalienable,\n",
      "Epoch 4/10 Iteration: 51100 Avg. Training loss: 4.0737 0.2508 sec/batch\n",
      "Epoch 4/10 Iteration: 51200 Avg. Training loss: 4.0773 0.2478 sec/batch\n",
      "Epoch 4/10 Iteration: 51300 Avg. Training loss: 4.0691 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 51400 Avg. Training loss: 4.1104 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 51500 Avg. Training loss: 4.0752 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 51600 Avg. Training loss: 4.0089 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 51700 Avg. Training loss: 4.0097 0.2493 sec/batch\n",
      "Epoch 4/10 Iteration: 51800 Avg. Training loss: 4.0546 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 51900 Avg. Training loss: 4.0110 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 52000 Avg. Training loss: 4.0697 0.2490 sec/batch\n",
      "Nearest to that: note, claimed, suggested, argue, argues, fact, noted, suggest,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, references, blacklisting, hipparchus, sprague, denslow,\n",
      "Nearest to he: himself, wrote, where, eton, thoroughbred, dimitrov, ceau, attended,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, pontryagin, practically, things,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, myself, think, world, wept, my,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, guessing, contemplated, prioritization, mistaking,\n",
      "Nearest to woman: sonnets, defensemen, butyrate, man, est, gentle, resident, nationality,\n",
      "Nearest to dr: usability, citation, insurgent, purported, zaman, hexafluoride, rms, kovacs,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, twelver, boasts, iceberg, analogical,\n",
      "Nearest to scale: large, neu, casualties, richter, small, duddy, culpa, irradiation,\n",
      "Nearest to test: ozone, ban, desertification, hoare, wastes, hazardous, globes, donors,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, graveyard, milked, marshland,\n",
      "Nearest to road: interscope, hypocrisy, traffic, athletics, offbeat, farrar, inner, theodosian,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, indian, inalienable,\n",
      "Epoch 4/10 Iteration: 52100 Avg. Training loss: 4.0909 0.2528 sec/batch\n",
      "Epoch 4/10 Iteration: 52200 Avg. Training loss: 4.0681 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 52300 Avg. Training loss: 4.0537 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 52400 Avg. Training loss: 4.0966 0.2488 sec/batch\n",
      "Epoch 4/10 Iteration: 52500 Avg. Training loss: 4.0430 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 52600 Avg. Training loss: 4.0193 0.2493 sec/batch\n",
      "Epoch 4/10 Iteration: 52700 Avg. Training loss: 4.0609 0.2479 sec/batch\n",
      "Epoch 4/10 Iteration: 52800 Avg. Training loss: 3.9804 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 52900 Avg. Training loss: 4.0367 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 53000 Avg. Training loss: 4.0888 0.2491 sec/batch\n",
      "Nearest to that: note, claimed, suggested, argue, argues, announced, noted, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, sprague, hipparchus, denslow,\n",
      "Nearest to he: himself, where, wrote, attended, aps, dimitrov, eton, thoroughbred,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, pontryagin, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, think, myself, world, wept, my,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, guessing, prioritization, regularly, contemplated,\n",
      "Nearest to woman: sonnets, butyrate, man, est, defensemen, gentle, nationality, resident,\n",
      "Nearest to dr: insurgent, citation, usability, purported, zaman, rms, kovacs, baronetcy,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, boasts, twelver, analogical, iceberg,\n",
      "Nearest to scale: large, neu, richter, casualties, small, duddy, culpa, irradiation,\n",
      "Nearest to test: ozone, ban, hoare, desertification, globes, donors, hazardous, aberdeen,\n",
      "Nearest to lived: short, parton, eunuch, rabindranath, kingdoms, milked, epochs, graveyard,\n",
      "Nearest to road: interscope, hypocrisy, athletics, traffic, offbeat, farrar, inner, waterbury,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, indian, intelsat,\n",
      "Epoch 4/10 Iteration: 53100 Avg. Training loss: 3.8854 0.2511 sec/batch\n",
      "Epoch 4/10 Iteration: 53200 Avg. Training loss: 4.0957 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 53300 Avg. Training loss: 4.0957 0.2478 sec/batch\n",
      "Epoch 4/10 Iteration: 53400 Avg. Training loss: 4.0478 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 53500 Avg. Training loss: 4.0277 0.2478 sec/batch\n",
      "Epoch 4/10 Iteration: 53600 Avg. Training loss: 4.0267 0.2493 sec/batch\n",
      "Epoch 4/10 Iteration: 53700 Avg. Training loss: 3.9705 0.2487 sec/batch\n",
      "Epoch 4/10 Iteration: 53800 Avg. Training loss: 4.0695 0.2496 sec/batch\n",
      "Epoch 4/10 Iteration: 53900 Avg. Training loss: 4.0822 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 54000 Avg. Training loss: 4.0648 0.2492 sec/batch\n",
      "Nearest to that: suggested, claimed, note, argue, suggest, noted, fact, argues,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, references, blacklisting, sprague, hipparchus, synonyms,\n",
      "Nearest to he: himself, wrote, attended, eton, where, dimitrov, aps, nihonshoki,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, practically, assorted, things,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, my, world, wept,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, guessing, prioritization, regularly, mistaking,\n",
      "Nearest to woman: sonnets, butyrate, est, man, nationality, gentle, defensemen, resident,\n",
      "Nearest to dr: insurgent, citation, usability, zaman, purported, kovacs, rms, hexafluoride,\n",
      "Nearest to mainly: emulsions, primarily, evangelism, analogical, boasts, twelver, reinforced, coals,\n",
      "Nearest to scale: large, richter, neu, casualties, small, duddy, culpa, irradiation,\n",
      "Nearest to test: ozone, ban, desertification, hoare, hazardous, globes, abort, donors,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, milked, kingdoms, worked, epochs,\n",
      "Nearest to road: hypocrisy, interscope, traffic, offbeat, athletics, farrar, inner, theodosian,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, indian, intelsat,\n",
      "Epoch 4/10 Iteration: 54100 Avg. Training loss: 4.0863 0.2512 sec/batch\n",
      "Epoch 4/10 Iteration: 54200 Avg. Training loss: 4.0908 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 54300 Avg. Training loss: 4.1014 0.2503 sec/batch\n",
      "Epoch 4/10 Iteration: 54400 Avg. Training loss: 4.0554 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 54500 Avg. Training loss: 4.0130 0.2489 sec/batch\n",
      "Epoch 4/10 Iteration: 54600 Avg. Training loss: 4.0623 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 54700 Avg. Training loss: 4.0965 0.2495 sec/batch\n",
      "Epoch 4/10 Iteration: 54800 Avg. Training loss: 4.0682 0.2491 sec/batch\n",
      "Epoch 4/10 Iteration: 54900 Avg. Training loss: 4.0485 0.2478 sec/batch\n",
      "Epoch 4/10 Iteration: 55000 Avg. Training loss: 4.0839 0.2488 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, announced, noted, suggest, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, sprague, hipparchus, denslow,\n",
      "Nearest to he: himself, attended, dimitrov, eton, where, thoroughbred, wrote, ceau,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, practically, pontryagin, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, world, my, wept,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, guessing, prioritization, contemplated, mistaking,\n",
      "Nearest to woman: sonnets, man, butyrate, nationality, defensemen, est, gentle, resident,\n",
      "Nearest to dr: insurgent, citation, usability, zaman, rms, purported, hexafluoride, kovacs,\n",
      "Nearest to mainly: emulsions, primarily, evangelism, analogical, reinforced, twelver, boasts, coals,\n",
      "Nearest to scale: large, neu, richter, casualties, small, duddy, culpa, irradiation,\n",
      "Nearest to test: ban, ozone, desertification, hoare, hazardous, globes, donors, abort,\n",
      "Nearest to lived: short, eunuch, parton, kingdoms, rabindranath, milked, epochs, slips,\n",
      "Nearest to road: hypocrisy, traffic, interscope, farrar, offbeat, theodosian, interconnect, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, indian, intelsat,\n",
      "Epoch 4/10 Iteration: 55100 Avg. Training loss: 4.0909 0.2525 sec/batch\n",
      "Epoch 4/10 Iteration: 55200 Avg. Training loss: 4.0592 0.2486 sec/batch\n",
      "Epoch 4/10 Iteration: 55300 Avg. Training loss: 4.0957 0.2480 sec/batch\n",
      "Epoch 4/10 Iteration: 55400 Avg. Training loss: 4.0691 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 55500 Avg. Training loss: 4.0485 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 55600 Avg. Training loss: 3.9554 0.2486 sec/batch\n",
      "Epoch 4/10 Iteration: 55700 Avg. Training loss: 4.0464 0.2472 sec/batch\n",
      "Epoch 4/10 Iteration: 55800 Avg. Training loss: 4.0469 0.2488 sec/batch\n",
      "Epoch 4/10 Iteration: 55900 Avg. Training loss: 3.9922 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 56000 Avg. Training loss: 4.0607 0.2498 sec/batch\n",
      "Nearest to that: note, claimed, suggested, argue, announced, suggest, noted, argued,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, ratified,\n",
      "Nearest to also: see, list, known, blacklisting, references, external, muscovite, denslow,\n",
      "Nearest to he: himself, dimitrov, attended, wrote, eton, nihonshoki, ceau, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, pontryagin, practically, assorted,\n",
      "Nearest to used: describe, extensively, commonly, refer, widely, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, world, my, ve,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, prioritization, mistaking, contemplated, guessing,\n",
      "Nearest to woman: sonnets, butyrate, man, gentle, est, defensemen, nationality, fertility,\n",
      "Nearest to dr: insurgent, citation, usability, zaman, purported, rms, hexafluoride, mummy,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, boasts, evangelism, analogical, coals, activate,\n",
      "Nearest to scale: large, neu, casualties, richter, small, culpa, duddy, irradiation,\n",
      "Nearest to test: ban, ozone, desertification, hoare, hazardous, globes, donors, dumping,\n",
      "Nearest to lived: short, parton, eunuch, kingdoms, rabindranath, epochs, milked, monumenta,\n",
      "Nearest to road: traffic, hypocrisy, interscope, farrar, roads, athletics, offbeat, haight,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, landing, intelsat,\n",
      "Epoch 4/10 Iteration: 56100 Avg. Training loss: 4.0516 0.2506 sec/batch\n",
      "Epoch 4/10 Iteration: 56200 Avg. Training loss: 4.0180 0.2479 sec/batch\n",
      "Epoch 4/10 Iteration: 56300 Avg. Training loss: 4.1228 0.2496 sec/batch\n",
      "Epoch 4/10 Iteration: 56400 Avg. Training loss: 3.8173 0.2489 sec/batch\n",
      "Epoch 4/10 Iteration: 56500 Avg. Training loss: 3.7321 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 56600 Avg. Training loss: 3.9752 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 56700 Avg. Training loss: 3.8453 0.2481 sec/batch\n",
      "Epoch 4/10 Iteration: 56800 Avg. Training loss: 3.9765 0.2488 sec/batch\n",
      "Epoch 4/10 Iteration: 56900 Avg. Training loss: 4.0341 0.2498 sec/batch\n",
      "Epoch 4/10 Iteration: 57000 Avg. Training loss: 4.0442 0.2495 sec/batch\n",
      "Nearest to that: suggested, claimed, note, announced, argue, suggest, argued, noted,\n",
      "Nearest to not: did, does, do, necessarily, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, denslow, sprague, called,\n",
      "Nearest to he: himself, where, attended, wrote, eton, thoroughbred, sevenfold, nihonshoki,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, assorted, things,\n",
      "Nearest to used: describe, extensively, commonly, refer, widely, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, think, myself, world, my, wept,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, contemplated, arrived, discussing, regularly,\n",
      "Nearest to woman: sonnets, man, butyrate, defensemen, gentle, est, nationality, resident,\n",
      "Nearest to dr: insurgent, citation, zaman, usability, rms, purported, blok, baronetcy,\n",
      "Nearest to mainly: emulsions, primarily, evangelism, boasts, reinforced, analogical, chronological, pagans,\n",
      "Nearest to scale: large, neu, richter, casualties, small, duddy, irradiation, serfdom,\n",
      "Nearest to test: ban, ozone, hoare, desertification, hazardous, globes, aberdeen, wastes,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, milked, worked, epochs,\n",
      "Nearest to road: hypocrisy, traffic, interscope, farrar, haight, roads, offbeat, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, indian, intelsat,\n",
      "Epoch 4/10 Iteration: 57100 Avg. Training loss: 4.1006 0.2515 sec/batch\n",
      "Epoch 4/10 Iteration: 57200 Avg. Training loss: 4.0245 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 57300 Avg. Training loss: 4.1112 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 57400 Avg. Training loss: 4.1283 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 57500 Avg. Training loss: 4.1227 0.2498 sec/batch\n",
      "Epoch 4/10 Iteration: 57600 Avg. Training loss: 4.0910 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 57700 Avg. Training loss: 4.0720 0.2479 sec/batch\n",
      "Epoch 4/10 Iteration: 57800 Avg. Training loss: 3.9307 0.2500 sec/batch\n",
      "Epoch 4/10 Iteration: 57900 Avg. Training loss: 4.0277 0.2487 sec/batch\n",
      "Epoch 4/10 Iteration: 58000 Avg. Training loss: 4.0619 0.2485 sec/batch\n",
      "Nearest to that: suggested, note, claimed, announced, argue, suggest, fact, argues,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, ratified,\n",
      "Nearest to also: see, list, known, blacklisting, references, muscovite, sprague, called,\n",
      "Nearest to he: himself, attended, wrote, where, ceau, eton, thoroughbred, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, pontryagin, srebrenica,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, think, myself, world, my, ve,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, discussing, arrived, guessing, regularly,\n",
      "Nearest to woman: sonnets, man, butyrate, defensemen, gentle, nationality, resident, fertility,\n",
      "Nearest to dr: insurgent, zaman, citation, usability, rms, purported, hexafluoride, blok,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, analogical, boasts, twelver, multisyllabic,\n",
      "Nearest to scale: large, richter, neu, casualties, small, duddy, irradiation, culpa,\n",
      "Nearest to test: ozone, ban, desertification, hazardous, hoare, globes, helens, dumping,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, who, milked, worked,\n",
      "Nearest to road: traffic, hypocrisy, interscope, roads, farrar, haight, athletics, offbeat,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, indian, intelsat,\n",
      "Epoch 4/10 Iteration: 58100 Avg. Training loss: 4.0500 0.2519 sec/batch\n",
      "Epoch 4/10 Iteration: 58200 Avg. Training loss: 4.0331 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 58300 Avg. Training loss: 4.0461 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 58400 Avg. Training loss: 4.0622 0.2485 sec/batch\n",
      "Epoch 4/10 Iteration: 58500 Avg. Training loss: 4.1016 0.2490 sec/batch\n",
      "Epoch 4/10 Iteration: 58600 Avg. Training loss: 4.0330 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 58700 Avg. Training loss: 4.0620 0.2482 sec/batch\n",
      "Epoch 4/10 Iteration: 58800 Avg. Training loss: 3.8983 0.2486 sec/batch\n",
      "Epoch 4/10 Iteration: 58900 Avg. Training loss: 3.8957 0.2480 sec/batch\n",
      "Epoch 4/10 Iteration: 59000 Avg. Training loss: 4.0670 0.2478 sec/batch\n",
      "Nearest to that: claimed, note, suggested, argue, suggest, announced, fact, argues,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, ratified,\n",
      "Nearest to also: see, list, known, blacklisting, references, external, sprague, muscovite,\n",
      "Nearest to he: himself, attended, wrote, where, eton, sevenfold, thoroughbred, ceau,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, pontryagin, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, think, world, my, war,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, arrived, discussing, guessing, regularly,\n",
      "Nearest to woman: sonnets, man, defensemen, butyrate, nationality, gentle, est, children,\n",
      "Nearest to dr: insurgent, citation, zaman, purported, usability, rms, blok, hexafluoride,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, boasts, analogical, twelver, multisyllabic,\n",
      "Nearest to scale: large, richter, neu, casualties, small, irradiation, duddy, lv,\n",
      "Nearest to test: ozone, ban, hazardous, desertification, hoare, wastes, dumping, globes,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, milked, who, epochs,\n",
      "Nearest to road: hypocrisy, traffic, roads, interscope, farrar, offbeat, haight, interconnect,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, indian, landing, intelsat,\n",
      "Epoch 4/10 Iteration: 59100 Avg. Training loss: 4.0428 0.2524 sec/batch\n",
      "Epoch 4/10 Iteration: 59200 Avg. Training loss: 3.8758 0.2484 sec/batch\n",
      "Epoch 4/10 Iteration: 59300 Avg. Training loss: 4.0603 0.2497 sec/batch\n",
      "Epoch 4/10 Iteration: 59400 Avg. Training loss: 4.0803 0.2486 sec/batch\n",
      "Epoch 4/10 Iteration: 59500 Avg. Training loss: 4.0401 0.2477 sec/batch\n",
      "Epoch 4/10 Iteration: 59600 Avg. Training loss: 4.0340 0.2492 sec/batch\n",
      "Epoch 4/10 Iteration: 59700 Avg. Training loss: 3.9905 0.2476 sec/batch\n",
      "Epoch 4/10 Iteration: 59800 Avg. Training loss: 4.0193 0.2483 sec/batch\n",
      "Epoch 4/10 Iteration: 59900 Avg. Training loss: 4.0508 0.2486 sec/batch\n",
      "Epoch 4/10 Iteration: 60000 Avg. Training loss: 4.0852 0.2491 sec/batch\n",
      "Nearest to that: suggested, claimed, note, announced, argue, suggest, noted, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, denslow, external, sprague,\n",
      "Nearest to he: himself, eton, where, attended, wrote, thoroughbred, sevenfold, ceau,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, assorted, pontryagin,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, myself, world, think, war, my,\n",
      "Nearest to when: asked, kidman, makeshift, guessing, mitra, discussing, arrived, mistaking,\n",
      "Nearest to woman: sonnets, man, defensemen, nationality, gentle, butyrate, young, fertility,\n",
      "Nearest to dr: insurgent, usability, citation, purported, rms, zaman, sierpinski, hexafluoride,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, evangelism, boasts, analogical, coals, twelver,\n",
      "Nearest to scale: large, richter, neu, casualties, small, scales, tuning, duddy,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, hoare, wastes, dumping, helens,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, milked, who, worked,\n",
      "Nearest to road: traffic, hypocrisy, roads, interscope, athletics, farrar, interconnect, offbeat,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, landing, indian, intelsat,\n",
      "Epoch 4/10 Iteration: 60100 Avg. Training loss: 4.1241 0.2523 sec/batch\n",
      "Epoch 5/10 Iteration: 60200 Avg. Training loss: 4.0958 0.1697 sec/batch\n",
      "Epoch 5/10 Iteration: 60300 Avg. Training loss: 4.0566 0.2474 sec/batch\n",
      "Epoch 5/10 Iteration: 60400 Avg. Training loss: 4.0157 0.2507 sec/batch\n",
      "Epoch 5/10 Iteration: 60500 Avg. Training loss: 4.0632 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 60600 Avg. Training loss: 4.0581 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 60700 Avg. Training loss: 3.9833 0.2496 sec/batch\n",
      "Epoch 5/10 Iteration: 60800 Avg. Training loss: 4.0213 0.2494 sec/batch\n",
      "Epoch 5/10 Iteration: 60900 Avg. Training loss: 4.0634 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 61000 Avg. Training loss: 3.9507 0.2488 sec/batch\n",
      "Nearest to that: claimed, suggested, note, argue, announced, noted, suggest, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, references, denslow, called, blacklisting, hipparchus,\n",
      "Nearest to he: himself, wrote, attended, eton, where, thoroughbred, ceau, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, assorted, things,\n",
      "Nearest to used: describe, extensively, commonly, widely, term, refer, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, world, myself, think, my, war,\n",
      "Nearest to when: asked, kidman, makeshift, mitra, guessing, arrived, discussing, regularly,\n",
      "Nearest to woman: sonnets, man, defensemen, nationality, gentle, butyrate, young, resident,\n",
      "Nearest to dr: insurgent, usability, citation, sierpinski, rms, zaman, purported, blok,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, boasts, analogical, evangelism, muskets, pagans,\n",
      "Nearest to scale: large, richter, casualties, neu, small, tuning, duddy, scales,\n",
      "Nearest to test: ban, ozone, hoare, hazardous, desertification, abort, cricket, helens,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, milked, epochs, worked,\n",
      "Nearest to road: traffic, hypocrisy, roads, interscope, athletics, farrar, middlesex, interconnect,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, intelsat, indian,\n",
      "Epoch 5/10 Iteration: 61100 Avg. Training loss: 3.9743 0.2521 sec/batch\n",
      "Epoch 5/10 Iteration: 61200 Avg. Training loss: 3.9882 0.2497 sec/batch\n",
      "Epoch 5/10 Iteration: 61300 Avg. Training loss: 4.0613 0.2490 sec/batch\n",
      "Epoch 5/10 Iteration: 61400 Avg. Training loss: 4.0467 0.2478 sec/batch\n",
      "Epoch 5/10 Iteration: 61500 Avg. Training loss: 4.0351 0.2480 sec/batch\n",
      "Epoch 5/10 Iteration: 61600 Avg. Training loss: 4.0237 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 61700 Avg. Training loss: 4.0431 0.2494 sec/batch\n",
      "Epoch 5/10 Iteration: 61800 Avg. Training loss: 4.0565 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 61900 Avg. Training loss: 4.0626 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 62000 Avg. Training loss: 4.0739 0.2495 sec/batch\n",
      "Nearest to that: claimed, suggested, note, announced, argue, noted, suggest, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, denslow, references, blacklisting, called, hipparchus,\n",
      "Nearest to he: himself, where, wrote, thoroughbred, eton, sevenfold, attended, joscelin,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, things, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, term, refer, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, myself, world, think, my, war,\n",
      "Nearest to when: asked, kidman, mitra, makeshift, guessing, prioritization, discussing, arrived,\n",
      "Nearest to woman: sonnets, man, defensemen, gentle, nationality, butyrate, resident, young,\n",
      "Nearest to dr: insurgent, citation, usability, zaman, sierpinski, blok, purported, rms,\n",
      "Nearest to mainly: emulsions, primarily, reinforced, boasts, analogical, evangelism, muskets, twelver,\n",
      "Nearest to scale: large, richter, casualties, neu, small, duddy, scales, culpa,\n",
      "Nearest to test: ozone, ban, hoare, hazardous, desertification, cricket, abort, lunda,\n",
      "Nearest to lived: short, eunuch, parton, kingdoms, rabindranath, worked, milked, epochs,\n",
      "Nearest to road: hypocrisy, roads, traffic, middlesex, farrar, interscope, interstate, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, landing, intelsat, indian,\n",
      "Epoch 5/10 Iteration: 62100 Avg. Training loss: 4.0565 0.2522 sec/batch\n",
      "Epoch 5/10 Iteration: 62200 Avg. Training loss: 4.0567 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 62300 Avg. Training loss: 3.9632 0.2491 sec/batch\n",
      "Epoch 5/10 Iteration: 62400 Avg. Training loss: 3.9877 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 62500 Avg. Training loss: 4.0687 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 62600 Avg. Training loss: 4.0832 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 62700 Avg. Training loss: 4.0752 0.2482 sec/batch\n",
      "Epoch 5/10 Iteration: 62800 Avg. Training loss: 4.0585 0.2510 sec/batch\n",
      "Epoch 5/10 Iteration: 62900 Avg. Training loss: 4.0813 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 63000 Avg. Training loss: 4.0539 0.2484 sec/batch\n",
      "Nearest to that: note, suggested, claimed, announced, argue, noted, suggest, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, blacklisting, denslow, called, references, sprague,\n",
      "Nearest to he: himself, thoroughbred, where, eton, wrote, attended, disciplinarian, orcinus,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, practically, assorted, things,\n",
      "Nearest to used: describe, extensively, commonly, widely, term, refer, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, think, world, myself, my, ve,\n",
      "Nearest to when: asked, kidman, mitra, makeshift, guessing, regularly, mistaking, arrived,\n",
      "Nearest to woman: sonnets, man, defensemen, gentle, nationality, butyrate, fertility, young,\n",
      "Nearest to dr: insurgent, zaman, citation, blok, usability, purported, sierpinski, physician,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, boasts, twelver, muskets, evangelism,\n",
      "Nearest to scale: large, casualties, richter, neu, small, culpa, duddy, scales,\n",
      "Nearest to test: ban, ozone, hoare, hazardous, cricket, matches, desertification, aberdeen,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, worked, epochs, milked,\n",
      "Nearest to road: hypocrisy, roads, traffic, farrar, athletics, inner, interscope, hildebrand,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, intelsat, landing, indian,\n",
      "Epoch 5/10 Iteration: 63100 Avg. Training loss: 4.0450 0.2517 sec/batch\n",
      "Epoch 5/10 Iteration: 63200 Avg. Training loss: 4.0720 0.2497 sec/batch\n",
      "Epoch 5/10 Iteration: 63300 Avg. Training loss: 4.0920 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 63400 Avg. Training loss: 4.0602 0.2477 sec/batch\n",
      "Epoch 5/10 Iteration: 63500 Avg. Training loss: 4.0599 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 63600 Avg. Training loss: 4.0176 0.2494 sec/batch\n",
      "Epoch 5/10 Iteration: 63700 Avg. Training loss: 4.0271 0.2482 sec/batch\n",
      "Epoch 5/10 Iteration: 63800 Avg. Training loss: 3.9155 0.2478 sec/batch\n",
      "Epoch 5/10 Iteration: 63900 Avg. Training loss: 3.9213 0.2496 sec/batch\n",
      "Epoch 5/10 Iteration: 64000 Avg. Training loss: 4.0596 0.2500 sec/batch\n",
      "Nearest to that: note, suggested, claimed, argue, fact, argues, announced, noted,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, called, blacklisting, references, denslow, external,\n",
      "Nearest to he: himself, thoroughbred, wrote, where, eton, attended, sevenfold, ceau,\n",
      "Nearest to all: almost, virtually, nearly, listing, pollinator, practically, things, ketones,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, think, myself, world, my, ve,\n",
      "Nearest to when: asked, kidman, mitra, guessing, makeshift, regularly, discussing, prioritization,\n",
      "Nearest to woman: sonnets, man, defensemen, gentle, nationality, fertility, est, children,\n",
      "Nearest to dr: insurgent, blok, citation, zaman, sierpinski, usability, purported, rms,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, muskets, twelver, evangelism, boasts,\n",
      "Nearest to scale: large, casualties, richter, neu, small, duddy, culpa, economies,\n",
      "Nearest to test: ban, ozone, hazardous, hoare, desertification, cricket, dumping, matches,\n",
      "Nearest to lived: short, eunuch, rabindranath, kingdoms, parton, milked, epochs, worked,\n",
      "Nearest to road: traffic, roads, hypocrisy, interstate, middlesex, inner, farrar, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, landing,\n",
      "Epoch 5/10 Iteration: 64100 Avg. Training loss: 4.0176 0.2509 sec/batch\n",
      "Epoch 5/10 Iteration: 64200 Avg. Training loss: 4.0330 0.2495 sec/batch\n",
      "Epoch 5/10 Iteration: 64300 Avg. Training loss: 4.0107 0.2494 sec/batch\n",
      "Epoch 5/10 Iteration: 64400 Avg. Training loss: 4.0518 0.2492 sec/batch\n",
      "Epoch 5/10 Iteration: 64500 Avg. Training loss: 4.0486 0.2500 sec/batch\n",
      "Epoch 5/10 Iteration: 64600 Avg. Training loss: 3.9208 0.2491 sec/batch\n",
      "Epoch 5/10 Iteration: 64700 Avg. Training loss: 4.0699 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 64800 Avg. Training loss: 4.0445 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 64900 Avg. Training loss: 4.0744 0.2500 sec/batch\n",
      "Epoch 5/10 Iteration: 65000 Avg. Training loss: 4.0591 0.2494 sec/batch\n",
      "Nearest to that: note, suggested, claimed, argue, noted, suggest, announced, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, called, blacklisting, references, denslow, duns,\n",
      "Nearest to he: himself, thoroughbred, wrote, sevenfold, eton, where, ceau, orcinus,\n",
      "Nearest to all: virtually, almost, nearly, listing, pollinator, practically, things, pontryagin,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, think, world, myself, my, ve,\n",
      "Nearest to when: asked, kidman, mitra, guessing, makeshift, regularly, mistaking, arrived,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, est, fertility, butyrate,\n",
      "Nearest to dr: insurgent, citation, blok, purported, zaman, rms, usability, hexafluoride,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, muskets, multisyllabic, evangelism, boasts,\n",
      "Nearest to scale: large, casualties, richter, neu, duddy, small, culpa, conducive,\n",
      "Nearest to test: ban, ozone, hazardous, cricket, desertification, hoare, dumping, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, kingdoms, worked, milked, graveyard,\n",
      "Nearest to road: hypocrisy, traffic, roads, middlesex, farrar, athletics, interstate, waterbury,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, intelsat, indian, landing,\n",
      "Epoch 5/10 Iteration: 65100 Avg. Training loss: 4.0609 0.2530 sec/batch\n",
      "Epoch 5/10 Iteration: 65200 Avg. Training loss: 4.0527 0.2494 sec/batch\n",
      "Epoch 5/10 Iteration: 65300 Avg. Training loss: 4.0547 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 65400 Avg. Training loss: 4.0536 0.2487 sec/batch\n",
      "Epoch 5/10 Iteration: 65500 Avg. Training loss: 4.0964 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 65600 Avg. Training loss: 4.0783 0.2491 sec/batch\n",
      "Epoch 5/10 Iteration: 65700 Avg. Training loss: 3.9674 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 65800 Avg. Training loss: 4.0520 0.2487 sec/batch\n",
      "Epoch 5/10 Iteration: 65900 Avg. Training loss: 3.9646 0.2491 sec/batch\n",
      "Epoch 5/10 Iteration: 66000 Avg. Training loss: 4.0458 0.2483 sec/batch\n",
      "Nearest to that: note, suggested, claimed, fact, argue, noted, announced, argues,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, denslow, called, external,\n",
      "Nearest to he: himself, thoroughbred, wrote, sevenfold, eton, where, ceau, tutti,\n",
      "Nearest to all: almost, virtually, nearly, listing, practically, pollinator, things, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, think, world, myself, ve, my,\n",
      "Nearest to when: asked, kidman, mitra, guessing, makeshift, regularly, mistaking, especially,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, fertility, butyrate, children,\n",
      "Nearest to dr: insurgent, citation, rms, blok, usability, zaman, purported, physician,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, muskets, analogical, multisyllabic, evangelism, pagans,\n",
      "Nearest to scale: large, richter, casualties, neu, duddy, small, irradiation, culpa,\n",
      "Nearest to test: ban, ozone, desertification, hazardous, cricket, hoare, dumping, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, kingdoms, worked, milked, graveyard,\n",
      "Nearest to road: hypocrisy, traffic, roads, middlesex, farrar, athletics, hildebrand, inner,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, landing, intelsat,\n",
      "Epoch 5/10 Iteration: 66100 Avg. Training loss: 4.0632 0.2518 sec/batch\n",
      "Epoch 5/10 Iteration: 66200 Avg. Training loss: 4.0510 0.2476 sec/batch\n",
      "Epoch 5/10 Iteration: 66300 Avg. Training loss: 4.0657 0.2480 sec/batch\n",
      "Epoch 5/10 Iteration: 66400 Avg. Training loss: 4.0890 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 66500 Avg. Training loss: 4.0860 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 66600 Avg. Training loss: 4.0264 0.2480 sec/batch\n",
      "Epoch 5/10 Iteration: 66700 Avg. Training loss: 3.9490 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 66800 Avg. Training loss: 4.0741 0.2480 sec/batch\n",
      "Epoch 5/10 Iteration: 66900 Avg. Training loss: 3.9805 0.2496 sec/batch\n",
      "Epoch 5/10 Iteration: 67000 Avg. Training loss: 4.0455 0.2493 sec/batch\n",
      "Nearest to that: note, suggested, noted, claimed, fact, argues, argue, announced,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, references, blacklisting, called, denslow, daiko,\n",
      "Nearest to he: himself, wrote, thoroughbred, where, eton, sevenfold, tutti, favorably,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, pollinator, pontryagin,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, treat, interchangeably,\n",
      "Nearest to i: ii, e, am, think, myself, world, ve, my,\n",
      "Nearest to when: asked, kidman, mitra, guessing, makeshift, discussing, contemplated, arrived,\n",
      "Nearest to woman: sonnets, man, nationality, gentle, defensemen, fertility, est, butyrate,\n",
      "Nearest to dr: citation, insurgent, usability, blok, rms, zaman, purported, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, multisyllabic, analogical, muskets, evangelism, boasts,\n",
      "Nearest to scale: large, richter, casualties, neu, duddy, small, irradiation, culpa,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, cricket, dumping, hoare, wastes,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, kingdoms, worked, milked, graveyard,\n",
      "Nearest to road: hypocrisy, traffic, roads, middlesex, farrar, athletics, interstate, interscope,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, indian, intelsat, landing,\n",
      "Epoch 5/10 Iteration: 67100 Avg. Training loss: 4.0670 0.2519 sec/batch\n",
      "Epoch 5/10 Iteration: 67200 Avg. Training loss: 4.0486 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 67300 Avg. Training loss: 4.0198 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 67400 Avg. Training loss: 4.0812 0.2483 sec/batch\n",
      "Epoch 5/10 Iteration: 67500 Avg. Training loss: 4.0391 0.2490 sec/batch\n",
      "Epoch 5/10 Iteration: 67600 Avg. Training loss: 3.9925 0.2487 sec/batch\n",
      "Epoch 5/10 Iteration: 67700 Avg. Training loss: 4.0514 0.2476 sec/batch\n",
      "Epoch 5/10 Iteration: 67800 Avg. Training loss: 4.0110 0.2483 sec/batch\n",
      "Epoch 5/10 Iteration: 67900 Avg. Training loss: 4.0090 0.2490 sec/batch\n",
      "Epoch 5/10 Iteration: 68000 Avg. Training loss: 4.0135 0.2485 sec/batch\n",
      "Nearest to that: note, suggested, claimed, noted, announced, argues, fact, argued,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, called, denslow, sprague,\n",
      "Nearest to he: himself, where, thoroughbred, wrote, sevenfold, aps, discontented, joscelin,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, pollinator, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, think, world, myself, ve, my,\n",
      "Nearest to when: asked, kidman, mitra, makeshift, guessing, arrived, discussing, regularly,\n",
      "Nearest to woman: sonnets, man, nationality, gentle, defensemen, fertility, est, butyrate,\n",
      "Nearest to dr: citation, insurgent, blok, usability, rms, zaman, purported, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, evangelism, boasts, muskets, multisyllabic,\n",
      "Nearest to scale: large, richter, casualties, neu, duddy, small, scales, culpa,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, cricket, hoare, globes, tests,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, milked, kingdoms, worked, graveyard,\n",
      "Nearest to road: hypocrisy, traffic, farrar, athletics, hildebrand, roads, interstate, middlesex,\n",
      "Nearest to ocean: intersputnik, atlantic, posed, castilla, pacific, indian, landing, intelsat,\n",
      "Epoch 5/10 Iteration: 68100 Avg. Training loss: 3.9498 0.2526 sec/batch\n",
      "Epoch 5/10 Iteration: 68200 Avg. Training loss: 3.9952 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 68300 Avg. Training loss: 4.0770 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 68400 Avg. Training loss: 4.0559 0.2496 sec/batch\n",
      "Epoch 5/10 Iteration: 68500 Avg. Training loss: 4.0104 0.2490 sec/batch\n",
      "Epoch 5/10 Iteration: 68600 Avg. Training loss: 4.0370 0.2493 sec/batch\n",
      "Epoch 5/10 Iteration: 68700 Avg. Training loss: 3.9167 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 68800 Avg. Training loss: 4.0823 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 68900 Avg. Training loss: 4.0451 0.2492 sec/batch\n",
      "Epoch 5/10 Iteration: 69000 Avg. Training loss: 4.0644 0.2493 sec/batch\n",
      "Nearest to that: note, suggested, claimed, noted, fact, suggest, announced, argued,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, references, blacklisting, called, sprague, denslow,\n",
      "Nearest to he: himself, wrote, thoroughbred, where, attended, sevenfold, aps, eton,\n",
      "Nearest to all: almost, virtually, nearly, listing, practically, things, pollinator, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, treat, denote,\n",
      "Nearest to i: ii, am, e, think, myself, my, world, ve,\n",
      "Nearest to when: asked, kidman, mitra, arrived, guessing, regularly, makeshift, discussing,\n",
      "Nearest to woman: sonnets, nationality, man, gentle, defensemen, fertility, est, children,\n",
      "Nearest to dr: insurgent, blok, citation, rms, usability, zaman, sierpinski, kovacs,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, evangelism, pagans, muskets, coals,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, small, rt,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, hoare, cricket, dumping, wastes,\n",
      "Nearest to lived: short, eunuch, parton, rabindranath, milked, kingdoms, worked, graveyard,\n",
      "Nearest to road: hypocrisy, traffic, farrar, hildebrand, inner, interscope, interstate, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, landing,\n",
      "Epoch 5/10 Iteration: 69100 Avg. Training loss: 4.0496 0.2525 sec/batch\n",
      "Epoch 5/10 Iteration: 69200 Avg. Training loss: 4.0741 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 69300 Avg. Training loss: 4.1116 0.2492 sec/batch\n",
      "Epoch 5/10 Iteration: 69400 Avg. Training loss: 4.0527 0.2500 sec/batch\n",
      "Epoch 5/10 Iteration: 69500 Avg. Training loss: 3.9665 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 69600 Avg. Training loss: 4.0693 0.2505 sec/batch\n",
      "Epoch 5/10 Iteration: 69700 Avg. Training loss: 4.1004 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 69800 Avg. Training loss: 4.0361 0.2490 sec/batch\n",
      "Epoch 5/10 Iteration: 69900 Avg. Training loss: 4.0374 0.2483 sec/batch\n",
      "Epoch 5/10 Iteration: 70000 Avg. Training loss: 4.0511 0.2476 sec/batch\n",
      "Nearest to that: note, suggested, claimed, fact, noted, announced, suggest, argued,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, blacklisting, references, denslow, sprague, called,\n",
      "Nearest to he: himself, wrote, attended, where, sevenfold, eton, thoroughbred, nihonshoki,\n",
      "Nearest to all: almost, virtually, nearly, listing, practically, pontryagin, things, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, denote,\n",
      "Nearest to i: ii, am, e, world, think, myself, my, war,\n",
      "Nearest to when: asked, kidman, mitra, arrived, makeshift, especially, discussing, guessing,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, gentle, fertility, est, children,\n",
      "Nearest to dr: insurgent, blok, citation, rms, usability, zaman, sierpinski, physician,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, evangelism, analogical, pagans, muskets, coals,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, small, irradiation,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, hoare, cricket, wastes, dumping,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, milked, kingdoms, worked, graveyard,\n",
      "Nearest to road: hypocrisy, traffic, farrar, roads, hildebrand, interstate, interscope, inner,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, landing,\n",
      "Epoch 5/10 Iteration: 70100 Avg. Training loss: 4.0687 0.2516 sec/batch\n",
      "Epoch 5/10 Iteration: 70200 Avg. Training loss: 4.0563 0.2491 sec/batch\n",
      "Epoch 5/10 Iteration: 70300 Avg. Training loss: 4.0640 0.2495 sec/batch\n",
      "Epoch 5/10 Iteration: 70400 Avg. Training loss: 4.0712 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 70500 Avg. Training loss: 4.0523 0.2476 sec/batch\n",
      "Epoch 5/10 Iteration: 70600 Avg. Training loss: 3.9700 0.2486 sec/batch\n",
      "Epoch 5/10 Iteration: 70700 Avg. Training loss: 4.0137 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 70800 Avg. Training loss: 4.0312 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 70900 Avg. Training loss: 4.0259 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 71000 Avg. Training loss: 4.0013 0.2487 sec/batch\n",
      "Nearest to that: note, suggested, claimed, announced, noted, suggest, argue, fact,\n",
      "Nearest to not: did, does, necessarily, do, but, until, uncommon, yet,\n",
      "Nearest to also: see, list, known, external, references, called, denslow, blacklisting,\n",
      "Nearest to he: himself, eton, thoroughbred, wrote, sevenfold, where, attended, nihonshoki,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, pollinator, things, assorted,\n",
      "Nearest to used: describe, extensively, widely, commonly, refer, term, denote, treat,\n",
      "Nearest to i: ii, am, e, my, think, myself, world, ve,\n",
      "Nearest to when: asked, kidman, mitra, guessing, discussing, arrived, especially, makeshift,\n",
      "Nearest to woman: sonnets, nationality, man, fertility, gentle, defensemen, est, children,\n",
      "Nearest to dr: insurgent, blok, citation, usability, zaman, rms, sierpinski, physician,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, pagans, muskets, evangelism, analogical, twelver,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, small, irradiation,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, wastes, dumping, hoare, tests,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, kingdoms, milked, epochs, worked,\n",
      "Nearest to road: hypocrisy, traffic, roads, farrar, interscope, hildebrand, middlesex, interstate,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, landing,\n",
      "Epoch 5/10 Iteration: 71100 Avg. Training loss: 4.0192 0.2517 sec/batch\n",
      "Epoch 5/10 Iteration: 71200 Avg. Training loss: 4.0030 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 71300 Avg. Training loss: 4.0951 0.2497 sec/batch\n",
      "Epoch 5/10 Iteration: 71400 Avg. Training loss: 3.9383 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 71500 Avg. Training loss: 3.5780 0.2481 sec/batch\n",
      "Epoch 5/10 Iteration: 71600 Avg. Training loss: 3.9404 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 71700 Avg. Training loss: 3.8328 0.2494 sec/batch\n",
      "Epoch 5/10 Iteration: 71800 Avg. Training loss: 3.9687 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 71900 Avg. Training loss: 4.0078 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 72000 Avg. Training loss: 4.0356 0.2497 sec/batch\n",
      "Nearest to that: note, suggested, claimed, announced, suggest, fact, noted, argue,\n",
      "Nearest to not: did, does, do, necessarily, but, until, yet, uncommon,\n",
      "Nearest to also: see, list, known, blacklisting, denslow, called, references, sprague,\n",
      "Nearest to he: himself, where, attended, eton, wrote, thoroughbred, sevenfold, discontented,\n",
      "Nearest to all: virtually, almost, listing, nearly, practically, things, pollinator, assorted,\n",
      "Nearest to used: describe, extensively, commonly, refer, widely, term, treat, interchangeably,\n",
      "Nearest to i: ii, am, e, think, my, world, myself, ve,\n",
      "Nearest to when: asked, kidman, arrived, makeshift, discussing, mitra, guessing, regularly,\n",
      "Nearest to woman: sonnets, nationality, man, gentle, defensemen, fertility, est, butyrate,\n",
      "Nearest to dr: blok, citation, insurgent, zaman, usability, physician, rms, kovacs,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, evangelism, pagans, muskets, analogical, coals,\n",
      "Nearest to scale: large, richter, casualties, neu, duddy, scales, small, irradiation,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, wastes, dumping, hoare, tests,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, milked, worked, kingdoms, graveyard,\n",
      "Nearest to road: hypocrisy, traffic, roads, farrar, interscope, hildebrand, haight, interstate,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, landing, intelsat,\n",
      "Epoch 5/10 Iteration: 72100 Avg. Training loss: 4.0554 0.2518 sec/batch\n",
      "Epoch 5/10 Iteration: 72200 Avg. Training loss: 4.0352 0.2481 sec/batch\n",
      "Epoch 5/10 Iteration: 72300 Avg. Training loss: 4.1072 0.2493 sec/batch\n",
      "Epoch 5/10 Iteration: 72400 Avg. Training loss: 4.0853 0.2473 sec/batch\n",
      "Epoch 5/10 Iteration: 72500 Avg. Training loss: 4.1124 0.2478 sec/batch\n",
      "Epoch 5/10 Iteration: 72600 Avg. Training loss: 4.0910 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 72700 Avg. Training loss: 4.0542 0.2488 sec/batch\n",
      "Epoch 5/10 Iteration: 72800 Avg. Training loss: 4.0064 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 72900 Avg. Training loss: 3.9124 0.2481 sec/batch\n",
      "Epoch 5/10 Iteration: 73000 Avg. Training loss: 4.0545 0.2486 sec/batch\n",
      "Nearest to that: note, suggested, claimed, announced, fact, argues, suggest, argued,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, references, called, blacklisting, external, daiko,\n",
      "Nearest to he: himself, eton, wrote, attended, where, sevenfold, thoroughbred, nihonshoki,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, pollinator, things, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, denote, treat,\n",
      "Nearest to i: ii, am, e, think, myself, world, my, ve,\n",
      "Nearest to when: asked, kidman, arrived, discussing, makeshift, mitra, especially, happens,\n",
      "Nearest to woman: sonnets, nationality, man, fertility, defensemen, gentle, est, children,\n",
      "Nearest to dr: blok, insurgent, citation, zaman, usability, physician, sierpinski, rms,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, evangelism, pagans, analogical, exclusively, multisyllabic,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, small, irradiation,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, tests, hoare, wastes, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, milked, kingdoms, worked, graveyard,\n",
      "Nearest to road: traffic, hypocrisy, roads, hildebrand, interscope, farrar, haight, middlesex,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, landing, intelsat,\n",
      "Epoch 5/10 Iteration: 73100 Avg. Training loss: 4.0500 0.2518 sec/batch\n",
      "Epoch 5/10 Iteration: 73200 Avg. Training loss: 3.9780 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 73300 Avg. Training loss: 4.0804 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 73400 Avg. Training loss: 4.0186 0.2490 sec/batch\n",
      "Epoch 5/10 Iteration: 73500 Avg. Training loss: 4.1128 0.2475 sec/batch\n",
      "Epoch 5/10 Iteration: 73600 Avg. Training loss: 4.0079 0.2487 sec/batch\n",
      "Epoch 5/10 Iteration: 73700 Avg. Training loss: 4.0607 0.2492 sec/batch\n",
      "Epoch 5/10 Iteration: 73800 Avg. Training loss: 3.9225 0.2481 sec/batch\n",
      "Epoch 5/10 Iteration: 73900 Avg. Training loss: 3.8339 0.2489 sec/batch\n",
      "Epoch 5/10 Iteration: 74000 Avg. Training loss: 4.0401 0.2491 sec/batch\n",
      "Nearest to that: note, suggested, claimed, fact, announced, argue, suggest, noted,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, external, references, called, blacklisting, daiko,\n",
      "Nearest to he: himself, wrote, eton, where, nihonshoki, sevenfold, tutti, attended,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, pollinator, assorted,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, treat, denote,\n",
      "Nearest to i: ii, am, e, think, world, myself, my, ve,\n",
      "Nearest to when: asked, kidman, arrived, discussing, makeshift, guessing, happens, especially,\n",
      "Nearest to woman: nationality, sonnets, fertility, man, children, est, gentle, defensemen,\n",
      "Nearest to dr: blok, citation, zaman, insurgent, physician, sierpinski, usability, rms,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, mawson, multisyllabic, pagans, analogical,\n",
      "Nearest to scale: large, richter, casualties, neu, duddy, scales, small, irradiation,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, tests, wastes, dumping, hoare,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, kingdoms, milked, worked, monumenta,\n",
      "Nearest to road: traffic, hypocrisy, roads, interscope, interstate, farrar, hildebrand, haight,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, eutelsat,\n",
      "Epoch 5/10 Iteration: 74100 Avg. Training loss: 4.0396 0.2529 sec/batch\n",
      "Epoch 5/10 Iteration: 74200 Avg. Training loss: 3.9579 0.2477 sec/batch\n",
      "Epoch 5/10 Iteration: 74300 Avg. Training loss: 3.9550 0.2484 sec/batch\n",
      "Epoch 5/10 Iteration: 74400 Avg. Training loss: 4.0166 0.2476 sec/batch\n",
      "Epoch 5/10 Iteration: 74500 Avg. Training loss: 4.0461 0.2485 sec/batch\n",
      "Epoch 5/10 Iteration: 74600 Avg. Training loss: 4.0690 0.2497 sec/batch\n",
      "Epoch 5/10 Iteration: 74700 Avg. Training loss: 3.9677 0.2478 sec/batch\n",
      "Epoch 5/10 Iteration: 74800 Avg. Training loss: 3.9318 0.2481 sec/batch\n",
      "Epoch 5/10 Iteration: 74900 Avg. Training loss: 4.0878 0.2493 sec/batch\n",
      "Epoch 5/10 Iteration: 75000 Avg. Training loss: 4.0264 0.2479 sec/batch\n",
      "Nearest to that: note, suggested, announced, claimed, fact, argue, noted, suggest,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, external, called, references, daiko, blacklisting,\n",
      "Nearest to he: himself, eton, where, wrote, attended, thoroughbred, nihonshoki, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, treat,\n",
      "Nearest to i: ii, am, e, think, world, myself, my, war,\n",
      "Nearest to when: asked, kidman, discussing, arrived, guessing, makeshift, mitra, especially,\n",
      "Nearest to woman: sonnets, nationality, man, children, fertility, gentle, defensemen, est,\n",
      "Nearest to dr: blok, insurgent, citation, sierpinski, usability, zaman, physician, rms,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, evangelism, exclusively, twelver, boasts,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, rt, tuning,\n",
      "Nearest to test: ban, ozone, hazardous, desertification, tests, hoare, dumping, cricket,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, kingdoms, milked, worked, graveyard,\n",
      "Nearest to road: traffic, hypocrisy, roads, farrar, athletics, hildebrand, interstate, middlesex,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, landing,\n",
      "Epoch 5/10 Iteration: 75100 Avg. Training loss: 4.1104 0.2516 sec/batch\n",
      "Epoch 6/10 Iteration: 75200 Avg. Training loss: 4.0619 0.0877 sec/batch\n",
      "Epoch 6/10 Iteration: 75300 Avg. Training loss: 4.0700 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 75400 Avg. Training loss: 4.0011 0.2493 sec/batch\n",
      "Epoch 6/10 Iteration: 75500 Avg. Training loss: 4.0620 0.2491 sec/batch\n",
      "Epoch 6/10 Iteration: 75600 Avg. Training loss: 4.0551 0.2493 sec/batch\n",
      "Epoch 6/10 Iteration: 75700 Avg. Training loss: 3.9750 0.2497 sec/batch\n",
      "Epoch 6/10 Iteration: 75800 Avg. Training loss: 4.0266 0.2491 sec/batch\n",
      "Epoch 6/10 Iteration: 75900 Avg. Training loss: 4.0072 0.2484 sec/batch\n",
      "Epoch 6/10 Iteration: 76000 Avg. Training loss: 3.9689 0.2486 sec/batch\n",
      "Nearest to that: suggested, note, claimed, noted, fact, announced, argue, argues,\n",
      "Nearest to not: did, does, do, necessarily, but, uncommon, until, yet,\n",
      "Nearest to also: see, list, known, references, external, denslow, called, daiko,\n",
      "Nearest to he: himself, eton, wrote, where, nihonshoki, attended, thoroughbred, joscelin,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pontryagin,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, denote, treat,\n",
      "Nearest to i: ii, am, e, think, my, world, myself, iii,\n",
      "Nearest to when: asked, kidman, arrived, discussing, guessing, mitra, especially, makeshift,\n",
      "Nearest to woman: sonnets, nationality, man, gentle, defensemen, children, fertility, butyrate,\n",
      "Nearest to dr: blok, sierpinski, usability, citation, insurgent, zaman, physician, rms,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, pagans, evangelism, analogical, muskets, twelver,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, rt, culpa,\n",
      "Nearest to test: ban, ozone, hazardous, tests, hoare, cricket, desertification, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, kingdoms, parton, milked, worked, monumenta,\n",
      "Nearest to road: traffic, hypocrisy, roads, farrar, middlesex, interstate, athletics, hildebrand,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, landing, intelsat,\n",
      "Epoch 6/10 Iteration: 76100 Avg. Training loss: 3.9657 0.2521 sec/batch\n",
      "Epoch 6/10 Iteration: 76200 Avg. Training loss: 3.9762 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 76300 Avg. Training loss: 4.0585 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 76400 Avg. Training loss: 4.0175 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 76500 Avg. Training loss: 4.0215 0.2484 sec/batch\n",
      "Epoch 6/10 Iteration: 76600 Avg. Training loss: 4.0049 0.2486 sec/batch\n",
      "Epoch 6/10 Iteration: 76700 Avg. Training loss: 4.0165 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 76800 Avg. Training loss: 4.0709 0.2478 sec/batch\n",
      "Epoch 6/10 Iteration: 76900 Avg. Training loss: 4.0205 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 77000 Avg. Training loss: 4.0719 0.2495 sec/batch\n",
      "Nearest to that: suggested, note, claimed, noted, announced, argue, fact, suggest,\n",
      "Nearest to not: did, does, do, necessarily, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, denslow, external, references, called, daiko,\n",
      "Nearest to he: himself, where, eton, nihonshoki, wrote, thoroughbred, sevenfold, whom,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, widely, commonly, refer, term, treat, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, myself, iii,\n",
      "Nearest to when: asked, discussing, kidman, guessing, arrived, mitra, makeshift, especially,\n",
      "Nearest to woman: sonnets, nationality, man, gentle, defensemen, children, fertility, butyrate,\n",
      "Nearest to dr: blok, sierpinski, zaman, citation, insurgent, physician, usability, rms,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, muskets, analogical, evangelism, pagans,\n",
      "Nearest to scale: large, casualties, richter, neu, scales, duddy, rt, irradiation,\n",
      "Nearest to test: ban, ozone, hazardous, tests, cricket, hoare, desertification, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, kingdoms, milked, worked, parton, monumenta,\n",
      "Nearest to road: hypocrisy, traffic, middlesex, roads, interstate, hildebrand, athletics, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, landing, intelsat,\n",
      "Epoch 6/10 Iteration: 77100 Avg. Training loss: 4.0366 0.2514 sec/batch\n",
      "Epoch 6/10 Iteration: 77200 Avg. Training loss: 4.0622 0.2499 sec/batch\n",
      "Epoch 6/10 Iteration: 77300 Avg. Training loss: 4.0184 0.2515 sec/batch\n",
      "Epoch 6/10 Iteration: 77400 Avg. Training loss: 3.8854 0.2485 sec/batch\n",
      "Epoch 6/10 Iteration: 77500 Avg. Training loss: 4.0804 0.2492 sec/batch\n",
      "Epoch 6/10 Iteration: 77600 Avg. Training loss: 4.0376 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 77700 Avg. Training loss: 4.0781 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 77800 Avg. Training loss: 4.0483 0.2475 sec/batch\n",
      "Epoch 6/10 Iteration: 77900 Avg. Training loss: 4.0518 0.2485 sec/batch\n",
      "Epoch 6/10 Iteration: 78000 Avg. Training loss: 4.0359 0.2473 sec/batch\n",
      "Nearest to that: note, suggested, noted, announced, claimed, argue, fact, argues,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, called, daiko, denslow, blacklisting, references,\n",
      "Nearest to he: himself, where, thoroughbred, eton, wrote, nihonshoki, attended, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, denote,\n",
      "Nearest to i: ii, am, e, think, my, world, myself, ve,\n",
      "Nearest to when: asked, kidman, guessing, discussing, arrived, mitra, makeshift, referring,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, children, fertility, est,\n",
      "Nearest to dr: blok, insurgent, sierpinski, citation, zaman, physician, kovacs, usability,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, pagans, analogical, muskets, twelver, exclusively,\n",
      "Nearest to scale: large, casualties, richter, neu, scales, duddy, irradiation, rt,\n",
      "Nearest to test: ban, hazardous, ozone, tests, cricket, hoare, matches, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, kingdoms, milked, parton, worked, epochs,\n",
      "Nearest to road: traffic, hypocrisy, roads, interstate, athletics, hildebrand, middlesex, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, landing,\n",
      "Epoch 6/10 Iteration: 78100 Avg. Training loss: 4.0406 0.2511 sec/batch\n",
      "Epoch 6/10 Iteration: 78200 Avg. Training loss: 4.0351 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 78300 Avg. Training loss: 4.0733 0.2485 sec/batch\n",
      "Epoch 6/10 Iteration: 78400 Avg. Training loss: 4.0653 0.2500 sec/batch\n",
      "Epoch 6/10 Iteration: 78500 Avg. Training loss: 4.0652 0.2493 sec/batch\n",
      "Epoch 6/10 Iteration: 78600 Avg. Training loss: 3.9927 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 78700 Avg. Training loss: 4.0198 0.2493 sec/batch\n",
      "Epoch 6/10 Iteration: 78800 Avg. Training loss: 3.9377 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 78900 Avg. Training loss: 3.8572 0.2494 sec/batch\n",
      "Epoch 6/10 Iteration: 79000 Avg. Training loss: 4.0219 0.2480 sec/batch\n",
      "Nearest to that: note, suggested, noted, argue, claimed, fact, argues, announced,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, references, called, blacklisting, external, denslow,\n",
      "Nearest to he: himself, thoroughbred, eton, where, nihonshoki, sevenfold, wrote, composing,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, am, e, think, my, myself, world, war,\n",
      "Nearest to when: asked, kidman, guessing, discussing, mitra, arrived, referring, especially,\n",
      "Nearest to woman: sonnets, nationality, man, fertility, defensemen, children, gentle, est,\n",
      "Nearest to dr: blok, insurgent, citation, zaman, physician, sierpinski, usability, kovacs,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, muskets, pagans, exclusively, twelver,\n",
      "Nearest to scale: large, casualties, richter, neu, scales, duddy, irradiation, rt,\n",
      "Nearest to test: ban, hazardous, ozone, tests, cricket, hoare, desertification, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, milked, kingdoms, worked, living,\n",
      "Nearest to road: traffic, hypocrisy, roads, middlesex, interstate, hildebrand, farrar, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, ost,\n",
      "Epoch 6/10 Iteration: 79100 Avg. Training loss: 4.0482 0.2517 sec/batch\n",
      "Epoch 6/10 Iteration: 79200 Avg. Training loss: 3.9909 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 79300 Avg. Training loss: 4.0325 0.2496 sec/batch\n",
      "Epoch 6/10 Iteration: 79400 Avg. Training loss: 4.0267 0.2486 sec/batch\n",
      "Epoch 6/10 Iteration: 79500 Avg. Training loss: 4.0009 0.2478 sec/batch\n",
      "Epoch 6/10 Iteration: 79600 Avg. Training loss: 4.0367 0.2482 sec/batch\n",
      "Epoch 6/10 Iteration: 79700 Avg. Training loss: 3.9015 0.2482 sec/batch\n",
      "Epoch 6/10 Iteration: 79800 Avg. Training loss: 4.0640 0.2483 sec/batch\n",
      "Epoch 6/10 Iteration: 79900 Avg. Training loss: 4.0495 0.2499 sec/batch\n",
      "Epoch 6/10 Iteration: 80000 Avg. Training loss: 4.0624 0.2484 sec/batch\n",
      "Nearest to that: note, suggested, noted, claimed, fact, announced, argue, argues,\n",
      "Nearest to not: did, does, do, necessarily, uncommon, but, until, yet,\n",
      "Nearest to also: see, list, known, called, references, blacklisting, external, daiko,\n",
      "Nearest to he: himself, thoroughbred, eton, nihonshoki, where, wrote, sevenfold, transmutation,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, widely, commonly, refer, term, denote, frequently,\n",
      "Nearest to i: ii, e, am, think, world, myself, my, ve,\n",
      "Nearest to when: asked, kidman, guessing, mitra, arrived, happens, discussing, referring,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, children, fertility, est,\n",
      "Nearest to dr: blok, kovacs, citation, zaman, physician, insurgent, usability, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, muskets, exclusively, pagans, twelver,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, irradiation, rt,\n",
      "Nearest to test: ban, hazardous, ozone, tests, cricket, desertification, hoare, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, milked, worked, kingdoms, living,\n",
      "Nearest to road: traffic, hypocrisy, middlesex, roads, interstate, athletics, farrar, hildebrand,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, indian, intelsat, landing,\n",
      "Epoch 6/10 Iteration: 80100 Avg. Training loss: 4.0470 0.2518 sec/batch\n",
      "Epoch 6/10 Iteration: 80200 Avg. Training loss: 4.0597 0.2486 sec/batch\n",
      "Epoch 6/10 Iteration: 80300 Avg. Training loss: 4.0394 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 80400 Avg. Training loss: 4.0252 0.2474 sec/batch\n",
      "Epoch 6/10 Iteration: 80500 Avg. Training loss: 4.0824 0.2496 sec/batch\n",
      "Epoch 6/10 Iteration: 80600 Avg. Training loss: 4.0628 0.2491 sec/batch\n",
      "Epoch 6/10 Iteration: 80700 Avg. Training loss: 3.9783 0.2487 sec/batch\n",
      "Epoch 6/10 Iteration: 80800 Avg. Training loss: 3.9950 0.2492 sec/batch\n",
      "Epoch 6/10 Iteration: 80900 Avg. Training loss: 4.0277 0.2480 sec/batch\n",
      "Epoch 6/10 Iteration: 81000 Avg. Training loss: 3.9622 0.2481 sec/batch\n",
      "Nearest to that: note, suggested, fact, noted, announced, claimed, argue, argues,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, blacklisting, called, external, references, denslow,\n",
      "Nearest to he: himself, thoroughbred, eton, where, wrote, nihonshoki, sevenfold, resurrecting,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, refer, widely, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, myself, my, ve,\n",
      "Nearest to when: asked, kidman, guessing, discussing, arrived, mitra, happens, especially,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, fertility, children, gentle, est,\n",
      "Nearest to dr: blok, physician, kovacs, rms, zaman, citation, sierpinski, insurgent,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, muskets, analogical, exclusively, pagans, evangelism,\n",
      "Nearest to scale: large, richter, casualties, scales, neu, irradiation, rt, duddy,\n",
      "Nearest to test: ban, hazardous, ozone, tests, cricket, globes, desertification, hoare,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, parton, milked, kingdoms, graveyard,\n",
      "Nearest to road: traffic, hypocrisy, middlesex, athletics, roads, farrar, hildebrand, interstate,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, indian, intelsat, landing,\n",
      "Epoch 6/10 Iteration: 81100 Avg. Training loss: 4.0517 0.2519 sec/batch\n",
      "Epoch 6/10 Iteration: 81200 Avg. Training loss: 4.0301 0.2481 sec/batch\n",
      "Epoch 6/10 Iteration: 81300 Avg. Training loss: 4.0560 0.2502 sec/batch\n",
      "Epoch 6/10 Iteration: 81400 Avg. Training loss: 4.0596 0.2483 sec/batch\n",
      "Epoch 6/10 Iteration: 81500 Avg. Training loss: 4.0756 0.2494 sec/batch\n",
      "Epoch 6/10 Iteration: 81600 Avg. Training loss: 4.0165 0.2483 sec/batch\n",
      "Epoch 6/10 Iteration: 81700 Avg. Training loss: 3.9215 0.2494 sec/batch\n",
      "Epoch 6/10 Iteration: 81800 Avg. Training loss: 4.0595 0.2494 sec/batch\n",
      "Epoch 6/10 Iteration: 81900 Avg. Training loss: 3.9837 0.2486 sec/batch\n",
      "Epoch 6/10 Iteration: 82000 Avg. Training loss: 4.0058 0.2492 sec/batch\n",
      "Nearest to that: note, suggested, noted, fact, argue, argues, claimed, argued,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, references, called, blacklisting, external, denslow,\n",
      "Nearest to he: himself, eton, wrote, thoroughbred, where, transmutation, nihonshoki, tutti,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, pollinator,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, my, myself, world, ve,\n",
      "Nearest to when: asked, kidman, guessing, discussing, arrived, mitra, happens, referring,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, fertility, gentle, children, butyrate,\n",
      "Nearest to dr: blok, physician, citation, kovacs, usability, zaman, rms, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, exclusively, muskets, twelver, mawson,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, rt, duddy, irradiation,\n",
      "Nearest to test: ban, hazardous, ozone, tests, cricket, globes, desertification, wastes,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, worked, milked, kingdoms, graveyard,\n",
      "Nearest to road: traffic, hypocrisy, middlesex, farrar, hildebrand, interstate, interscope, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, indian, launches, ost,\n",
      "Epoch 6/10 Iteration: 82100 Avg. Training loss: 4.0657 0.2517 sec/batch\n",
      "Epoch 6/10 Iteration: 82200 Avg. Training loss: 4.0522 0.2497 sec/batch\n",
      "Epoch 6/10 Iteration: 82300 Avg. Training loss: 4.0179 0.2498 sec/batch\n",
      "Epoch 6/10 Iteration: 82400 Avg. Training loss: 4.0510 0.2487 sec/batch\n",
      "Epoch 6/10 Iteration: 82500 Avg. Training loss: 4.0478 0.2483 sec/batch\n",
      "Epoch 6/10 Iteration: 82600 Avg. Training loss: 3.9734 0.2496 sec/batch\n",
      "Epoch 6/10 Iteration: 82700 Avg. Training loss: 4.0328 0.2498 sec/batch\n",
      "Epoch 6/10 Iteration: 82800 Avg. Training loss: 4.0379 0.2502 sec/batch\n",
      "Epoch 6/10 Iteration: 82900 Avg. Training loss: 3.9506 0.2485 sec/batch\n",
      "Epoch 6/10 Iteration: 83000 Avg. Training loss: 4.0096 0.2489 sec/batch\n",
      "Nearest to that: note, suggested, noted, argues, argued, announced, fact, suggest,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, blacklisting, references, called, external, sprague,\n",
      "Nearest to he: himself, where, thoroughbred, eton, sevenfold, transmutation, wrote, aps,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, am, e, think, world, my, myself, ve,\n",
      "Nearest to when: asked, kidman, guessing, arrived, discussing, happens, mitra, especially,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, fertility, children, est,\n",
      "Nearest to dr: blok, physician, kovacs, citation, zaman, rms, usability, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, analogical, evangelism, democratisation, muskets,\n",
      "Nearest to scale: large, richter, casualties, scales, neu, duddy, rt, irradiation,\n",
      "Nearest to test: ban, hazardous, tests, cricket, ozone, globes, hoare, desertification,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, worked, milked, kingdoms, living,\n",
      "Nearest to road: hypocrisy, traffic, hildebrand, middlesex, extractive, roads, farrar, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, castilla, pacific, indian, ost, intelsat,\n",
      "Epoch 6/10 Iteration: 83100 Avg. Training loss: 4.0536 0.2514 sec/batch\n",
      "Epoch 6/10 Iteration: 83200 Avg. Training loss: 3.8770 0.2491 sec/batch\n",
      "Epoch 6/10 Iteration: 83300 Avg. Training loss: 4.0648 0.2499 sec/batch\n",
      "Epoch 6/10 Iteration: 83400 Avg. Training loss: 4.0352 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 83500 Avg. Training loss: 4.0209 0.2481 sec/batch\n",
      "Epoch 6/10 Iteration: 83600 Avg. Training loss: 4.0553 0.2493 sec/batch\n",
      "Epoch 6/10 Iteration: 83700 Avg. Training loss: 3.8851 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 83800 Avg. Training loss: 4.0612 0.2492 sec/batch\n",
      "Epoch 6/10 Iteration: 83900 Avg. Training loss: 4.0072 0.2476 sec/batch\n",
      "Epoch 6/10 Iteration: 84000 Avg. Training loss: 4.0727 0.2496 sec/batch\n",
      "Nearest to that: note, suggested, noted, suggest, argue, fact, argued, claimed,\n",
      "Nearest to not: did, does, do, necessarily, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, references, blacklisting, sprague, called, synonyms,\n",
      "Nearest to he: himself, thoroughbred, where, wrote, eton, sevenfold, transmutation, aps,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, assorted, pollinator,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, coined,\n",
      "Nearest to i: ii, am, e, think, my, world, myself, ve,\n",
      "Nearest to when: asked, kidman, arrived, guessing, discussing, happens, especially, referring,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, gentle, fertility, children, est,\n",
      "Nearest to dr: blok, physician, kovacs, citation, zaman, rms, sierpinski, usability,\n",
      "Nearest to mainly: primarily, emulsions, analogical, reinforced, evangelism, exclusively, democratisation, pagans,\n",
      "Nearest to scale: large, richter, casualties, neu, scales, duddy, rt, tornadoes,\n",
      "Nearest to test: ban, hazardous, tests, ozone, cricket, hoare, desertification, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, parton, living, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, farrar, extractive, donovan, roads, middlesex,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, castilla, intelsat, eutelsat,\n",
      "Epoch 6/10 Iteration: 84100 Avg. Training loss: 4.0393 0.2507 sec/batch\n",
      "Epoch 6/10 Iteration: 84200 Avg. Training loss: 4.0409 0.2501 sec/batch\n",
      "Epoch 6/10 Iteration: 84300 Avg. Training loss: 4.0903 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 84400 Avg. Training loss: 4.0593 0.2491 sec/batch\n",
      "Epoch 6/10 Iteration: 84500 Avg. Training loss: 4.0044 0.2475 sec/batch\n",
      "Epoch 6/10 Iteration: 84600 Avg. Training loss: 3.9872 0.2484 sec/batch\n",
      "Epoch 6/10 Iteration: 84700 Avg. Training loss: 4.0572 0.2484 sec/batch\n",
      "Epoch 6/10 Iteration: 84800 Avg. Training loss: 4.0497 0.2485 sec/batch\n",
      "Epoch 6/10 Iteration: 84900 Avg. Training loss: 4.0547 0.2481 sec/batch\n",
      "Epoch 6/10 Iteration: 85000 Avg. Training loss: 4.0153 0.2493 sec/batch\n",
      "Nearest to that: note, suggested, noted, claimed, argue, announced, fact, suggest,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, blacklisting, references, sprague, external, called,\n",
      "Nearest to he: himself, thoroughbred, where, eton, sevenfold, transmutation, wrote, discontented,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, assorted, pollinator,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, am, e, my, think, world, war, myself,\n",
      "Nearest to when: asked, discussing, kidman, arrived, guessing, especially, referring, mitra,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, gentle, fertility, children, est,\n",
      "Nearest to dr: blok, physician, zaman, kovacs, citation, rms, insurgent, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, analogical, reinforced, exclusively, evangelism, paternoster, mawson,\n",
      "Nearest to scale: large, richter, casualties, scales, neu, duddy, rt, irradiation,\n",
      "Nearest to test: ban, hazardous, tests, ozone, hoare, cricket, desertification, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, milked, worked, parton, living, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, farrar, interstate, extractive, middlesex, roads,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, castilla, indian, intelsat, eutelsat,\n",
      "Epoch 6/10 Iteration: 85100 Avg. Training loss: 4.0563 0.2512 sec/batch\n",
      "Epoch 6/10 Iteration: 85200 Avg. Training loss: 4.0609 0.2495 sec/batch\n",
      "Epoch 6/10 Iteration: 85300 Avg. Training loss: 4.0361 0.2498 sec/batch\n",
      "Epoch 6/10 Iteration: 85400 Avg. Training loss: 4.0648 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 85500 Avg. Training loss: 4.0513 0.2499 sec/batch\n",
      "Epoch 6/10 Iteration: 85600 Avg. Training loss: 4.0102 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 85700 Avg. Training loss: 3.9486 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 85800 Avg. Training loss: 4.0058 0.2486 sec/batch\n",
      "Epoch 6/10 Iteration: 85900 Avg. Training loss: 4.0307 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 86000 Avg. Training loss: 3.9738 0.2473 sec/batch\n",
      "Nearest to that: note, suggested, argue, noted, announced, argued, claimed, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, references, external, called, blacklisting, synonyms,\n",
      "Nearest to he: himself, thoroughbred, where, sevenfold, eton, transmutation, wrote, discontented,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, my, world, think, myself, ve,\n",
      "Nearest to when: asked, discussing, kidman, arrived, referring, especially, guessing, occurs,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, gentle, fertility, children, est,\n",
      "Nearest to dr: blok, physician, zaman, kovacs, citation, insurgent, rms, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, exclusively, mawson, evangelism, coals,\n",
      "Nearest to scale: large, richter, casualties, scales, neu, rt, duddy, small,\n",
      "Nearest to test: ban, hazardous, tests, ozone, desertification, hoare, dumping, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, parton, milked, living, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, extractive, roads, middlesex, hildebrand, interstate, donovan,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, castilla, intelsat, eutelsat,\n",
      "Epoch 6/10 Iteration: 86100 Avg. Training loss: 4.0328 0.2510 sec/batch\n",
      "Epoch 6/10 Iteration: 86200 Avg. Training loss: 4.0053 0.2499 sec/batch\n",
      "Epoch 6/10 Iteration: 86300 Avg. Training loss: 4.0624 0.2477 sec/batch\n",
      "Epoch 6/10 Iteration: 86400 Avg. Training loss: 4.0004 0.2486 sec/batch\n",
      "Epoch 6/10 Iteration: 86500 Avg. Training loss: 3.6941 0.2483 sec/batch\n",
      "Epoch 6/10 Iteration: 86600 Avg. Training loss: 3.7734 0.2472 sec/batch\n",
      "Epoch 6/10 Iteration: 86700 Avg. Training loss: 3.8543 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 86800 Avg. Training loss: 3.9001 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 86900 Avg. Training loss: 3.9565 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 87000 Avg. Training loss: 4.0299 0.2490 sec/batch\n",
      "Nearest to that: note, suggested, announced, argue, noted, claimed, suggest, argued,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, blacklisting, called, references, lemnos, sprague,\n",
      "Nearest to he: himself, where, attended, thoroughbred, eton, sevenfold, wrote, discontented,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, am, e, my, think, world, myself, ve,\n",
      "Nearest to when: asked, arrived, kidman, discussing, referring, especially, mitra, makeshift,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, fertility, children, est,\n",
      "Nearest to dr: blok, physician, zaman, kovacs, rms, insurgent, citation, sierpinski,\n",
      "Nearest to mainly: primarily, emulsions, analogical, reinforced, exclusively, evangelism, paternoster, coals,\n",
      "Nearest to scale: large, richter, casualties, scales, neu, duddy, rt, small,\n",
      "Nearest to test: ban, hazardous, tests, ozone, desertification, globes, dumping, hoare,\n",
      "Nearest to lived: short, eunuch, worked, rabindranath, parton, milked, living, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, donovan, extractive, roads, middlesex, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 6/10 Iteration: 87100 Avg. Training loss: 4.0355 0.2514 sec/batch\n",
      "Epoch 6/10 Iteration: 87200 Avg. Training loss: 4.0362 0.2487 sec/batch\n",
      "Epoch 6/10 Iteration: 87300 Avg. Training loss: 4.0553 0.2483 sec/batch\n",
      "Epoch 6/10 Iteration: 87400 Avg. Training loss: 4.0893 0.2485 sec/batch\n",
      "Epoch 6/10 Iteration: 87500 Avg. Training loss: 4.0815 0.2488 sec/batch\n",
      "Epoch 6/10 Iteration: 87600 Avg. Training loss: 4.0768 0.2493 sec/batch\n",
      "Epoch 6/10 Iteration: 87700 Avg. Training loss: 4.0696 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 87800 Avg. Training loss: 4.0450 0.2492 sec/batch\n",
      "Epoch 6/10 Iteration: 87900 Avg. Training loss: 3.8474 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 88000 Avg. Training loss: 4.0475 0.2487 sec/batch\n",
      "Nearest to that: note, suggested, announced, argued, argue, noted, claimed, suggest,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, external, references, blacklisting, called, daiko,\n",
      "Nearest to he: himself, attended, thoroughbred, where, wrote, eton, sevenfold, resurrecting,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, denote, interchangeably,\n",
      "Nearest to i: ii, am, e, think, my, world, myself, ve,\n",
      "Nearest to when: asked, arrived, discussing, kidman, especially, referring, happens, guessing,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, fertility, gentle, children, est,\n",
      "Nearest to dr: blok, physician, zaman, insurgent, sierpinski, kovacs, citation, rms,\n",
      "Nearest to mainly: primarily, emulsions, analogical, reinforced, evangelism, exclusively, mawson, spruce,\n",
      "Nearest to scale: large, richter, casualties, scales, neu, duddy, rt, small,\n",
      "Nearest to test: ban, hazardous, tests, ozone, desertification, dumping, globes, wastes,\n",
      "Nearest to lived: short, eunuch, rabindranath, parton, worked, milked, living, monumenta,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, donovan, roads, middlesex, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 6/10 Iteration: 88100 Avg. Training loss: 4.0531 0.2519 sec/batch\n",
      "Epoch 6/10 Iteration: 88200 Avg. Training loss: 3.9812 0.2483 sec/batch\n",
      "Epoch 6/10 Iteration: 88300 Avg. Training loss: 4.0333 0.2480 sec/batch\n",
      "Epoch 6/10 Iteration: 88400 Avg. Training loss: 3.9866 0.2490 sec/batch\n",
      "Epoch 6/10 Iteration: 88500 Avg. Training loss: 4.0889 0.2492 sec/batch\n",
      "Epoch 6/10 Iteration: 88600 Avg. Training loss: 4.0441 0.2506 sec/batch\n",
      "Epoch 6/10 Iteration: 88700 Avg. Training loss: 4.0270 0.2494 sec/batch\n",
      "Epoch 6/10 Iteration: 88800 Avg. Training loss: 3.9685 0.2481 sec/batch\n",
      "Epoch 6/10 Iteration: 88900 Avg. Training loss: 3.8333 0.2485 sec/batch\n",
      "Epoch 6/10 Iteration: 89000 Avg. Training loss: 3.9230 0.2491 sec/batch\n",
      "Nearest to that: note, suggested, argue, fact, claimed, suggest, announced, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, references, links, called, blacklisting,\n",
      "Nearest to he: himself, thoroughbred, where, wrote, attended, eton, sevenfold, resurrecting,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, denote, interchangeably,\n",
      "Nearest to i: ii, am, e, think, my, world, myself, war,\n",
      "Nearest to when: asked, arrived, discussing, kidman, referring, especially, happens, guessing,\n",
      "Nearest to woman: nationality, sonnets, man, children, fertility, defensemen, est, gentle,\n",
      "Nearest to dr: blok, physician, zaman, insurgent, sierpinski, kovacs, acc, citation,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, analogical, mawson, spruce, democratisation,\n",
      "Nearest to scale: large, richter, casualties, scales, neu, rt, duddy, small,\n",
      "Nearest to test: ban, hazardous, tests, ozone, dumping, wastes, desertification, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, parton, living, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, donovan, farrar, interstate, roads,\n",
      "Nearest to ocean: atlantic, intersputnik, indian, posed, pacific, castilla, intelsat, eutelsat,\n",
      "Epoch 6/10 Iteration: 89100 Avg. Training loss: 4.0334 0.2525 sec/batch\n",
      "Epoch 6/10 Iteration: 89200 Avg. Training loss: 4.0203 0.2492 sec/batch\n",
      "Epoch 6/10 Iteration: 89300 Avg. Training loss: 3.9060 0.2481 sec/batch\n",
      "Epoch 6/10 Iteration: 89400 Avg. Training loss: 4.0189 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 89500 Avg. Training loss: 4.0454 0.2481 sec/batch\n",
      "Epoch 6/10 Iteration: 89600 Avg. Training loss: 4.0287 0.2492 sec/batch\n",
      "Epoch 6/10 Iteration: 89700 Avg. Training loss: 3.9895 0.2489 sec/batch\n",
      "Epoch 6/10 Iteration: 89800 Avg. Training loss: 3.9787 0.2478 sec/batch\n",
      "Epoch 6/10 Iteration: 89900 Avg. Training loss: 4.0074 0.2487 sec/batch\n",
      "Epoch 6/10 Iteration: 90000 Avg. Training loss: 4.0124 0.2498 sec/batch\n",
      "Nearest to that: note, suggested, fact, announced, argue, noted, claimed, suggest,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, external, called, blacklisting, references, daiko,\n",
      "Nearest to he: himself, thoroughbred, where, wrote, attended, eton, sevenfold, resurrecting,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, world, think, my, war, ve,\n",
      "Nearest to when: asked, discussing, arrived, especially, kidman, guessing, referring, happens,\n",
      "Nearest to woman: nationality, sonnets, man, defensemen, children, fertility, gentle, noun,\n",
      "Nearest to dr: blok, physician, zaman, insurgent, sierpinski, kovacs, acc, positioned,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, exclusively, mawson, evangelism, democratisation,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, duddy, fifths,\n",
      "Nearest to test: ban, hazardous, tests, ozone, cricket, dumping, wastes, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, parton, living, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, donovan, interstate, farrar, middlesex,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, ost,\n",
      "Epoch 6/10 Iteration: 90100 Avg. Training loss: 4.0919 0.2523 sec/batch\n",
      "Epoch 7/10 Iteration: 90200 Avg. Training loss: 4.0647 0.0054 sec/batch\n",
      "Epoch 7/10 Iteration: 90300 Avg. Training loss: 4.0667 0.2490 sec/batch\n",
      "Epoch 7/10 Iteration: 90400 Avg. Training loss: 3.9915 0.2490 sec/batch\n",
      "Epoch 7/10 Iteration: 90500 Avg. Training loss: 4.0281 0.2492 sec/batch\n",
      "Epoch 7/10 Iteration: 90600 Avg. Training loss: 4.0596 0.2495 sec/batch\n",
      "Epoch 7/10 Iteration: 90700 Avg. Training loss: 3.9665 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 90800 Avg. Training loss: 4.0144 0.2494 sec/batch\n",
      "Epoch 7/10 Iteration: 90900 Avg. Training loss: 4.0177 0.2479 sec/batch\n",
      "Epoch 7/10 Iteration: 91000 Avg. Training loss: 4.0117 0.2499 sec/batch\n",
      "Nearest to that: note, suggested, claimed, fact, argue, noted, argued, suggest,\n",
      "Nearest to not: did, does, necessarily, do, but, uncommon, yet, until,\n",
      "Nearest to also: see, list, known, external, references, called, daiko, denslow,\n",
      "Nearest to he: himself, wrote, thoroughbred, where, eton, attended, resurrecting, sevenfold,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, commonly, refer, term, denote, treat,\n",
      "Nearest to i: ii, e, am, world, think, my, iii, myself,\n",
      "Nearest to when: asked, discussing, arrived, referring, especially, happens, guessing, kidman,\n",
      "Nearest to woman: nationality, sonnets, man, defensemen, children, gentle, fertility, women,\n",
      "Nearest to dr: blok, zaman, physician, insurgent, sierpinski, kovacs, acc, citation,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, analogical, exclusively, democratisation, evangelism, mawson,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, duddy, fifths,\n",
      "Nearest to test: ban, tests, hazardous, cricket, globes, matches, hoare, ozone,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, parton, living, long,\n",
      "Nearest to road: traffic, hypocrisy, extractive, hildebrand, donovan, interstate, middlesex, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 7/10 Iteration: 91100 Avg. Training loss: 3.9300 0.2514 sec/batch\n",
      "Epoch 7/10 Iteration: 91200 Avg. Training loss: 3.9504 0.2497 sec/batch\n",
      "Epoch 7/10 Iteration: 91300 Avg. Training loss: 4.0015 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 91400 Avg. Training loss: 4.0190 0.2496 sec/batch\n",
      "Epoch 7/10 Iteration: 91500 Avg. Training loss: 4.0031 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 91600 Avg. Training loss: 3.9921 0.2479 sec/batch\n",
      "Epoch 7/10 Iteration: 91700 Avg. Training loss: 3.9923 0.2505 sec/batch\n",
      "Epoch 7/10 Iteration: 91800 Avg. Training loss: 4.0485 0.2485 sec/batch\n",
      "Epoch 7/10 Iteration: 91900 Avg. Training loss: 4.0328 0.2471 sec/batch\n",
      "Epoch 7/10 Iteration: 92000 Avg. Training loss: 4.0415 0.2493 sec/batch\n",
      "Nearest to that: note, suggested, claimed, argue, noted, fact, argued, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, external, called, denslow, references, daiko,\n",
      "Nearest to he: himself, where, wrote, thoroughbred, eton, attended, nihonshoki, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, commonly, refer, term, treat, interchangeably,\n",
      "Nearest to i: ii, e, am, my, world, think, myself, war,\n",
      "Nearest to when: asked, discussing, arrived, referring, especially, guessing, kidman, occurs,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, gentle, children, fertility, women,\n",
      "Nearest to dr: blok, zaman, physician, sierpinski, insurgent, kovacs, acc, positioned,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, analogical, muskets, democratisation, cerebrospinal,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, duddy, rt, sizeable,\n",
      "Nearest to test: ban, tests, hazardous, cricket, globes, ozone, hoare, matches,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, parton, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, extractive, hildebrand, middlesex, donovan, farrar, interstate,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 7/10 Iteration: 92100 Avg. Training loss: 4.0448 0.2512 sec/batch\n",
      "Epoch 7/10 Iteration: 92200 Avg. Training loss: 4.0471 0.2486 sec/batch\n",
      "Epoch 7/10 Iteration: 92300 Avg. Training loss: 4.0481 0.2499 sec/batch\n",
      "Epoch 7/10 Iteration: 92400 Avg. Training loss: 3.8232 0.2487 sec/batch\n",
      "Epoch 7/10 Iteration: 92500 Avg. Training loss: 4.0889 0.2497 sec/batch\n",
      "Epoch 7/10 Iteration: 92600 Avg. Training loss: 4.0304 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 92700 Avg. Training loss: 4.0500 0.2477 sec/batch\n",
      "Epoch 7/10 Iteration: 92800 Avg. Training loss: 4.0518 0.2494 sec/batch\n",
      "Epoch 7/10 Iteration: 92900 Avg. Training loss: 4.0476 0.2486 sec/batch\n",
      "Epoch 7/10 Iteration: 93000 Avg. Training loss: 4.0400 0.2488 sec/batch\n",
      "Nearest to that: note, suggested, noted, argue, announced, argued, claimed, fact,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, called, external, daiko, blacklisting, denslow,\n",
      "Nearest to he: himself, where, thoroughbred, wrote, eton, attended, resurrecting, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, pollinator,\n",
      "Nearest to used: describe, extensively, commonly, widely, refer, term, interchangeably, denote,\n",
      "Nearest to i: ii, am, e, my, think, world, myself, ve,\n",
      "Nearest to when: asked, discussing, guessing, arrived, referring, especially, kidman, mitra,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, children, fertility, women,\n",
      "Nearest to dr: blok, physician, zaman, kovacs, sierpinski, positioned, insurgent, acc,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, analogical, muskets, cerebrospinal, twelver,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, duddy, rt, beltline,\n",
      "Nearest to test: ban, hazardous, tests, cricket, wastes, hoare, matches, ozone,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, kingdoms, parton,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, roads, donovan, athletics, racing,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 7/10 Iteration: 93100 Avg. Training loss: 4.0304 0.2521 sec/batch\n",
      "Epoch 7/10 Iteration: 93200 Avg. Training loss: 4.0003 0.2477 sec/batch\n",
      "Epoch 7/10 Iteration: 93300 Avg. Training loss: 4.0740 0.2481 sec/batch\n",
      "Epoch 7/10 Iteration: 93400 Avg. Training loss: 4.0607 0.2494 sec/batch\n",
      "Epoch 7/10 Iteration: 93500 Avg. Training loss: 4.0427 0.2490 sec/batch\n",
      "Epoch 7/10 Iteration: 93600 Avg. Training loss: 3.9912 0.2474 sec/batch\n",
      "Epoch 7/10 Iteration: 93700 Avg. Training loss: 4.0254 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 93800 Avg. Training loss: 3.9863 0.2496 sec/batch\n",
      "Epoch 7/10 Iteration: 93900 Avg. Training loss: 3.8307 0.2482 sec/batch\n",
      "Epoch 7/10 Iteration: 94000 Avg. Training loss: 3.9810 0.2498 sec/batch\n",
      "Nearest to that: note, suggested, argue, noted, fact, claimed, argued, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, external, references, called, blacklisting, but,\n",
      "Nearest to he: himself, thoroughbred, where, transmutation, wrote, eton, resurrecting, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, commonly, widely, term, interchangeably, denote,\n",
      "Nearest to i: ii, e, am, my, think, world, war, myself,\n",
      "Nearest to when: asked, discussing, referring, guessing, especially, arrived, kidman, occurs,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, children, fertility, gentle, noun,\n",
      "Nearest to dr: blok, zaman, insurgent, physician, sierpinski, kovacs, positioned, acc,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, analogical, muskets, paternoster, twelver,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, workforce, duddy,\n",
      "Nearest to test: ban, hazardous, tests, cricket, wastes, ozone, hoare, dumping,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, kingdoms, parton,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, roads, interstate, donovan, middlesex,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 7/10 Iteration: 94100 Avg. Training loss: 4.0419 0.2520 sec/batch\n",
      "Epoch 7/10 Iteration: 94200 Avg. Training loss: 3.9926 0.2474 sec/batch\n",
      "Epoch 7/10 Iteration: 94300 Avg. Training loss: 4.0306 0.2485 sec/batch\n",
      "Epoch 7/10 Iteration: 94400 Avg. Training loss: 4.0082 0.2499 sec/batch\n",
      "Epoch 7/10 Iteration: 94500 Avg. Training loss: 4.0075 0.2486 sec/batch\n",
      "Epoch 7/10 Iteration: 94600 Avg. Training loss: 4.0213 0.2481 sec/batch\n",
      "Epoch 7/10 Iteration: 94700 Avg. Training loss: 3.9118 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 94800 Avg. Training loss: 4.0261 0.2469 sec/batch\n",
      "Epoch 7/10 Iteration: 94900 Avg. Training loss: 4.0369 0.2486 sec/batch\n",
      "Epoch 7/10 Iteration: 95000 Avg. Training loss: 4.0575 0.2487 sec/batch\n",
      "Nearest to that: note, suggested, noted, argue, fact, claimed, argued, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, until, yet,\n",
      "Nearest to also: see, list, known, called, external, references, blacklisting, daiko,\n",
      "Nearest to he: himself, thoroughbred, transmutation, eton, wrote, where, orcinus, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, commonly, refer, term, denote, frequently,\n",
      "Nearest to i: ii, e, am, world, think, my, ve, war,\n",
      "Nearest to when: asked, discussing, referring, guessing, arrived, kidman, mitra, happens,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, children, fertility, women,\n",
      "Nearest to dr: blok, insurgent, kovacs, physician, zaman, acc, sierpinski, positioned,\n",
      "Nearest to mainly: primarily, emulsions, reinforced, exclusively, analogical, muskets, democratisation, paternoster,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, duddy, tornadoes,\n",
      "Nearest to test: ban, hazardous, tests, cricket, wastes, ozone, hoare, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, living, parton, milked, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, middlesex, donovan, extractive, athletics, racing,\n",
      "Nearest to ocean: atlantic, posed, intersputnik, indian, pacific, castilla, intelsat, ost,\n",
      "Epoch 7/10 Iteration: 95100 Avg. Training loss: 4.0365 0.2520 sec/batch\n",
      "Epoch 7/10 Iteration: 95200 Avg. Training loss: 4.0384 0.2492 sec/batch\n",
      "Epoch 7/10 Iteration: 95300 Avg. Training loss: 4.0444 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 95400 Avg. Training loss: 4.0197 0.2493 sec/batch\n",
      "Epoch 7/10 Iteration: 95500 Avg. Training loss: 4.0560 0.2481 sec/batch\n",
      "Epoch 7/10 Iteration: 95600 Avg. Training loss: 4.0726 0.2477 sec/batch\n",
      "Epoch 7/10 Iteration: 95700 Avg. Training loss: 3.9694 0.2494 sec/batch\n",
      "Epoch 7/10 Iteration: 95800 Avg. Training loss: 4.0141 0.2502 sec/batch\n",
      "Epoch 7/10 Iteration: 95900 Avg. Training loss: 4.0268 0.2473 sec/batch\n",
      "Epoch 7/10 Iteration: 96000 Avg. Training loss: 3.9627 0.2477 sec/batch\n",
      "Nearest to that: note, suggested, noted, argue, fact, argued, suggest, claimed,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, until,\n",
      "Nearest to also: see, list, known, blacklisting, called, external, references, daiko,\n",
      "Nearest to he: himself, thoroughbred, wrote, transmutation, eton, where, sevenfold, popularize,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, except,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, treat,\n",
      "Nearest to i: ii, e, am, world, my, think, war, ve,\n",
      "Nearest to when: asked, discussing, referring, especially, guessing, kidman, occurs, happens,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, children, fertility, gentle, women,\n",
      "Nearest to dr: blok, physician, kovacs, insurgent, zaman, surgery, sierpinski, positioned,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, reinforced, analogical, muskets, democratisation, elas,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, tornadoes, rt, duddy,\n",
      "Nearest to test: ban, hazardous, tests, cricket, wastes, dumping, ozone, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, parton, living, long,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, middlesex, donovan, extractive, athletics, farrar,\n",
      "Nearest to ocean: atlantic, posed, intersputnik, pacific, indian, castilla, intelsat, ost,\n",
      "Epoch 7/10 Iteration: 96100 Avg. Training loss: 4.0377 0.2519 sec/batch\n",
      "Epoch 7/10 Iteration: 96200 Avg. Training loss: 4.0217 0.2495 sec/batch\n",
      "Epoch 7/10 Iteration: 96300 Avg. Training loss: 4.0428 0.2480 sec/batch\n",
      "Epoch 7/10 Iteration: 96400 Avg. Training loss: 4.0199 0.2485 sec/batch\n",
      "Epoch 7/10 Iteration: 96500 Avg. Training loss: 4.0662 0.2499 sec/batch\n",
      "Epoch 7/10 Iteration: 96600 Avg. Training loss: 4.0392 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 96700 Avg. Training loss: 3.9725 0.2477 sec/batch\n",
      "Epoch 7/10 Iteration: 96800 Avg. Training loss: 3.9812 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 96900 Avg. Training loss: 4.0278 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 97000 Avg. Training loss: 3.9762 0.2485 sec/batch\n",
      "Nearest to that: note, suggested, argue, noted, fact, suggest, argued, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, confused,\n",
      "Nearest to also: see, list, known, external, references, called, blacklisting, daiko,\n",
      "Nearest to he: himself, transmutation, eton, thoroughbred, wrote, where, popularize, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: e, ii, am, my, think, world, war, ve,\n",
      "Nearest to when: asked, discussing, referring, especially, guessing, kidman, arrived, occurs,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, fertility, gentle, children, women,\n",
      "Nearest to dr: blok, kovacs, zaman, physician, insurgent, citation, sierpinski, positioned,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, reinforced, analogical, democratisation, elas, muskets,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, hazardous, tests, wastes, cricket, dumping, ozone, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, parton, living, long,\n",
      "Nearest to road: traffic, hypocrisy, extractive, hildebrand, routes, middlesex, athletics, interstate,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, ost,\n",
      "Epoch 7/10 Iteration: 97100 Avg. Training loss: 4.0471 0.2509 sec/batch\n",
      "Epoch 7/10 Iteration: 97200 Avg. Training loss: 4.0520 0.2493 sec/batch\n",
      "Epoch 7/10 Iteration: 97300 Avg. Training loss: 4.0143 0.2485 sec/batch\n",
      "Epoch 7/10 Iteration: 97400 Avg. Training loss: 4.0151 0.2476 sec/batch\n",
      "Epoch 7/10 Iteration: 97500 Avg. Training loss: 4.0506 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 97600 Avg. Training loss: 4.0054 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 97700 Avg. Training loss: 3.9664 0.2485 sec/batch\n",
      "Epoch 7/10 Iteration: 97800 Avg. Training loss: 4.0283 0.2490 sec/batch\n",
      "Epoch 7/10 Iteration: 97900 Avg. Training loss: 3.9369 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 98000 Avg. Training loss: 3.9934 0.2483 sec/batch\n",
      "Nearest to that: note, suggested, noted, suggest, argue, fact, argued, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, called, references, blacklisting, sprague,\n",
      "Nearest to he: himself, transmutation, where, thoroughbred, eton, popularize, nihonshoki, aps,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, interchangeably, denote,\n",
      "Nearest to i: ii, e, am, think, my, world, ve, war,\n",
      "Nearest to when: asked, discussing, kidman, guessing, referring, especially, arrived, occurs,\n",
      "Nearest to woman: sonnets, man, nationality, defensemen, gentle, fertility, children, women,\n",
      "Nearest to dr: blok, kovacs, zaman, physician, positioned, sierpinski, citation, insurgent,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, reinforced, analogical, democratisation, elas, paternoster,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, duddy, tornadoes, rt,\n",
      "Nearest to test: ban, cricket, tests, hazardous, matches, wastes, hoare, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, milked, worked, living, parton, beleaguered,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, donovan, farrar, interstate, routes,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 7/10 Iteration: 98100 Avg. Training loss: 4.0598 0.2516 sec/batch\n",
      "Epoch 7/10 Iteration: 98200 Avg. Training loss: 3.8386 0.2488 sec/batch\n",
      "Epoch 7/10 Iteration: 98300 Avg. Training loss: 4.0675 0.2488 sec/batch\n",
      "Epoch 7/10 Iteration: 98400 Avg. Training loss: 4.0644 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 98500 Avg. Training loss: 4.0107 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 98600 Avg. Training loss: 4.0000 0.2488 sec/batch\n",
      "Epoch 7/10 Iteration: 98700 Avg. Training loss: 3.9882 0.2490 sec/batch\n",
      "Epoch 7/10 Iteration: 98800 Avg. Training loss: 3.9223 0.2486 sec/batch\n",
      "Epoch 7/10 Iteration: 98900 Avg. Training loss: 4.0310 0.2500 sec/batch\n",
      "Epoch 7/10 Iteration: 99000 Avg. Training loss: 4.0491 0.2490 sec/batch\n",
      "Nearest to that: note, suggested, suggest, noted, fact, argue, claimed, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, external, references, called, blacklisting, sprague,\n",
      "Nearest to he: himself, transmutation, thoroughbred, wrote, where, vitis, eton, sevenfold,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, my, world, myself, ve,\n",
      "Nearest to when: asked, discussing, referring, arrived, kidman, guessing, especially, occurs,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, fertility, children, gentle, noun,\n",
      "Nearest to dr: blok, kovacs, zaman, sierpinski, positioned, berimbau, insurgent, acc,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, democratisation, paternoster, muskets, reinforced,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, tornadoes, rt, duddy,\n",
      "Nearest to test: ban, tests, hazardous, cricket, wastes, globes, dumping, ozone,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, parton, beleaguered,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, donovan, extractive, farrar, interstate, quarreling,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, eutelsat,\n",
      "Epoch 7/10 Iteration: 99100 Avg. Training loss: 4.0359 0.2519 sec/batch\n",
      "Epoch 7/10 Iteration: 99200 Avg. Training loss: 4.0472 0.2482 sec/batch\n",
      "Epoch 7/10 Iteration: 99300 Avg. Training loss: 4.0571 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 99400 Avg. Training loss: 4.0743 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 99500 Avg. Training loss: 4.0179 0.2486 sec/batch\n",
      "Epoch 7/10 Iteration: 99600 Avg. Training loss: 3.9652 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 99700 Avg. Training loss: 4.0476 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 99800 Avg. Training loss: 4.0618 0.2494 sec/batch\n",
      "Epoch 7/10 Iteration: 99900 Avg. Training loss: 4.0313 0.2476 sec/batch\n",
      "Epoch 7/10 Iteration: 100000 Avg. Training loss: 4.0243 0.2494 sec/batch\n",
      "Nearest to that: note, suggested, suggest, noted, fact, argue, claimed, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, external, references, called, blacklisting, sprague,\n",
      "Nearest to he: himself, thoroughbred, transmutation, where, wrote, eton, sevenfold, vitis,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, war, world, my, think, iii,\n",
      "Nearest to when: asked, discussing, especially, arrived, kidman, referring, occurs, guessing,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, fertility, gentle, children, noun,\n",
      "Nearest to dr: blok, kovacs, zaman, positioned, sierpinski, acc, insurgent, physician,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, democratisation, reinforced, paternoster, mawson,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, tornadoes, rt, duddy,\n",
      "Nearest to test: ban, tests, hazardous, cricket, wastes, globes, matches, dumping,\n",
      "Nearest to lived: short, eunuch, rabindranath, milked, worked, living, beleaguered, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, donovan, extractive, interstate, quarreling, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, castilla, indian, intelsat, eutelsat,\n",
      "Epoch 7/10 Iteration: 100100 Avg. Training loss: 4.0527 0.2519 sec/batch\n",
      "Epoch 7/10 Iteration: 100200 Avg. Training loss: 4.0598 0.2487 sec/batch\n",
      "Epoch 7/10 Iteration: 100300 Avg. Training loss: 4.0244 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 100400 Avg. Training loss: 4.0604 0.2480 sec/batch\n",
      "Epoch 7/10 Iteration: 100500 Avg. Training loss: 4.0336 0.2491 sec/batch\n",
      "Epoch 7/10 Iteration: 100600 Avg. Training loss: 4.0092 0.2501 sec/batch\n",
      "Epoch 7/10 Iteration: 100700 Avg. Training loss: 3.9137 0.2482 sec/batch\n",
      "Epoch 7/10 Iteration: 100800 Avg. Training loss: 4.0327 0.2481 sec/batch\n",
      "Epoch 7/10 Iteration: 100900 Avg. Training loss: 4.0038 0.2491 sec/batch\n",
      "Epoch 7/10 Iteration: 101000 Avg. Training loss: 3.9729 0.2480 sec/batch\n",
      "Nearest to that: note, suggested, suggest, noted, argue, fact, argued, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, external, references, called, blacklisting, siais,\n",
      "Nearest to he: himself, eton, thoroughbred, transmutation, where, vitis, sevenfold, nihonshoki,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, commonly, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, world, war, my, think, myself,\n",
      "Nearest to when: asked, discussing, referring, especially, kidman, guessing, arrived, occurs,\n",
      "Nearest to woman: sonnets, nationality, man, defensemen, fertility, children, gentle, noun,\n",
      "Nearest to dr: blok, zaman, kovacs, sierpinski, physician, positioned, acc, sword,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, analogical, democratisation, mostly, paternoster, mawson,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, duddy, sizeable,\n",
      "Nearest to test: ban, tests, hazardous, cricket, wastes, dumping, ozone, desertification,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, living, milked, kingdoms, long,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, donovan, roads, athletics, quarreling,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, ost,\n",
      "Epoch 7/10 Iteration: 101100 Avg. Training loss: 4.0189 0.2511 sec/batch\n",
      "Epoch 7/10 Iteration: 101200 Avg. Training loss: 4.0128 0.2482 sec/batch\n",
      "Epoch 7/10 Iteration: 101300 Avg. Training loss: 3.9758 0.2486 sec/batch\n",
      "Epoch 7/10 Iteration: 101400 Avg. Training loss: 4.0879 0.2490 sec/batch\n",
      "Epoch 7/10 Iteration: 101500 Avg. Training loss: 3.7542 0.2493 sec/batch\n",
      "Epoch 7/10 Iteration: 101600 Avg. Training loss: 3.6776 0.2482 sec/batch\n",
      "Epoch 7/10 Iteration: 101700 Avg. Training loss: 3.9356 0.2485 sec/batch\n",
      "Epoch 7/10 Iteration: 101800 Avg. Training loss: 3.8029 0.2477 sec/batch\n",
      "Epoch 7/10 Iteration: 101900 Avg. Training loss: 3.9369 0.2488 sec/batch\n",
      "Epoch 7/10 Iteration: 102000 Avg. Training loss: 4.0106 0.2483 sec/batch\n",
      "Nearest to that: suggested, note, suggest, argue, noted, argued, announced, fact,\n",
      "Nearest to not: did, does, do, necessarily, uncommon, but, yet, until,\n",
      "Nearest to also: see, list, known, called, blacklisting, references, external, lemnos,\n",
      "Nearest to he: himself, where, thoroughbred, wrote, attended, transmutation, sevenfold, eton,\n",
      "Nearest to all: virtually, nearly, listing, almost, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, am, e, my, world, think, myself, war,\n",
      "Nearest to when: asked, arrived, discussing, kidman, referring, especially, occurs, guessing,\n",
      "Nearest to woman: man, nationality, sonnets, defensemen, fertility, gentle, children, noun,\n",
      "Nearest to dr: blok, zaman, kovacs, physician, sierpinski, acc, citation, positioned,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, democratisation, mostly, paternoster, evangelism,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, tests, hazardous, cricket, globes, wastes, matches, obsoletes,\n",
      "Nearest to lived: short, eunuch, worked, rabindranath, milked, living, parton, long,\n",
      "Nearest to road: hypocrisy, traffic, hildebrand, donovan, extractive, quarreling, athletics, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, pacific, indian, castilla, intelsat, ost,\n",
      "Epoch 7/10 Iteration: 102100 Avg. Training loss: 4.0260 0.2521 sec/batch\n",
      "Epoch 7/10 Iteration: 102200 Avg. Training loss: 4.0651 0.2492 sec/batch\n",
      "Epoch 7/10 Iteration: 102300 Avg. Training loss: 3.9908 0.2496 sec/batch\n",
      "Epoch 7/10 Iteration: 102400 Avg. Training loss: 4.0813 0.2487 sec/batch\n",
      "Epoch 7/10 Iteration: 102500 Avg. Training loss: 4.0721 0.2500 sec/batch\n",
      "Epoch 7/10 Iteration: 102600 Avg. Training loss: 4.0949 0.2499 sec/batch\n",
      "Epoch 7/10 Iteration: 102700 Avg. Training loss: 4.0473 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 102800 Avg. Training loss: 4.0528 0.2487 sec/batch\n",
      "Epoch 7/10 Iteration: 102900 Avg. Training loss: 3.8959 0.2502 sec/batch\n",
      "Epoch 7/10 Iteration: 103000 Avg. Training loss: 4.0144 0.2490 sec/batch\n",
      "Nearest to that: note, suggested, suggest, argue, argued, announced, fact, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, references, called, blacklisting, daiko,\n",
      "Nearest to he: himself, where, attended, wrote, thoroughbred, eton, sevenfold, whom,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, commonly, refer, widely, term, denote, frequently,\n",
      "Nearest to i: ii, e, am, world, think, my, myself, war,\n",
      "Nearest to when: asked, discussing, arrived, kidman, referring, especially, happens, mitra,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, gentle, noun, children,\n",
      "Nearest to dr: blok, zaman, kovacs, acc, positioned, sierpinski, physician, rcn,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, mostly, democratisation, reinforced, cerebrospinal,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, sizeable, small,\n",
      "Nearest to test: ban, hazardous, tests, cricket, dumping, ozone, wastes, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, parton, beleaguered,\n",
      "Nearest to road: hypocrisy, traffic, hildebrand, donovan, extractive, quarreling, athletics, interstate,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, intelsat, castilla, ost,\n",
      "Epoch 7/10 Iteration: 103100 Avg. Training loss: 4.0227 0.2512 sec/batch\n",
      "Epoch 7/10 Iteration: 103200 Avg. Training loss: 4.0193 0.2494 sec/batch\n",
      "Epoch 7/10 Iteration: 103300 Avg. Training loss: 3.9850 0.2484 sec/batch\n",
      "Epoch 7/10 Iteration: 103400 Avg. Training loss: 4.0174 0.2480 sec/batch\n",
      "Epoch 7/10 Iteration: 103500 Avg. Training loss: 4.0355 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 103600 Avg. Training loss: 4.0754 0.2481 sec/batch\n",
      "Epoch 7/10 Iteration: 103700 Avg. Training loss: 3.9885 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 103800 Avg. Training loss: 4.0259 0.2498 sec/batch\n",
      "Epoch 7/10 Iteration: 103900 Avg. Training loss: 3.8501 0.2479 sec/batch\n",
      "Epoch 7/10 Iteration: 104000 Avg. Training loss: 3.8632 0.2480 sec/batch\n",
      "Nearest to that: note, suggested, suggest, argue, fact, announced, claimed, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, references, called, links, daiko,\n",
      "Nearest to he: himself, where, wrote, attended, sevenfold, thoroughbred, whom, eton,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, commonly, widely, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, myself,\n",
      "Nearest to when: asked, discussing, arrived, kidman, happens, referring, especially, occurs,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, children, noun, women,\n",
      "Nearest to dr: blok, zaman, acc, positioned, physician, sierpinski, kovacs, insurgent,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, mostly, analogical, reinforced, democratisation, cerebrospinal,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, small, workforce,\n",
      "Nearest to test: ban, hazardous, tests, dumping, cricket, ozone, wastes, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, monks, parton,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, donovan, extractive, quarreling, routes, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, indian, pacific, posed, intelsat, castilla, ost,\n",
      "Epoch 7/10 Iteration: 104100 Avg. Training loss: 4.0284 0.2510 sec/batch\n",
      "Epoch 7/10 Iteration: 104200 Avg. Training loss: 4.0142 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 104300 Avg. Training loss: 3.8531 0.2488 sec/batch\n",
      "Epoch 7/10 Iteration: 104400 Avg. Training loss: 4.0148 0.2489 sec/batch\n",
      "Epoch 7/10 Iteration: 104500 Avg. Training loss: 4.0503 0.2482 sec/batch\n",
      "Epoch 7/10 Iteration: 104600 Avg. Training loss: 4.0067 0.2490 sec/batch\n",
      "Epoch 7/10 Iteration: 104700 Avg. Training loss: 3.9915 0.2493 sec/batch\n",
      "Epoch 7/10 Iteration: 104800 Avg. Training loss: 3.9563 0.2483 sec/batch\n",
      "Epoch 7/10 Iteration: 104900 Avg. Training loss: 3.9984 0.2480 sec/batch\n",
      "Epoch 7/10 Iteration: 105000 Avg. Training loss: 4.0188 0.2486 sec/batch\n",
      "Nearest to that: note, suggested, argue, suggest, fact, announced, noted, claimed,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, references, called, daiko, denslow,\n",
      "Nearest to he: himself, where, wrote, attended, sevenfold, thoroughbred, whom, dike,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, commonly, widely, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, world, my, war, think, iii,\n",
      "Nearest to when: asked, discussing, arrived, happens, guessing, especially, referring, kidman,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, women, children, noun,\n",
      "Nearest to dr: blok, zaman, sierpinski, positioned, physician, acc, rcn, kovacs,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, reinforced, democratisation, mostly, cerebrospinal,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, duddy, small,\n",
      "Nearest to test: ban, tests, hazardous, cricket, obsoletes, wastes, dumping, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, kingdoms, monks,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, donovan, routes, athletics, middlesex,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, castilla, intelsat, ost,\n",
      "Epoch 7/10 Iteration: 105100 Avg. Training loss: 4.0467 0.2529 sec/batch\n",
      "Epoch 7/10 Iteration: 105200 Avg. Training loss: 4.0894 0.2483 sec/batch\n",
      "Epoch 8/10 Iteration: 105300 Avg. Training loss: 4.0605 0.1726 sec/batch\n",
      "Epoch 8/10 Iteration: 105400 Avg. Training loss: 4.0378 0.2497 sec/batch\n",
      "Epoch 8/10 Iteration: 105500 Avg. Training loss: 3.9619 0.2482 sec/batch\n",
      "Epoch 8/10 Iteration: 105600 Avg. Training loss: 4.0359 0.2494 sec/batch\n",
      "Epoch 8/10 Iteration: 105700 Avg. Training loss: 4.0270 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 105800 Avg. Training loss: 3.9562 0.2494 sec/batch\n",
      "Epoch 8/10 Iteration: 105900 Avg. Training loss: 3.9848 0.2490 sec/batch\n",
      "Epoch 8/10 Iteration: 106000 Avg. Training loss: 4.0227 0.2493 sec/batch\n",
      "Nearest to that: suggested, note, fact, argue, suggest, claimed, believe, noted,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, called, daiko, denslow, references,\n",
      "Nearest to he: himself, wrote, where, whom, sevenfold, thoroughbred, nihonshoki, attended,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, my, world, iii, myself,\n",
      "Nearest to when: asked, discussing, arrived, referring, guessing, especially, happens, kidman,\n",
      "Nearest to woman: nationality, sonnets, man, defensemen, women, fertility, children, noun,\n",
      "Nearest to dr: blok, sierpinski, kovacs, zaman, positioned, rcn, physician, acc,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, democratisation, mostly, cerebrospinal, reinforced,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, duddy, beltline,\n",
      "Nearest to test: ban, tests, cricket, hazardous, matches, globes, obsoletes, hoare,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, kingdoms, long,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, routes, racing, donovan, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, castilla, intelsat, ost,\n",
      "Epoch 8/10 Iteration: 106100 Avg. Training loss: 3.9211 0.2510 sec/batch\n",
      "Epoch 8/10 Iteration: 106200 Avg. Training loss: 3.9348 0.2511 sec/batch\n",
      "Epoch 8/10 Iteration: 106300 Avg. Training loss: 3.9559 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 106400 Avg. Training loss: 4.0348 0.2487 sec/batch\n",
      "Epoch 8/10 Iteration: 106500 Avg. Training loss: 4.0240 0.2477 sec/batch\n",
      "Epoch 8/10 Iteration: 106600 Avg. Training loss: 4.0060 0.2469 sec/batch\n",
      "Epoch 8/10 Iteration: 106700 Avg. Training loss: 4.0022 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 106800 Avg. Training loss: 4.0098 0.2496 sec/batch\n",
      "Epoch 8/10 Iteration: 106900 Avg. Training loss: 4.0270 0.2479 sec/batch\n",
      "Epoch 8/10 Iteration: 107000 Avg. Training loss: 4.0414 0.2488 sec/batch\n",
      "Nearest to that: suggested, note, suggest, fact, argue, claimed, noted, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, external, denslow, daiko, references,\n",
      "Nearest to he: himself, where, wrote, whom, nihonshoki, sevenfold, attended, thoroughbred,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, my, world, iii, think, war,\n",
      "Nearest to when: asked, discussing, referring, arrived, guessing, occurs, especially, happens,\n",
      "Nearest to woman: nationality, sonnets, man, defensemen, fertility, noun, children, gentle,\n",
      "Nearest to dr: blok, zaman, sierpinski, positioned, rcn, kovacs, physician, acc,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, cerebrospinal, reinforced, mostly, muskets,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, duddy, beltline,\n",
      "Nearest to test: ban, tests, cricket, hazardous, matches, globes, hoare, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, kingdoms, monks,\n",
      "Nearest to road: hypocrisy, traffic, hildebrand, routes, extractive, racing, accident, roads,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, castilla, intelsat, ost,\n",
      "Epoch 8/10 Iteration: 107100 Avg. Training loss: 4.0414 0.2511 sec/batch\n",
      "Epoch 8/10 Iteration: 107200 Avg. Training loss: 4.0149 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 107300 Avg. Training loss: 4.0219 0.2511 sec/batch\n",
      "Epoch 8/10 Iteration: 107400 Avg. Training loss: 3.9093 0.2487 sec/batch\n",
      "Epoch 8/10 Iteration: 107500 Avg. Training loss: 3.9717 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 107600 Avg. Training loss: 4.0388 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 107700 Avg. Training loss: 4.0537 0.2497 sec/batch\n",
      "Epoch 8/10 Iteration: 107800 Avg. Training loss: 4.0444 0.2497 sec/batch\n",
      "Epoch 8/10 Iteration: 107900 Avg. Training loss: 4.0475 0.2472 sec/batch\n",
      "Epoch 8/10 Iteration: 108000 Avg. Training loss: 4.0604 0.2489 sec/batch\n",
      "Nearest to that: note, suggested, argue, fact, suggest, noted, claimed, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, daiko, external, denslow, references,\n",
      "Nearest to he: himself, where, thoroughbred, wrote, attended, resurrecting, orcinus, whom,\n",
      "Nearest to all: almost, virtually, nearly, listing, practically, assorted, things, except,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, my, world, war, iii,\n",
      "Nearest to when: asked, discussing, guessing, referring, arrived, kidman, occurs, happens,\n",
      "Nearest to woman: nationality, sonnets, man, defensemen, fertility, children, noun, gentle,\n",
      "Nearest to dr: blok, sierpinski, zaman, kovacs, physician, positioned, rcn, acc,\n",
      "Nearest to mainly: primarily, emulsions, exclusively, analogical, reinforced, mostly, cerebrospinal, twelver,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, duddy, tornadoes,\n",
      "Nearest to test: ban, tests, cricket, hazardous, matches, globes, hoare, dumping,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, kingdoms, monks, living,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, routes, extractive, racing, interstate, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, castilla, intelsat, ost,\n",
      "Epoch 8/10 Iteration: 108100 Avg. Training loss: 4.0103 0.2513 sec/batch\n",
      "Epoch 8/10 Iteration: 108200 Avg. Training loss: 4.0109 0.2492 sec/batch\n",
      "Epoch 8/10 Iteration: 108300 Avg. Training loss: 4.0477 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 108400 Avg. Training loss: 4.0602 0.2497 sec/batch\n",
      "Epoch 8/10 Iteration: 108500 Avg. Training loss: 4.0360 0.2495 sec/batch\n",
      "Epoch 8/10 Iteration: 108600 Avg. Training loss: 4.0333 0.2479 sec/batch\n",
      "Epoch 8/10 Iteration: 108700 Avg. Training loss: 3.9903 0.2494 sec/batch\n",
      "Epoch 8/10 Iteration: 108800 Avg. Training loss: 3.9967 0.2482 sec/batch\n",
      "Epoch 8/10 Iteration: 108900 Avg. Training loss: 3.8779 0.2482 sec/batch\n",
      "Epoch 8/10 Iteration: 109000 Avg. Training loss: 3.8940 0.2502 sec/batch\n",
      "Nearest to that: note, suggested, argue, suggest, fact, noted, argues, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, called, daiko, references, denslow,\n",
      "Nearest to he: himself, where, thoroughbred, wrote, eton, transmutation, attended, resurrecting,\n",
      "Nearest to all: almost, virtually, nearly, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, denote, term, interchangeably,\n",
      "Nearest to i: ii, e, am, think, my, world, war, myself,\n",
      "Nearest to when: asked, discussing, referring, guessing, occurs, arrived, especially, kidman,\n",
      "Nearest to woman: nationality, sonnets, defensemen, man, fertility, children, noun, women,\n",
      "Nearest to dr: blok, physician, sierpinski, zaman, positioned, kovacs, acc, rcn,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, mostly, analogical, reinforced, twelver, democratisation,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, duddy, tornadoes,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, matches, wastes, dumping,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, long, kingdoms,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, routes, interstate, extractive, racing, athletics,\n",
      "Nearest to ocean: atlantic, intersputnik, indian, posed, pacific, intelsat, castilla, ost,\n",
      "Epoch 8/10 Iteration: 109100 Avg. Training loss: 4.0355 0.2520 sec/batch\n",
      "Epoch 8/10 Iteration: 109200 Avg. Training loss: 3.9920 0.2476 sec/batch\n",
      "Epoch 8/10 Iteration: 109300 Avg. Training loss: 4.0132 0.2500 sec/batch\n",
      "Epoch 8/10 Iteration: 109400 Avg. Training loss: 3.9755 0.2481 sec/batch\n",
      "Epoch 8/10 Iteration: 109500 Avg. Training loss: 4.0200 0.2481 sec/batch\n",
      "Epoch 8/10 Iteration: 109600 Avg. Training loss: 4.0130 0.2483 sec/batch\n",
      "Epoch 8/10 Iteration: 109700 Avg. Training loss: 3.8895 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 109800 Avg. Training loss: 4.0306 0.2487 sec/batch\n",
      "Epoch 8/10 Iteration: 109900 Avg. Training loss: 4.0151 0.2495 sec/batch\n",
      "Epoch 8/10 Iteration: 110000 Avg. Training loss: 4.0427 0.2484 sec/batch\n",
      "Nearest to that: note, suggested, argue, suggest, noted, fact, argues, claimed,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, daiko, external, references, denslow,\n",
      "Nearest to he: himself, where, thoroughbred, wrote, transmutation, orcinus, whom, nihonshoki,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, denote, term, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, ve,\n",
      "Nearest to when: asked, discussing, referring, guessing, arrived, occurs, happens, kidman,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, women, children, gentle,\n",
      "Nearest to dr: blok, physician, kovacs, positioned, sierpinski, acc, zaman, rcn,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, analogical, mostly, democratisation, reinforced, cerebrospinal,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, duddy, tornadoes,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, matches, wastes, dumping,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, milked, living, kingdoms, cannibal,\n",
      "Nearest to road: hypocrisy, traffic, hildebrand, athletics, donovan, extractive, interstate, routes,\n",
      "Nearest to ocean: atlantic, posed, intersputnik, indian, castilla, pacific, intelsat, ost,\n",
      "Epoch 8/10 Iteration: 110100 Avg. Training loss: 4.0340 0.2517 sec/batch\n",
      "Epoch 8/10 Iteration: 110200 Avg. Training loss: 4.0205 0.2503 sec/batch\n",
      "Epoch 8/10 Iteration: 110300 Avg. Training loss: 4.0208 0.2486 sec/batch\n",
      "Epoch 8/10 Iteration: 110400 Avg. Training loss: 4.0151 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 110500 Avg. Training loss: 4.0239 0.2491 sec/batch\n",
      "Epoch 8/10 Iteration: 110600 Avg. Training loss: 4.0754 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 110700 Avg. Training loss: 4.0548 0.2486 sec/batch\n",
      "Epoch 8/10 Iteration: 110800 Avg. Training loss: 3.9416 0.2487 sec/batch\n",
      "Epoch 8/10 Iteration: 110900 Avg. Training loss: 4.0367 0.2490 sec/batch\n",
      "Epoch 8/10 Iteration: 111000 Avg. Training loss: 3.9294 0.2478 sec/batch\n",
      "Nearest to that: note, suggested, argue, fact, noted, suggest, argued, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, confused,\n",
      "Nearest to also: see, list, known, called, external, blacklisting, daiko, references,\n",
      "Nearest to he: himself, where, thoroughbred, transmutation, wrote, popularize, whom, vitis,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, denote, term, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, iii,\n",
      "Nearest to when: asked, discussing, referring, guessing, especially, occurs, kidman, arrived,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, women, children, noun,\n",
      "Nearest to dr: blok, kovacs, physician, sierpinski, rcn, positioned, acc, zaman,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, analogical, democratisation, reinforced, mostly, muskets,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, wastes, dumping, matches,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, long, milked, story, kingdoms,\n",
      "Nearest to road: hypocrisy, traffic, hildebrand, athletics, routes, donovan, middlesex, interstate,\n",
      "Nearest to ocean: atlantic, intersputnik, posed, indian, pacific, castilla, intelsat, ost,\n",
      "Epoch 8/10 Iteration: 111100 Avg. Training loss: 4.0065 0.2516 sec/batch\n",
      "Epoch 8/10 Iteration: 111200 Avg. Training loss: 4.0509 0.2490 sec/batch\n",
      "Epoch 8/10 Iteration: 111300 Avg. Training loss: 4.0276 0.2476 sec/batch\n",
      "Epoch 8/10 Iteration: 111400 Avg. Training loss: 4.0342 0.2498 sec/batch\n",
      "Epoch 8/10 Iteration: 111500 Avg. Training loss: 4.0581 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 111600 Avg. Training loss: 4.0505 0.2493 sec/batch\n",
      "Epoch 8/10 Iteration: 111700 Avg. Training loss: 3.9994 0.2482 sec/batch\n",
      "Epoch 8/10 Iteration: 111800 Avg. Training loss: 3.9159 0.2491 sec/batch\n",
      "Epoch 8/10 Iteration: 111900 Avg. Training loss: 4.0352 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 112000 Avg. Training loss: 3.9552 0.2492 sec/batch\n",
      "Nearest to that: note, argue, suggested, suggest, noted, argues, fact, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, external, references, blacklisting, sprague,\n",
      "Nearest to he: himself, transmutation, wrote, where, thoroughbred, popularize, acknowledges, sevenfold,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, denote, commonly, term, interchangeably,\n",
      "Nearest to i: e, ii, am, think, my, world, war, myself,\n",
      "Nearest to when: asked, discussing, referring, guessing, especially, occurs, happens, arrived,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, women, children, noun,\n",
      "Nearest to dr: blok, kovacs, physician, positioned, sierpinski, rcn, zaman, acc,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, mostly, analogical, democratisation, largely, hazara,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, tests, cricket, hazardous, wastes, globes, dumping, desertification,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, long, milked, story, coney,\n",
      "Nearest to road: hypocrisy, traffic, routes, hildebrand, interstate, athletics, extractive, middlesex,\n",
      "Nearest to ocean: atlantic, intersputnik, indian, posed, pacific, castilla, intelsat, eutelsat,\n",
      "Epoch 8/10 Iteration: 112100 Avg. Training loss: 4.0231 0.2516 sec/batch\n",
      "Epoch 8/10 Iteration: 112200 Avg. Training loss: 4.0397 0.2481 sec/batch\n",
      "Epoch 8/10 Iteration: 112300 Avg. Training loss: 4.0303 0.2493 sec/batch\n",
      "Epoch 8/10 Iteration: 112400 Avg. Training loss: 3.9838 0.2497 sec/batch\n",
      "Epoch 8/10 Iteration: 112500 Avg. Training loss: 4.0443 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 112600 Avg. Training loss: 4.0118 0.2475 sec/batch\n",
      "Epoch 8/10 Iteration: 112700 Avg. Training loss: 3.9723 0.2492 sec/batch\n",
      "Epoch 8/10 Iteration: 112800 Avg. Training loss: 4.0119 0.2486 sec/batch\n",
      "Epoch 8/10 Iteration: 112900 Avg. Training loss: 3.9830 0.2479 sec/batch\n",
      "Epoch 8/10 Iteration: 113000 Avg. Training loss: 3.9767 0.2471 sec/batch\n",
      "Nearest to that: note, suggested, suggest, argue, noted, argued, fact, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, external, references, blacklisting, sprague,\n",
      "Nearest to he: himself, where, transmutation, nihonshoki, sevenfold, thoroughbred, whom, wrote,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, denote, commonly, term, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, war, my, myself,\n",
      "Nearest to when: asked, discussing, guessing, especially, referring, occurs, arrived, kidman,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, women, fertility, children, gentle,\n",
      "Nearest to dr: blok, kovacs, physician, acc, rcn, sierpinski, surgery, positioned,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, analogical, mostly, democratisation, elas, reinforced,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, dumping, matches, wastes,\n",
      "Nearest to lived: short, eunuch, rabindranath, milked, long, worked, living, monks,\n",
      "Nearest to road: hypocrisy, traffic, hildebrand, extractive, athletics, routes, interstate, donovan,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, indian, posed, castilla, intelsat, eutelsat,\n",
      "Epoch 8/10 Iteration: 113100 Avg. Training loss: 4.0119 0.2512 sec/batch\n",
      "Epoch 8/10 Iteration: 113200 Avg. Training loss: 3.9334 0.2500 sec/batch\n",
      "Epoch 8/10 Iteration: 113300 Avg. Training loss: 3.9691 0.2479 sec/batch\n",
      "Epoch 8/10 Iteration: 113400 Avg. Training loss: 4.0486 0.2492 sec/batch\n",
      "Epoch 8/10 Iteration: 113500 Avg. Training loss: 4.0421 0.2494 sec/batch\n",
      "Epoch 8/10 Iteration: 113600 Avg. Training loss: 3.9780 0.2497 sec/batch\n",
      "Epoch 8/10 Iteration: 113700 Avg. Training loss: 4.0049 0.2491 sec/batch\n",
      "Epoch 8/10 Iteration: 113800 Avg. Training loss: 3.8939 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 113900 Avg. Training loss: 4.0535 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 114000 Avg. Training loss: 4.0328 0.2481 sec/batch\n",
      "Nearest to that: note, suggested, suggest, noted, argue, argued, fact, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, external, references, sprague, blacklisting,\n",
      "Nearest to he: himself, transmutation, where, wrote, thoroughbred, whom, nihonshoki, sevenfold,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, term, denote, commonly, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, myself,\n",
      "Nearest to when: asked, discussing, especially, arrived, guessing, referring, kidman, occurs,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, noun, women, children,\n",
      "Nearest to dr: blok, kovacs, sierpinski, physician, positioned, zaman, acc, surgery,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, mostly, analogical, democratisation, largely, elas,\n",
      "Nearest to scale: large, richter, scales, casualties, neu, tornadoes, rt, celsius,\n",
      "Nearest to test: ban, tests, hazardous, cricket, globes, wastes, dumping, obsoletes,\n",
      "Nearest to lived: short, eunuch, milked, rabindranath, worked, long, living, monks,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, athletics, accident, interstate, routes,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, indian, posed, intelsat, castilla, eutelsat,\n",
      "Epoch 8/10 Iteration: 114100 Avg. Training loss: 4.0262 0.2513 sec/batch\n",
      "Epoch 8/10 Iteration: 114200 Avg. Training loss: 4.0160 0.2482 sec/batch\n",
      "Epoch 8/10 Iteration: 114300 Avg. Training loss: 4.0405 0.2500 sec/batch\n",
      "Epoch 8/10 Iteration: 114400 Avg. Training loss: 4.0820 0.2497 sec/batch\n",
      "Epoch 8/10 Iteration: 114500 Avg. Training loss: 4.0188 0.2498 sec/batch\n",
      "Epoch 8/10 Iteration: 114600 Avg. Training loss: 3.9543 0.2491 sec/batch\n",
      "Epoch 8/10 Iteration: 114700 Avg. Training loss: 4.0335 0.2502 sec/batch\n",
      "Epoch 8/10 Iteration: 114800 Avg. Training loss: 4.0695 0.2495 sec/batch\n",
      "Epoch 8/10 Iteration: 114900 Avg. Training loss: 4.0064 0.2473 sec/batch\n",
      "Epoch 8/10 Iteration: 115000 Avg. Training loss: 4.0133 0.2474 sec/batch\n",
      "Nearest to that: note, suggested, suggest, fact, noted, argue, argued, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, called, references, sprague, blacklisting,\n",
      "Nearest to he: himself, transmutation, where, whom, nihonshoki, sevenfold, wrote, thoroughbred,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, denote, commonly, term, interchangeably,\n",
      "Nearest to i: ii, e, am, think, war, world, my, iii,\n",
      "Nearest to when: asked, discussing, arrived, especially, referring, guessing, kidman, occurs,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, women, noun, children,\n",
      "Nearest to dr: blok, physician, sierpinski, positioned, acc, kovacs, zaman, surgery,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, mostly, analogical, democratisation, elas, evangelism,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, tornadoes, rt, workforce,\n",
      "Nearest to test: ban, tests, hazardous, cricket, dumping, wastes, globes, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, milked, monks, living, worked, story,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, accident, interstate, donovan, farrar,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, indian, intelsat, castilla, eutelsat,\n",
      "Epoch 8/10 Iteration: 115100 Avg. Training loss: 4.0271 0.2515 sec/batch\n",
      "Epoch 8/10 Iteration: 115200 Avg. Training loss: 4.0472 0.2486 sec/batch\n",
      "Epoch 8/10 Iteration: 115300 Avg. Training loss: 4.0350 0.2487 sec/batch\n",
      "Epoch 8/10 Iteration: 115400 Avg. Training loss: 4.0439 0.2486 sec/batch\n",
      "Epoch 8/10 Iteration: 115500 Avg. Training loss: 4.0442 0.2491 sec/batch\n",
      "Epoch 8/10 Iteration: 115600 Avg. Training loss: 4.0183 0.2495 sec/batch\n",
      "Epoch 8/10 Iteration: 115700 Avg. Training loss: 3.9496 0.2495 sec/batch\n",
      "Epoch 8/10 Iteration: 115800 Avg. Training loss: 3.9864 0.2487 sec/batch\n",
      "Epoch 8/10 Iteration: 115900 Avg. Training loss: 4.0116 0.2478 sec/batch\n",
      "Epoch 8/10 Iteration: 116000 Avg. Training loss: 4.0090 0.2486 sec/batch\n",
      "Nearest to that: note, suggest, suggested, noted, argue, argued, fact, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, external, references, but, blacklisting,\n",
      "Nearest to he: himself, transmutation, thoroughbred, where, nihonshoki, wrote, acknowledges, sevenfold,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, denote, commonly, term, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, myself,\n",
      "Nearest to when: asked, discussing, referring, especially, occurs, arrived, kidman, guessing,\n",
      "Nearest to woman: nationality, sonnets, man, fertility, defensemen, children, women, noun,\n",
      "Nearest to dr: blok, physician, positioned, zaman, sierpinski, acc, kovacs, surgery,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, mostly, democratisation, analogical, hazara, largely,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, tests, hazardous, cricket, dumping, wastes, desertification, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, milked, living, long, inhabited, monks,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, accident, interstate, athletics, quarreling,\n",
      "Nearest to ocean: atlantic, posed, intersputnik, pacific, indian, castilla, intelsat, ost,\n",
      "Epoch 8/10 Iteration: 116100 Avg. Training loss: 3.9813 0.2511 sec/batch\n",
      "Epoch 8/10 Iteration: 116200 Avg. Training loss: 3.9855 0.2486 sec/batch\n",
      "Epoch 8/10 Iteration: 116300 Avg. Training loss: 3.9898 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 116400 Avg. Training loss: 4.0691 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 116500 Avg. Training loss: 3.9135 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 116600 Avg. Training loss: 3.5700 0.2485 sec/batch\n",
      "Epoch 8/10 Iteration: 116700 Avg. Training loss: 3.8969 0.2483 sec/batch\n",
      "Epoch 8/10 Iteration: 116800 Avg. Training loss: 3.7767 0.2479 sec/batch\n",
      "Epoch 8/10 Iteration: 116900 Avg. Training loss: 3.9397 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 117000 Avg. Training loss: 3.9745 0.2493 sec/batch\n",
      "Nearest to that: note, suggest, suggested, argued, noted, argue, announced, fact,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, references, he, blacklisting, lemnos,\n",
      "Nearest to he: himself, where, thoroughbred, wrote, transmutation, also, nihonshoki, discontented,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, denote, term, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, myself,\n",
      "Nearest to when: asked, discussing, arrived, kidman, referring, especially, occurs, happens,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, fertility, women, children, noun,\n",
      "Nearest to dr: blok, physician, kovacs, zaman, sierpinski, positioned, acc, surgery,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, mostly, analogical, elas, democratisation, predominantly,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, tests, hazardous, cricket, dumping, globes, wastes, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, story, long, milked, worked, living,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, accident, donovan, extractive, interstate, waterbury,\n",
      "Nearest to ocean: atlantic, posed, pacific, intersputnik, indian, intelsat, castilla, ost,\n",
      "Epoch 8/10 Iteration: 117100 Avg. Training loss: 4.0043 0.2517 sec/batch\n",
      "Epoch 8/10 Iteration: 117200 Avg. Training loss: 4.0144 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 117300 Avg. Training loss: 4.0081 0.2490 sec/batch\n",
      "Epoch 8/10 Iteration: 117400 Avg. Training loss: 4.0770 0.2498 sec/batch\n",
      "Epoch 8/10 Iteration: 117500 Avg. Training loss: 4.0676 0.2504 sec/batch\n",
      "Epoch 8/10 Iteration: 117600 Avg. Training loss: 4.0826 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 117700 Avg. Training loss: 4.0746 0.2494 sec/batch\n",
      "Epoch 8/10 Iteration: 117800 Avg. Training loss: 4.0225 0.2480 sec/batch\n",
      "Epoch 8/10 Iteration: 117900 Avg. Training loss: 3.9787 0.2498 sec/batch\n",
      "Epoch 8/10 Iteration: 118000 Avg. Training loss: 3.8847 0.2479 sec/batch\n",
      "Nearest to that: note, suggest, suggested, announced, argue, argued, fact, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, references, external, called, below, but,\n",
      "Nearest to he: himself, where, wrote, thoroughbred, acknowledges, attended, sevenfold, cheltenham,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, frequently,\n",
      "Nearest to i: ii, e, am, think, world, my, war, myself,\n",
      "Nearest to when: asked, discussing, arrived, referring, happens, kidman, especially, occurs,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, fertility, women, noun, children,\n",
      "Nearest to dr: blok, physician, kovacs, zaman, surgery, positioned, sierpinski, rcn,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, largely, predominantly, analogical, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, neu, rt, tornadoes, duddy,\n",
      "Nearest to test: ban, tests, hazardous, dumping, cricket, globes, desertification, wastes,\n",
      "Nearest to lived: short, eunuch, rabindranath, story, milked, living, worked, long,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, interstate, routes, accident, quarreling,\n",
      "Nearest to ocean: atlantic, intersputnik, pacific, posed, indian, intelsat, castilla, ost,\n",
      "Epoch 8/10 Iteration: 118100 Avg. Training loss: 4.0207 0.2511 sec/batch\n",
      "Epoch 8/10 Iteration: 118200 Avg. Training loss: 4.0350 0.2493 sec/batch\n",
      "Epoch 8/10 Iteration: 118300 Avg. Training loss: 3.9333 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 118400 Avg. Training loss: 4.0631 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 118500 Avg. Training loss: 3.9794 0.2483 sec/batch\n",
      "Epoch 8/10 Iteration: 118600 Avg. Training loss: 4.0774 0.2496 sec/batch\n",
      "Epoch 8/10 Iteration: 118700 Avg. Training loss: 3.9798 0.2490 sec/batch\n",
      "Epoch 8/10 Iteration: 118800 Avg. Training loss: 4.0285 0.2487 sec/batch\n",
      "Epoch 8/10 Iteration: 118900 Avg. Training loss: 3.8916 0.2477 sec/batch\n",
      "Epoch 8/10 Iteration: 119000 Avg. Training loss: 3.7994 0.2494 sec/batch\n",
      "Nearest to that: note, suggest, fact, suggested, argue, announced, believe, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, external, references, called, below, but,\n",
      "Nearest to he: himself, where, wrote, thoroughbred, nihonshoki, sevenfold, cheltenham, acknowledges,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, commonly, term, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, war, world, my, myself,\n",
      "Nearest to when: asked, discussing, arrived, especially, referring, happens, kidman, guessing,\n",
      "Nearest to woman: nationality, defensemen, man, fertility, sonnets, children, noun, women,\n",
      "Nearest to dr: blok, physician, zaman, kovacs, positioned, sierpinski, acc, rcn,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, elas, predominantly, democratisation,\n",
      "Nearest to scale: large, richter, scales, casualties, rt, neu, tornadoes, small,\n",
      "Nearest to test: ban, tests, hazardous, cricket, dumping, globes, wastes, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, long, milked, story, coney, monks,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, accident, interstate, routes, vehicles,\n",
      "Nearest to ocean: atlantic, indian, pacific, intersputnik, posed, intelsat, castilla, ost,\n",
      "Epoch 8/10 Iteration: 119100 Avg. Training loss: 4.0061 0.2518 sec/batch\n",
      "Epoch 8/10 Iteration: 119200 Avg. Training loss: 3.9955 0.2490 sec/batch\n",
      "Epoch 8/10 Iteration: 119300 Avg. Training loss: 3.9247 0.2481 sec/batch\n",
      "Epoch 8/10 Iteration: 119400 Avg. Training loss: 3.9515 0.2488 sec/batch\n",
      "Epoch 8/10 Iteration: 119500 Avg. Training loss: 3.9967 0.2484 sec/batch\n",
      "Epoch 8/10 Iteration: 119600 Avg. Training loss: 4.0218 0.2482 sec/batch\n",
      "Epoch 8/10 Iteration: 119700 Avg. Training loss: 4.0396 0.2482 sec/batch\n",
      "Epoch 8/10 Iteration: 119800 Avg. Training loss: 3.9461 0.2492 sec/batch\n",
      "Epoch 8/10 Iteration: 119900 Avg. Training loss: 3.9162 0.2489 sec/batch\n",
      "Epoch 8/10 Iteration: 120000 Avg. Training loss: 4.0637 0.2494 sec/batch\n",
      "Nearest to that: note, suggest, fact, argue, suggested, announced, noted, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, confused,\n",
      "Nearest to also: see, list, known, references, external, called, daiko, but,\n",
      "Nearest to he: himself, where, wrote, thoroughbred, cheltenham, suv, whom, attended,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, term, commonly, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, war, my, iii,\n",
      "Nearest to when: asked, discussing, arrived, especially, guessing, referring, happens, occurs,\n",
      "Nearest to woman: nationality, defensemen, man, sonnets, children, women, fertility, noun,\n",
      "Nearest to dr: blok, physician, positioned, sierpinski, zaman, kovacs, acc, rcn,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, elas, predominantly, democratisation, twelver,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, neu, tornadoes, bokuto,\n",
      "Nearest to test: ban, tests, cricket, hazardous, obsoletes, globes, dumping, matches,\n",
      "Nearest to lived: short, eunuch, rabindranath, long, coney, milked, living, story,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, extractive, accident, routes, donovan, interstate,\n",
      "Nearest to ocean: atlantic, indian, pacific, intersputnik, posed, intelsat, castilla, ost,\n",
      "Epoch 8/10 Iteration: 120100 Avg. Training loss: 4.0164 0.2509 sec/batch\n",
      "Epoch 8/10 Iteration: 120200 Avg. Training loss: 4.0974 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 120300 Avg. Training loss: 4.0362 0.0896 sec/batch\n",
      "Epoch 9/10 Iteration: 120400 Avg. Training loss: 4.0532 0.2485 sec/batch\n",
      "Epoch 9/10 Iteration: 120500 Avg. Training loss: 3.9574 0.2490 sec/batch\n",
      "Epoch 9/10 Iteration: 120600 Avg. Training loss: 4.0372 0.2514 sec/batch\n",
      "Epoch 9/10 Iteration: 120700 Avg. Training loss: 4.0265 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 120800 Avg. Training loss: 3.9430 0.2492 sec/batch\n",
      "Epoch 9/10 Iteration: 120900 Avg. Training loss: 3.9967 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 121000 Avg. Training loss: 3.9901 0.2481 sec/batch\n",
      "Nearest to that: note, argue, suggested, fact, noted, suggest, claimed, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, external, references, called, daiko, but,\n",
      "Nearest to he: himself, wrote, where, nihonshoki, whom, thoroughbred, cheltenham, also,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, term, commonly, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, war, my, iii,\n",
      "Nearest to when: asked, discussing, arrived, referring, happens, guessing, especially, occurs,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, women, children, noun, fertility,\n",
      "Nearest to dr: blok, physician, sierpinski, kovacs, zaman, positioned, acc, berimbau,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, elas, predominantly, democratisation, twelver,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, neu, bokuto, tornadoes,\n",
      "Nearest to test: tests, ban, cricket, hazardous, matches, globes, dumping, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, long, milked, coney, living, story,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, accident, routes, extractive, interstate, donovan,\n",
      "Nearest to ocean: atlantic, pacific, posed, indian, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 9/10 Iteration: 121100 Avg. Training loss: 3.9354 0.2527 sec/batch\n",
      "Epoch 9/10 Iteration: 121200 Avg. Training loss: 3.9446 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 121300 Avg. Training loss: 3.9452 0.2485 sec/batch\n",
      "Epoch 9/10 Iteration: 121400 Avg. Training loss: 4.0512 0.2492 sec/batch\n",
      "Epoch 9/10 Iteration: 121500 Avg. Training loss: 3.9895 0.2474 sec/batch\n",
      "Epoch 9/10 Iteration: 121600 Avg. Training loss: 3.9913 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 121700 Avg. Training loss: 3.9840 0.2477 sec/batch\n",
      "Epoch 9/10 Iteration: 121800 Avg. Training loss: 3.9937 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 121900 Avg. Training loss: 4.0625 0.2492 sec/batch\n",
      "Epoch 9/10 Iteration: 122000 Avg. Training loss: 3.9930 0.2488 sec/batch\n",
      "Nearest to that: note, argue, fact, suggested, suggest, noted, claimed, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, called, references, daiko, but,\n",
      "Nearest to he: himself, where, wrote, nihonshoki, cheltenham, whom, journeyed, his,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, commonly, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, iii, my, world, war,\n",
      "Nearest to when: asked, discussing, referring, especially, arrived, guessing, occurs, happens,\n",
      "Nearest to woman: nationality, man, sonnets, defensemen, women, fertility, children, noun,\n",
      "Nearest to dr: blok, physician, zaman, sierpinski, positioned, kovacs, acc, prof,\n",
      "Nearest to mainly: primarily, exclusively, emulsions, mostly, elas, predominantly, largely, democratisation,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, neu, bokuto, tornadoes,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, obsoletes, matches, dumping,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, long, milked, story, living,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, accident, extractive, interstate, routes, donovan,\n",
      "Nearest to ocean: atlantic, pacific, posed, indian, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 9/10 Iteration: 122100 Avg. Training loss: 4.0568 0.2517 sec/batch\n",
      "Epoch 9/10 Iteration: 122200 Avg. Training loss: 4.0064 0.2481 sec/batch\n",
      "Epoch 9/10 Iteration: 122300 Avg. Training loss: 4.0243 0.2500 sec/batch\n",
      "Epoch 9/10 Iteration: 122400 Avg. Training loss: 3.9852 0.2480 sec/batch\n",
      "Epoch 9/10 Iteration: 122500 Avg. Training loss: 3.8711 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 122600 Avg. Training loss: 4.0427 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 122700 Avg. Training loss: 4.0143 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 122800 Avg. Training loss: 4.0589 0.2502 sec/batch\n",
      "Epoch 9/10 Iteration: 122900 Avg. Training loss: 4.0287 0.2478 sec/batch\n",
      "Epoch 9/10 Iteration: 123000 Avg. Training loss: 4.0313 0.2494 sec/batch\n",
      "Nearest to that: note, argue, fact, suggested, noted, suggest, argued, claimed,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, daiko, external, references, but,\n",
      "Nearest to he: himself, where, wrote, thoroughbred, cheltenham, nihonshoki, also, dike,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, assorted, except,\n",
      "Nearest to used: describe, extensively, widely, refer, term, commonly, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, iii,\n",
      "Nearest to when: asked, discussing, referring, guessing, especially, arrived, occurs, happens,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, women, fertility, children, gentle,\n",
      "Nearest to dr: blok, physician, positioned, zaman, sierpinski, kovacs, berimbau, acc,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, largely, elas, democratisation, predominantly,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, neu, tornadoes, workforce,\n",
      "Nearest to test: tests, ban, cricket, hazardous, matches, globes, dumping, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, long, story, milked, living,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, extractive, interstate, accident, routes, donovan,\n",
      "Nearest to ocean: atlantic, pacific, posed, indian, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 9/10 Iteration: 123100 Avg. Training loss: 4.0204 0.2497 sec/batch\n",
      "Epoch 9/10 Iteration: 123200 Avg. Training loss: 4.0200 0.2484 sec/batch\n",
      "Epoch 9/10 Iteration: 123300 Avg. Training loss: 4.0056 0.2493 sec/batch\n",
      "Epoch 9/10 Iteration: 123400 Avg. Training loss: 4.0391 0.2474 sec/batch\n",
      "Epoch 9/10 Iteration: 123500 Avg. Training loss: 4.0494 0.2495 sec/batch\n",
      "Epoch 9/10 Iteration: 123600 Avg. Training loss: 4.0398 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 123700 Avg. Training loss: 3.9708 0.2489 sec/batch\n",
      "Epoch 9/10 Iteration: 123800 Avg. Training loss: 3.9909 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 123900 Avg. Training loss: 3.9207 0.2493 sec/batch\n",
      "Epoch 9/10 Iteration: 124000 Avg. Training loss: 3.8475 0.2485 sec/batch\n",
      "Nearest to that: note, argue, suggested, fact, noted, argued, argues, suggest,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, external, references, called, daiko, but,\n",
      "Nearest to he: himself, wrote, thoroughbred, where, popularize, nihonshoki, cheltenham, tribesmen,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, except,\n",
      "Nearest to used: describe, extensively, refer, widely, term, commonly, denote, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, war, my, iii,\n",
      "Nearest to when: asked, discussing, referring, especially, occurs, guessing, arrived, happens,\n",
      "Nearest to woman: nationality, defensemen, man, sonnets, women, fertility, children, noun,\n",
      "Nearest to dr: blok, physician, zaman, positioned, sierpinski, kovacs, acc, berimbau,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, predominantly, largely, elas, twelver,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, neu, tornadoes, workforce,\n",
      "Nearest to test: tests, ban, cricket, hazardous, matches, globes, dumping, obsoletes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, story, worked, milked, coney,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, interstate, extractive, routes, roads, accident,\n",
      "Nearest to ocean: atlantic, pacific, indian, intersputnik, posed, intelsat, castilla, ost,\n",
      "Epoch 9/10 Iteration: 124100 Avg. Training loss: 4.0111 0.2515 sec/batch\n",
      "Epoch 9/10 Iteration: 124200 Avg. Training loss: 4.0248 0.2496 sec/batch\n",
      "Epoch 9/10 Iteration: 124300 Avg. Training loss: 3.9573 0.2484 sec/batch\n",
      "Epoch 9/10 Iteration: 124400 Avg. Training loss: 3.9986 0.2485 sec/batch\n",
      "Epoch 9/10 Iteration: 124500 Avg. Training loss: 3.9923 0.2483 sec/batch\n",
      "Epoch 9/10 Iteration: 124600 Avg. Training loss: 3.9920 0.2493 sec/batch\n",
      "Epoch 9/10 Iteration: 124700 Avg. Training loss: 3.9891 0.2489 sec/batch\n",
      "Epoch 9/10 Iteration: 124800 Avg. Training loss: 3.9062 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 124900 Avg. Training loss: 4.0408 0.2485 sec/batch\n",
      "Epoch 9/10 Iteration: 125000 Avg. Training loss: 4.0091 0.2488 sec/batch\n",
      "Nearest to that: note, suggested, fact, argue, noted, suggest, argued, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, called, references, external, daiko, awakenings,\n",
      "Nearest to he: himself, wrote, where, thoroughbred, cheltenham, nihonshoki, popularize, transmutation,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, commonly, frequently,\n",
      "Nearest to i: ii, e, am, think, world, war, my, iii,\n",
      "Nearest to when: asked, referring, discussing, especially, occurs, happens, guessing, arrived,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, women, fertility, children, noun,\n",
      "Nearest to dr: blok, physician, kovacs, acc, positioned, zaman, rcn, berimbau,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, largely, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, workforce,\n",
      "Nearest to test: tests, ban, cricket, hazardous, globes, matches, dumping, wastes,\n",
      "Nearest to lived: short, eunuch, rabindranath, worked, long, living, cannibal, milked,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, interstate, accident, extractive, routes, donovan,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, intersputnik, intelsat, castilla, ost,\n",
      "Epoch 9/10 Iteration: 125100 Avg. Training loss: 4.0263 0.2511 sec/batch\n",
      "Epoch 9/10 Iteration: 125200 Avg. Training loss: 4.0221 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 125300 Avg. Training loss: 4.0428 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 125400 Avg. Training loss: 4.0309 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 125500 Avg. Training loss: 3.9975 0.2490 sec/batch\n",
      "Epoch 9/10 Iteration: 125600 Avg. Training loss: 4.0470 0.2490 sec/batch\n",
      "Epoch 9/10 Iteration: 125700 Avg. Training loss: 4.0614 0.2480 sec/batch\n",
      "Epoch 9/10 Iteration: 125800 Avg. Training loss: 3.9550 0.2483 sec/batch\n",
      "Epoch 9/10 Iteration: 125900 Avg. Training loss: 3.9865 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 126000 Avg. Training loss: 4.0031 0.2484 sec/batch\n",
      "Nearest to that: note, suggested, fact, argue, argued, noted, suggest, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, confused,\n",
      "Nearest to also: see, list, known, external, called, references, daiko, awakenings,\n",
      "Nearest to he: himself, wrote, where, popularize, thoroughbred, transmutation, cheltenham, his,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, except,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, commonly, interchangeably,\n",
      "Nearest to i: e, ii, am, world, think, my, war, iii,\n",
      "Nearest to when: asked, referring, discussing, occurs, especially, guessing, arrived, happens,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, women, fertility, children, noun,\n",
      "Nearest to dr: blok, physician, kovacs, positioned, berimbau, zaman, rcn, acc,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, largely, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, workforce,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, dumping, wastes, matches,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, worked, milked, coney, cannibal,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, interstate, middlesex, routes, accident, waterbury,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 9/10 Iteration: 126100 Avg. Training loss: 3.9441 0.2518 sec/batch\n",
      "Epoch 9/10 Iteration: 126200 Avg. Training loss: 4.0459 0.2490 sec/batch\n",
      "Epoch 9/10 Iteration: 126300 Avg. Training loss: 3.9979 0.2472 sec/batch\n",
      "Epoch 9/10 Iteration: 126400 Avg. Training loss: 4.0371 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 126500 Avg. Training loss: 4.0425 0.2497 sec/batch\n",
      "Epoch 9/10 Iteration: 126600 Avg. Training loss: 4.0585 0.2484 sec/batch\n",
      "Epoch 9/10 Iteration: 126700 Avg. Training loss: 3.9951 0.2479 sec/batch\n",
      "Epoch 9/10 Iteration: 126800 Avg. Training loss: 3.9016 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 126900 Avg. Training loss: 4.0453 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 127000 Avg. Training loss: 3.9681 0.2490 sec/batch\n",
      "Nearest to that: note, argue, suggested, fact, argued, suggest, noted, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, references, called, daiko, awakenings,\n",
      "Nearest to he: himself, wrote, where, popularize, transmutation, his, cheltenham, thoroughbred,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, interchangeably, commonly,\n",
      "Nearest to i: e, ii, am, think, world, war, iii, my,\n",
      "Nearest to when: asked, referring, discussing, occurs, especially, guessing, happens, kidman,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, women, fertility, children, noun,\n",
      "Nearest to dr: blok, physician, kovacs, berimbau, positioned, zaman, rcn, acc,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, largely, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, celsius,\n",
      "Nearest to test: ban, tests, cricket, hazardous, dumping, globes, obsoletes, wastes,\n",
      "Nearest to lived: short, eunuch, long, worked, rabindranath, milked, coney, story,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, routes, interstate, vehicles, streets, extractive,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, intersputnik, castilla, ost, intelsat,\n",
      "Epoch 9/10 Iteration: 127100 Avg. Training loss: 3.9799 0.2519 sec/batch\n",
      "Epoch 9/10 Iteration: 127200 Avg. Training loss: 4.0508 0.2490 sec/batch\n",
      "Epoch 9/10 Iteration: 127300 Avg. Training loss: 4.0297 0.2483 sec/batch\n",
      "Epoch 9/10 Iteration: 127400 Avg. Training loss: 3.9802 0.2495 sec/batch\n",
      "Epoch 9/10 Iteration: 127500 Avg. Training loss: 4.0354 0.2496 sec/batch\n",
      "Epoch 9/10 Iteration: 127600 Avg. Training loss: 4.0054 0.2480 sec/batch\n",
      "Epoch 9/10 Iteration: 127700 Avg. Training loss: 3.9682 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 127800 Avg. Training loss: 4.0153 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 127900 Avg. Training loss: 4.0067 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 128000 Avg. Training loss: 3.9354 0.2496 sec/batch\n",
      "Nearest to that: note, argue, suggested, fact, argued, noted, suggest, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, confused,\n",
      "Nearest to also: see, list, known, called, external, references, daiko, sprague,\n",
      "Nearest to he: himself, where, his, popularize, transmutation, nihonshoki, whom, wrote,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, commonly, interchangeably,\n",
      "Nearest to i: ii, e, am, war, world, think, my, iii,\n",
      "Nearest to when: asked, referring, discussing, occurs, especially, happens, arrived, guessing,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, women, fertility, children, noun,\n",
      "Nearest to dr: blok, physician, positioned, berimbau, kovacs, rcn, zaman, acc,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, largely, democratisation, predominantly, analogical,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, celsius,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, matches, dumping, obsoletes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, living, worked, milked, coney,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, interstate, routes, extractive, accident, donovan,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 9/10 Iteration: 128100 Avg. Training loss: 3.9742 0.2514 sec/batch\n",
      "Epoch 9/10 Iteration: 128200 Avg. Training loss: 4.0382 0.2481 sec/batch\n",
      "Epoch 9/10 Iteration: 128300 Avg. Training loss: 3.8406 0.2492 sec/batch\n",
      "Epoch 9/10 Iteration: 128400 Avg. Training loss: 4.0405 0.2493 sec/batch\n",
      "Epoch 9/10 Iteration: 128500 Avg. Training loss: 4.0115 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 128600 Avg. Training loss: 4.0040 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 128700 Avg. Training loss: 4.0318 0.2493 sec/batch\n",
      "Epoch 9/10 Iteration: 128800 Avg. Training loss: 3.8550 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 128900 Avg. Training loss: 4.0178 0.2489 sec/batch\n",
      "Epoch 9/10 Iteration: 129000 Avg. Training loss: 3.9867 0.2492 sec/batch\n",
      "Nearest to that: note, fact, suggested, argue, suggest, argues, argued, noted,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, confused,\n",
      "Nearest to also: see, list, known, external, references, called, sprague, daiko,\n",
      "Nearest to he: himself, his, where, wrote, popularize, transmutation, thoroughbred, whom,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, commonly, interchangeably,\n",
      "Nearest to i: ii, e, am, think, world, my, war, iii,\n",
      "Nearest to when: asked, discussing, referring, arrived, especially, guessing, happens, occurs,\n",
      "Nearest to woman: nationality, man, defensemen, fertility, sonnets, women, children, noun,\n",
      "Nearest to dr: blok, physician, berimbau, kovacs, positioned, zaman, rcn, acc,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, largely, analogical, democratisation, predominantly,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, celsius,\n",
      "Nearest to test: ban, tests, cricket, hazardous, matches, globes, obsoletes, dumping,\n",
      "Nearest to lived: short, eunuch, long, living, worked, milked, rabindranath, cannibal,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, interstate, extractive, routes, accident, vehicles,\n",
      "Nearest to ocean: atlantic, pacific, posed, intersputnik, indian, ost, castilla, intelsat,\n",
      "Epoch 9/10 Iteration: 129100 Avg. Training loss: 4.0447 0.2516 sec/batch\n",
      "Epoch 9/10 Iteration: 129200 Avg. Training loss: 4.0140 0.2501 sec/batch\n",
      "Epoch 9/10 Iteration: 129300 Avg. Training loss: 4.0381 0.2490 sec/batch\n",
      "Epoch 9/10 Iteration: 129400 Avg. Training loss: 4.0560 0.2492 sec/batch\n",
      "Epoch 9/10 Iteration: 129500 Avg. Training loss: 4.0390 0.2500 sec/batch\n",
      "Epoch 9/10 Iteration: 129600 Avg. Training loss: 3.9765 0.2483 sec/batch\n",
      "Epoch 9/10 Iteration: 129700 Avg. Training loss: 3.9772 0.2490 sec/batch\n",
      "Epoch 9/10 Iteration: 129800 Avg. Training loss: 4.0429 0.2495 sec/batch\n",
      "Epoch 9/10 Iteration: 129900 Avg. Training loss: 4.0376 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 130000 Avg. Training loss: 4.0448 0.2494 sec/batch\n",
      "Nearest to that: note, suggested, fact, suggest, argued, argue, believe, noted,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, yet, but, confused,\n",
      "Nearest to also: see, list, known, external, called, references, sprague, but,\n",
      "Nearest to he: himself, where, transmutation, wrote, nihonshoki, popularize, montpellier, his,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, except, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, interchangeably, commonly,\n",
      "Nearest to i: ii, e, am, think, world, war, my, iii,\n",
      "Nearest to when: asked, referring, discussing, especially, occurs, arrived, kidman, guessing,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, fertility, women, noun, children,\n",
      "Nearest to dr: blok, physician, kovacs, positioned, berimbau, zaman, acc, rcn,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, analogical, democratisation, largely, paternoster,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, celsius,\n",
      "Nearest to test: ban, tests, cricket, hazardous, globes, dumping, obsoletes, matches,\n",
      "Nearest to lived: short, eunuch, long, living, worked, rabindranath, cannibal, milked,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, interstate, routes, extractive, vehicles, accident,\n",
      "Nearest to ocean: atlantic, pacific, posed, indian, intersputnik, intelsat, castilla, ost,\n",
      "Epoch 9/10 Iteration: 130100 Avg. Training loss: 3.9942 0.2503 sec/batch\n",
      "Epoch 9/10 Iteration: 130200 Avg. Training loss: 4.0387 0.2483 sec/batch\n",
      "Epoch 9/10 Iteration: 130300 Avg. Training loss: 4.0403 0.2485 sec/batch\n",
      "Epoch 9/10 Iteration: 130400 Avg. Training loss: 4.0169 0.2494 sec/batch\n",
      "Epoch 9/10 Iteration: 130500 Avg. Training loss: 4.0415 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 130600 Avg. Training loss: 4.0280 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 130700 Avg. Training loss: 3.9938 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 130800 Avg. Training loss: 3.9178 0.2489 sec/batch\n",
      "Epoch 9/10 Iteration: 130900 Avg. Training loss: 3.9893 0.2482 sec/batch\n",
      "Epoch 9/10 Iteration: 131000 Avg. Training loss: 4.0103 0.2492 sec/batch\n",
      "Nearest to that: note, suggested, fact, argue, suggest, argued, believe, noted,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, external, references, called, but, below,\n",
      "Nearest to he: himself, transmutation, thoroughbred, wrote, where, popularize, nihonshoki, acknowledges,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, things, except, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, term, denote, interchangeably, frequently,\n",
      "Nearest to i: ii, e, am, world, think, my, war, iii,\n",
      "Nearest to when: asked, referring, discussing, occurs, especially, happens, arrived, kidman,\n",
      "Nearest to woman: nationality, man, defensemen, sonnets, fertility, women, noun, children,\n",
      "Nearest to dr: blok, physician, positioned, zaman, kovacs, berimbau, acc, prof,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, democratisation, predominantly, analogical, largely,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, celsius,\n",
      "Nearest to test: ban, tests, cricket, hazardous, obsoletes, globes, dumping, matches,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, living, worked, story, coney,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, extractive, interstate, accident, routes, vehicles,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, intersputnik, intelsat, ost, castilla,\n",
      "Epoch 9/10 Iteration: 131100 Avg. Training loss: 3.9580 0.2514 sec/batch\n",
      "Epoch 9/10 Iteration: 131200 Avg. Training loss: 4.0039 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 131300 Avg. Training loss: 3.9600 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 131400 Avg. Training loss: 4.0420 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 131500 Avg. Training loss: 3.9878 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 131600 Avg. Training loss: 3.6466 0.2484 sec/batch\n",
      "Epoch 9/10 Iteration: 131700 Avg. Training loss: 3.7507 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 131800 Avg. Training loss: 3.8378 0.2492 sec/batch\n",
      "Epoch 9/10 Iteration: 131900 Avg. Training loss: 3.8879 0.2497 sec/batch\n",
      "Epoch 9/10 Iteration: 132000 Avg. Training loss: 3.9400 0.2487 sec/batch\n",
      "Nearest to that: note, suggested, argue, suggest, fact, believe, argued, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, yet, confused,\n",
      "Nearest to also: see, list, known, called, references, he, external, but,\n",
      "Nearest to he: himself, wrote, where, also, thoroughbred, cheltenham, transmutation, whom,\n",
      "Nearest to all: nearly, listing, virtually, almost, practically, things, assorted, except,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, commonly, interchangeably,\n",
      "Nearest to i: ii, e, am, world, think, my, war, iii,\n",
      "Nearest to when: asked, discussing, referring, arrived, occurs, kidman, happens, especially,\n",
      "Nearest to woman: nationality, man, defensemen, women, sonnets, noun, fertility, children,\n",
      "Nearest to dr: blok, physician, kovacs, zaman, acc, positioned, berimbau, prof,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, analogical, democratisation, predominantly, fond,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, workforce,\n",
      "Nearest to test: ban, tests, cricket, hazardous, obsoletes, globes, dumping, matches,\n",
      "Nearest to lived: short, eunuch, long, living, story, worked, rabindranath, coney,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, accident, donovan, interstate, extractive, routes,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, intersputnik, ost, castilla, intelsat,\n",
      "Epoch 9/10 Iteration: 132100 Avg. Training loss: 4.0211 0.2523 sec/batch\n",
      "Epoch 9/10 Iteration: 132200 Avg. Training loss: 4.0067 0.2484 sec/batch\n",
      "Epoch 9/10 Iteration: 132300 Avg. Training loss: 4.0139 0.2489 sec/batch\n",
      "Epoch 9/10 Iteration: 132400 Avg. Training loss: 4.0397 0.2481 sec/batch\n",
      "Epoch 9/10 Iteration: 132500 Avg. Training loss: 4.0677 0.2496 sec/batch\n",
      "Epoch 9/10 Iteration: 132600 Avg. Training loss: 4.0624 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 132700 Avg. Training loss: 4.0667 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 132800 Avg. Training loss: 4.0432 0.2480 sec/batch\n",
      "Epoch 9/10 Iteration: 132900 Avg. Training loss: 4.0068 0.2479 sec/batch\n",
      "Epoch 9/10 Iteration: 133000 Avg. Training loss: 3.8270 0.2480 sec/batch\n",
      "Nearest to that: note, fact, argue, suggested, suggest, argued, believe, claimed,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, references, external, called, below, there,\n",
      "Nearest to he: himself, where, wrote, cheltenham, thoroughbred, popularize, acknowledges, discontented,\n",
      "Nearest to all: nearly, virtually, listing, almost, practically, things, except, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, term, commonly, denote, frequently,\n",
      "Nearest to i: ii, e, am, think, world, my, war, said,\n",
      "Nearest to when: asked, discussing, referring, arrived, happens, occurs, especially, kidman,\n",
      "Nearest to woman: nationality, man, defensemen, women, fertility, noun, children, sonnets,\n",
      "Nearest to dr: blok, physician, positioned, acc, zaman, kovacs, berimbau, rcn,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, analogical,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, yuga,\n",
      "Nearest to test: ban, tests, hazardous, cricket, dumping, obsoletes, globes, matches,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, settled, living, story, coney,\n",
      "Nearest to road: hildebrand, hypocrisy, traffic, extractive, accident, vehicles, donovan, interstate,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, intersputnik, castilla, ost, intelsat,\n",
      "Epoch 9/10 Iteration: 133100 Avg. Training loss: 4.0340 0.2512 sec/batch\n",
      "Epoch 9/10 Iteration: 133200 Avg. Training loss: 4.0397 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 133300 Avg. Training loss: 3.9473 0.2499 sec/batch\n",
      "Epoch 9/10 Iteration: 133400 Avg. Training loss: 4.0204 0.2485 sec/batch\n",
      "Epoch 9/10 Iteration: 133500 Avg. Training loss: 3.9584 0.2488 sec/batch\n",
      "Epoch 9/10 Iteration: 133600 Avg. Training loss: 4.0727 0.2478 sec/batch\n",
      "Epoch 9/10 Iteration: 133700 Avg. Training loss: 4.0099 0.2501 sec/batch\n",
      "Epoch 9/10 Iteration: 133800 Avg. Training loss: 3.9964 0.2482 sec/batch\n",
      "Epoch 9/10 Iteration: 133900 Avg. Training loss: 3.9487 0.2494 sec/batch\n",
      "Epoch 9/10 Iteration: 134000 Avg. Training loss: 3.7944 0.2494 sec/batch\n",
      "Nearest to that: note, fact, argue, suggest, suggested, announced, believe, claimed,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, external, references, called, below, daiko,\n",
      "Nearest to he: himself, where, wrote, cheltenham, attended, popularize, whom, colic,\n",
      "Nearest to all: nearly, virtually, almost, listing, practically, everywhere, things, except,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, denote, frequently, term,\n",
      "Nearest to i: ii, e, am, world, think, war, my, iii,\n",
      "Nearest to when: asked, discussing, referring, arrived, happens, especially, occurs, guessing,\n",
      "Nearest to woman: nationality, man, children, defensemen, fertility, women, noun, sonnets,\n",
      "Nearest to dr: blok, physician, positioned, acc, kovacs, zaman, berimbau, rcn,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, predominantly, democratisation, largely, elas,\n",
      "Nearest to scale: large, scales, richter, rt, casualties, tornadoes, neu, yuga,\n",
      "Nearest to test: ban, tests, cricket, hazardous, dumping, obsoletes, matches, globes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, story, coney, living, settled,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, extractive, accident, vehicles, routes, donovan,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 9/10 Iteration: 134100 Avg. Training loss: 3.9071 0.2527 sec/batch\n",
      "Epoch 9/10 Iteration: 134200 Avg. Training loss: 4.0024 0.2486 sec/batch\n",
      "Epoch 9/10 Iteration: 134300 Avg. Training loss: 4.0183 0.2494 sec/batch\n",
      "Epoch 9/10 Iteration: 134400 Avg. Training loss: 3.8731 0.2477 sec/batch\n",
      "Epoch 9/10 Iteration: 134500 Avg. Training loss: 3.9974 0.2489 sec/batch\n",
      "Epoch 9/10 Iteration: 134600 Avg. Training loss: 4.0305 0.2487 sec/batch\n",
      "Epoch 9/10 Iteration: 134700 Avg. Training loss: 3.9992 0.2491 sec/batch\n",
      "Epoch 9/10 Iteration: 134800 Avg. Training loss: 3.9548 0.2495 sec/batch\n",
      "Epoch 9/10 Iteration: 134900 Avg. Training loss: 3.9727 0.2489 sec/batch\n",
      "Epoch 9/10 Iteration: 135000 Avg. Training loss: 3.9750 0.2481 sec/batch\n",
      "Nearest to that: note, fact, suggested, argue, suggest, announced, claimed, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, called, external, references, daiko, below,\n",
      "Nearest to he: himself, where, wrote, cheltenham, whom, attended, mulhouse, guangzhou,\n",
      "Nearest to all: nearly, listing, virtually, almost, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, term, denote, commonly, interchangeably,\n",
      "Nearest to i: ii, e, am, world, war, think, iii, my,\n",
      "Nearest to when: asked, discussing, referring, arrived, happens, especially, occurs, guessing,\n",
      "Nearest to woman: nationality, man, women, defensemen, children, noun, fertility, sonnets,\n",
      "Nearest to dr: physician, blok, positioned, acc, rcn, berimbau, kovacs, zaman,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, democratisation, predominantly, elas, largely,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, yuga,\n",
      "Nearest to test: ban, tests, cricket, hazardous, matches, obsoletes, dumping, globes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, settled, living, coney, monks,\n",
      "Nearest to road: hildebrand, traffic, hypocrisy, accident, extractive, routes, racing, vehicles,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, intersputnik, castilla, ost, intelsat,\n",
      "Epoch 9/10 Iteration: 135100 Avg. Training loss: 3.9919 0.2541 sec/batch\n",
      "Epoch 9/10 Iteration: 135200 Avg. Training loss: 4.0686 0.2481 sec/batch\n",
      "Epoch 10/10 Iteration: 135300 Avg. Training loss: 4.0431 0.0079 sec/batch\n",
      "Epoch 10/10 Iteration: 135400 Avg. Training loss: 4.0349 0.2481 sec/batch\n",
      "Epoch 10/10 Iteration: 135500 Avg. Training loss: 3.9692 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 135600 Avg. Training loss: 4.0186 0.2493 sec/batch\n",
      "Epoch 10/10 Iteration: 135700 Avg. Training loss: 4.0399 0.2487 sec/batch\n",
      "Epoch 10/10 Iteration: 135800 Avg. Training loss: 3.9508 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 135900 Avg. Training loss: 3.9869 0.2496 sec/batch\n",
      "Epoch 10/10 Iteration: 136000 Avg. Training loss: 4.0106 0.2487 sec/batch\n",
      "Nearest to that: fact, note, argue, claimed, suggested, believe, suggest, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, external, references, called, daiko, but,\n",
      "Nearest to he: himself, where, wrote, cheltenham, whom, journeyed, nihonshoki, also,\n",
      "Nearest to all: nearly, listing, almost, virtually, practically, things, everywhere, assorted,\n",
      "Nearest to used: describe, extensively, refer, widely, term, denote, commonly, frequently,\n",
      "Nearest to i: ii, e, am, think, world, my, iii, war,\n",
      "Nearest to when: asked, discussing, referring, arrived, happens, especially, guessing, occurs,\n",
      "Nearest to woman: nationality, man, women, children, defensemen, sonnets, noun, fertility,\n",
      "Nearest to dr: physician, blok, acc, positioned, berimbau, zaman, prof, rcn,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, democratisation, predominantly, elas, largely,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, fifths,\n",
      "Nearest to test: tests, ban, cricket, matches, hazardous, obsoletes, globes, dumping,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, living, monks, coney, story,\n",
      "Nearest to road: traffic, hildebrand, accident, hypocrisy, racing, extractive, routes, vehicles,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, castilla, intersputnik, ost, intelsat,\n",
      "Epoch 10/10 Iteration: 136100 Avg. Training loss: 3.9708 0.2518 sec/batch\n",
      "Epoch 10/10 Iteration: 136200 Avg. Training loss: 3.9083 0.2491 sec/batch\n",
      "Epoch 10/10 Iteration: 136300 Avg. Training loss: 3.9201 0.2496 sec/batch\n",
      "Epoch 10/10 Iteration: 136400 Avg. Training loss: 3.9925 0.2497 sec/batch\n",
      "Epoch 10/10 Iteration: 136500 Avg. Training loss: 4.0042 0.2494 sec/batch\n",
      "Epoch 10/10 Iteration: 136600 Avg. Training loss: 4.0003 0.2490 sec/batch\n",
      "Epoch 10/10 Iteration: 136700 Avg. Training loss: 3.9732 0.2495 sec/batch\n",
      "Epoch 10/10 Iteration: 136800 Avg. Training loss: 3.9746 0.2488 sec/batch\n",
      "Epoch 10/10 Iteration: 136900 Avg. Training loss: 4.0267 0.2491 sec/batch\n",
      "Epoch 10/10 Iteration: 137000 Avg. Training loss: 4.0099 0.2482 sec/batch\n",
      "Nearest to that: fact, note, suggested, argue, claimed, suggest, announced, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, called, external, references, daiko, but,\n",
      "Nearest to he: himself, where, journeyed, wrote, whom, his, nihonshoki, cheltenham,\n",
      "Nearest to all: listing, nearly, almost, virtually, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, frequently, interchangeably,\n",
      "Nearest to i: ii, e, am, world, my, think, iii, war,\n",
      "Nearest to when: asked, discussing, referring, arrived, especially, happens, occurs, guessing,\n",
      "Nearest to woman: nationality, man, children, defensemen, sonnets, women, noun, fertility,\n",
      "Nearest to dr: physician, blok, positioned, acc, prof, zaman, rcn, berimbau,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, democratisation, predominantly, largely, cerebrospinal,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, neu, tornadoes, fifths,\n",
      "Nearest to test: ban, tests, cricket, matches, hazardous, obsoletes, globes, dumping,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, worked, monks, living, story,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, routes, accident, racing, extractive, interstate,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, castilla, intersputnik, ost, intelsat,\n",
      "Epoch 10/10 Iteration: 137100 Avg. Training loss: 4.0269 0.2520 sec/batch\n",
      "Epoch 10/10 Iteration: 137200 Avg. Training loss: 4.0204 0.2484 sec/batch\n",
      "Epoch 10/10 Iteration: 137300 Avg. Training loss: 4.0252 0.2495 sec/batch\n",
      "Epoch 10/10 Iteration: 137400 Avg. Training loss: 4.0248 0.2492 sec/batch\n",
      "Epoch 10/10 Iteration: 137500 Avg. Training loss: 3.7939 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 137600 Avg. Training loss: 4.0794 0.2483 sec/batch\n",
      "Epoch 10/10 Iteration: 137700 Avg. Training loss: 4.0096 0.2479 sec/batch\n",
      "Epoch 10/10 Iteration: 137800 Avg. Training loss: 4.0423 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 137900 Avg. Training loss: 4.0405 0.2499 sec/batch\n",
      "Epoch 10/10 Iteration: 138000 Avg. Training loss: 4.0318 0.2492 sec/batch\n",
      "Nearest to that: note, fact, argue, suggested, noted, argued, suggest, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, yet, but,\n",
      "Nearest to also: see, list, known, called, daiko, external, references, denslow,\n",
      "Nearest to he: himself, where, wrote, popularize, thoroughbred, cheltenham, journeyed, youngster,\n",
      "Nearest to all: virtually, almost, listing, nearly, practically, things, assorted, except,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, denote, term, frequently,\n",
      "Nearest to i: ii, e, am, think, world, my, war, iii,\n",
      "Nearest to when: asked, discussing, referring, guessing, happens, especially, arrived, occurs,\n",
      "Nearest to woman: nationality, man, children, defensemen, sonnets, women, fertility, noun,\n",
      "Nearest to dr: physician, blok, positioned, acc, zaman, rcn, prof, berimbau,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, fifths,\n",
      "Nearest to test: ban, tests, cricket, matches, hazardous, dumping, obsoletes, globes,\n",
      "Nearest to lived: short, eunuch, long, monks, rabindranath, living, worked, cannibal,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, racing, accident, routes, extractive, interstate,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 10/10 Iteration: 138100 Avg. Training loss: 4.0052 0.2523 sec/batch\n",
      "Epoch 10/10 Iteration: 138200 Avg. Training loss: 4.0136 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 138300 Avg. Training loss: 3.9788 0.2484 sec/batch\n",
      "Epoch 10/10 Iteration: 138400 Avg. Training loss: 4.0533 0.2490 sec/batch\n",
      "Epoch 10/10 Iteration: 138500 Avg. Training loss: 4.0488 0.2487 sec/batch\n",
      "Epoch 10/10 Iteration: 138600 Avg. Training loss: 4.0301 0.2490 sec/batch\n",
      "Epoch 10/10 Iteration: 138700 Avg. Training loss: 3.9719 0.2490 sec/batch\n",
      "Epoch 10/10 Iteration: 138800 Avg. Training loss: 4.0048 0.2495 sec/batch\n",
      "Epoch 10/10 Iteration: 138900 Avg. Training loss: 3.9607 0.2497 sec/batch\n",
      "Epoch 10/10 Iteration: 139000 Avg. Training loss: 3.8051 0.2495 sec/batch\n",
      "Nearest to that: note, fact, suggested, argue, noted, argues, argued, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, yet, but,\n",
      "Nearest to also: see, list, known, external, called, daiko, references, links,\n",
      "Nearest to he: himself, where, popularize, wrote, thoroughbred, orcinus, cheltenham, youngster,\n",
      "Nearest to all: virtually, almost, nearly, listing, practically, things, assorted, except,\n",
      "Nearest to used: describe, extensively, refer, widely, denote, term, frequently, commonly,\n",
      "Nearest to i: ii, e, am, think, world, my, war, iii,\n",
      "Nearest to when: asked, discussing, referring, guessing, happens, occurs, especially, arrived,\n",
      "Nearest to woman: nationality, man, children, defensemen, fertility, women, sonnets, noun,\n",
      "Nearest to dr: physician, blok, positioned, acc, zaman, rcn, berimbau, insurgent,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, notes,\n",
      "Nearest to test: ban, tests, cricket, matches, hazardous, dumping, globes, obsoletes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, monks, living, worked, cannibal,\n",
      "Nearest to road: traffic, routes, hypocrisy, hildebrand, accident, interstate, extractive, racing,\n",
      "Nearest to ocean: atlantic, indian, pacific, intersputnik, posed, castilla, intelsat, ost,\n",
      "Epoch 10/10 Iteration: 139100 Avg. Training loss: 3.9619 0.2542 sec/batch\n",
      "Epoch 10/10 Iteration: 139200 Avg. Training loss: 4.0342 0.2487 sec/batch\n",
      "Epoch 10/10 Iteration: 139300 Avg. Training loss: 3.9606 0.2476 sec/batch\n",
      "Epoch 10/10 Iteration: 139400 Avg. Training loss: 4.0302 0.2484 sec/batch\n",
      "Epoch 10/10 Iteration: 139500 Avg. Training loss: 3.9772 0.2489 sec/batch\n",
      "Epoch 10/10 Iteration: 139600 Avg. Training loss: 3.9845 0.2479 sec/batch\n",
      "Epoch 10/10 Iteration: 139700 Avg. Training loss: 3.9978 0.2481 sec/batch\n",
      "Epoch 10/10 Iteration: 139800 Avg. Training loss: 3.9117 0.2479 sec/batch\n",
      "Epoch 10/10 Iteration: 139900 Avg. Training loss: 4.0148 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 140000 Avg. Training loss: 4.0213 0.2486 sec/batch\n",
      "Nearest to that: note, fact, suggested, argue, noted, argues, suggest, argued,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, called, external, references, daiko, awakenings,\n",
      "Nearest to he: himself, popularize, where, his, orcinus, wrote, youngster, also,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, denote, frequently, term, commonly,\n",
      "Nearest to i: ii, e, am, think, world, my, war, iii,\n",
      "Nearest to when: asked, referring, discussing, happens, guessing, especially, occurs, arrived,\n",
      "Nearest to woman: nationality, man, children, women, defensemen, sonnets, fertility, noun,\n",
      "Nearest to dr: physician, blok, acc, rcn, positioned, kovacs, prof, zaman,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, analogical,\n",
      "Nearest to scale: large, scales, richter, casualties, tornadoes, rt, neu, yuga,\n",
      "Nearest to test: ban, tests, cricket, hazardous, dumping, matches, globes, obsoletes,\n",
      "Nearest to lived: short, eunuch, rabindranath, long, monks, cannibal, living, worked,\n",
      "Nearest to road: traffic, routes, hildebrand, hypocrisy, accident, interstate, racing, extractive,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, intersputnik, castilla, intelsat, ost,\n",
      "Epoch 10/10 Iteration: 140100 Avg. Training loss: 4.0299 0.2527 sec/batch\n",
      "Epoch 10/10 Iteration: 140200 Avg. Training loss: 4.0217 0.2495 sec/batch\n",
      "Epoch 10/10 Iteration: 140300 Avg. Training loss: 4.0146 0.2480 sec/batch\n",
      "Epoch 10/10 Iteration: 140400 Avg. Training loss: 4.0067 0.2493 sec/batch\n",
      "Epoch 10/10 Iteration: 140500 Avg. Training loss: 4.0101 0.2477 sec/batch\n",
      "Epoch 10/10 Iteration: 140600 Avg. Training loss: 4.0483 0.2493 sec/batch\n",
      "Epoch 10/10 Iteration: 140700 Avg. Training loss: 4.0507 0.2488 sec/batch\n",
      "Epoch 10/10 Iteration: 140800 Avg. Training loss: 3.9378 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 140900 Avg. Training loss: 4.0048 0.2490 sec/batch\n",
      "Epoch 10/10 Iteration: 141000 Avg. Training loss: 3.9935 0.2475 sec/batch\n",
      "Nearest to that: note, suggested, fact, argued, argue, argues, noted, indicate,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, called, external, daiko, references, below,\n",
      "Nearest to he: himself, popularize, where, his, acknowledges, also, wrote, thoroughbred,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, refer, widely, denote, frequently, term, commonly,\n",
      "Nearest to i: e, ii, am, world, my, think, war, iii,\n",
      "Nearest to when: asked, discussing, referring, occurs, especially, happens, guessing, arrived,\n",
      "Nearest to woman: nationality, man, children, defensemen, women, fertility, sonnets, noun,\n",
      "Nearest to dr: physician, blok, kovacs, positioned, acc, rcn, bibb, prof,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, democratisation, predominantly, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, yuga,\n",
      "Nearest to test: ban, tests, cricket, hazardous, matches, dumping, globes, obsoletes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, worked, cannibal, living, monks,\n",
      "Nearest to road: traffic, routes, hildebrand, hypocrisy, accident, racing, waterbury, extractive,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, castilla, ost, intersputnik, intelsat,\n",
      "Epoch 10/10 Iteration: 141100 Avg. Training loss: 3.9289 0.2517 sec/batch\n",
      "Epoch 10/10 Iteration: 141200 Avg. Training loss: 4.0248 0.2491 sec/batch\n",
      "Epoch 10/10 Iteration: 141300 Avg. Training loss: 4.0136 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 141400 Avg. Training loss: 4.0370 0.2487 sec/batch\n",
      "Epoch 10/10 Iteration: 141500 Avg. Training loss: 4.0086 0.2483 sec/batch\n",
      "Epoch 10/10 Iteration: 141600 Avg. Training loss: 4.0441 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 141700 Avg. Training loss: 4.0276 0.2475 sec/batch\n",
      "Epoch 10/10 Iteration: 141800 Avg. Training loss: 3.9519 0.2484 sec/batch\n",
      "Epoch 10/10 Iteration: 141900 Avg. Training loss: 3.9646 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 142000 Avg. Training loss: 4.0048 0.2478 sec/batch\n",
      "Nearest to that: note, suggested, fact, argued, argues, argue, noted, suggest,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, external, references, called, daiko, he,\n",
      "Nearest to he: himself, popularize, his, where, wrote, also, transmutation, acknowledges,\n",
      "Nearest to all: virtually, listing, nearly, almost, practically, things, everywhere, assorted,\n",
      "Nearest to used: describe, extensively, widely, refer, denote, term, interchangeably, frequently,\n",
      "Nearest to i: e, ii, am, think, world, my, war, said,\n",
      "Nearest to when: asked, discussing, referring, occurs, especially, guessing, happens, arrived,\n",
      "Nearest to woman: nationality, man, defensemen, children, women, fertility, sonnets, noun,\n",
      "Nearest to dr: blok, physician, positioned, kovacs, rcn, acc, seuss, character,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, rt, casualties, tornadoes, neu, grid,\n",
      "Nearest to test: ban, tests, cricket, matches, hazardous, obsoletes, globes, dumping,\n",
      "Nearest to lived: short, eunuch, long, worked, rabindranath, monks, story, living,\n",
      "Nearest to road: traffic, hypocrisy, hildebrand, routes, racing, accident, extractive, vehicles,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, castilla, intersputnik, ost, intelsat,\n",
      "Epoch 10/10 Iteration: 142100 Avg. Training loss: 3.9579 0.2520 sec/batch\n",
      "Epoch 10/10 Iteration: 142200 Avg. Training loss: 4.0280 0.2498 sec/batch\n",
      "Epoch 10/10 Iteration: 142300 Avg. Training loss: 4.0379 0.2488 sec/batch\n",
      "Epoch 10/10 Iteration: 142400 Avg. Training loss: 4.0105 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 142500 Avg. Training loss: 3.9928 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 142600 Avg. Training loss: 4.0426 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 142700 Avg. Training loss: 3.9990 0.2494 sec/batch\n",
      "Epoch 10/10 Iteration: 142800 Avg. Training loss: 3.9567 0.2494 sec/batch\n",
      "Epoch 10/10 Iteration: 142900 Avg. Training loss: 4.0031 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 143000 Avg. Training loss: 3.9114 0.2487 sec/batch\n",
      "Nearest to that: note, suggested, argued, fact, argue, noted, argues, believe,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, external, called, references, below, daiko,\n",
      "Nearest to he: himself, popularize, his, where, also, nihonshoki, mulhouse, transmutation,\n",
      "Nearest to all: virtually, listing, nearly, almost, practically, assorted, things, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, frequently, denote, term, commonly,\n",
      "Nearest to i: ii, e, am, world, war, my, think, iii,\n",
      "Nearest to when: asked, discussing, referring, occurs, especially, guessing, happens, arrived,\n",
      "Nearest to woman: nationality, man, defensemen, women, children, sonnets, noun, fertility,\n",
      "Nearest to dr: blok, physician, positioned, acc, rcn, kovacs, prof, zaman,\n",
      "Nearest to mainly: primarily, exclusively, mostly, emulsions, predominantly, largely, democratisation, analogical,\n",
      "Nearest to scale: large, scales, richter, tornadoes, rt, casualties, neu, yuga,\n",
      "Nearest to test: ban, tests, cricket, obsoletes, matches, hazardous, dumping, globes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, worked, living, monks, cannibal,\n",
      "Nearest to road: traffic, hildebrand, routes, hypocrisy, accident, extractive, racing, waterbury,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, castilla, intersputnik, ost, intelsat,\n",
      "Epoch 10/10 Iteration: 143100 Avg. Training loss: 3.9906 0.2505 sec/batch\n",
      "Epoch 10/10 Iteration: 143200 Avg. Training loss: 4.0374 0.2491 sec/batch\n",
      "Epoch 10/10 Iteration: 143300 Avg. Training loss: 3.8156 0.2489 sec/batch\n",
      "Epoch 10/10 Iteration: 143400 Avg. Training loss: 4.0447 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 143500 Avg. Training loss: 4.0373 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 143600 Avg. Training loss: 3.9963 0.2489 sec/batch\n",
      "Epoch 10/10 Iteration: 143700 Avg. Training loss: 3.9699 0.2500 sec/batch\n",
      "Epoch 10/10 Iteration: 143800 Avg. Training loss: 3.9694 0.2493 sec/batch\n",
      "Epoch 10/10 Iteration: 143900 Avg. Training loss: 3.9170 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 144000 Avg. Training loss: 4.0135 0.2498 sec/batch\n",
      "Nearest to that: note, suggested, fact, argued, argue, noted, believe, argues,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, external, called, references, below, jost,\n",
      "Nearest to he: himself, popularize, his, where, wrote, mulhouse, montpellier, transmutation,\n",
      "Nearest to all: virtually, nearly, listing, almost, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, denote, term, commonly, frequently,\n",
      "Nearest to i: ii, e, am, think, my, world, war, iii,\n",
      "Nearest to when: asked, discussing, referring, especially, guessing, happens, occurs, arrived,\n",
      "Nearest to woman: nationality, man, children, defensemen, fertility, noun, women, sonnets,\n",
      "Nearest to dr: blok, positioned, physician, acc, kovacs, rcn, zaman, berimbau,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, analogical,\n",
      "Nearest to scale: large, scales, richter, tornadoes, casualties, rt, neu, fifths,\n",
      "Nearest to test: ban, tests, cricket, matches, obsoletes, hazardous, dumping, globes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, living, worked, monks, cannibal,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, routes, accident, extractive, waterbury, nestled,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, intersputnik, castilla, ost, intelsat,\n",
      "Epoch 10/10 Iteration: 144100 Avg. Training loss: 4.0444 0.2519 sec/batch\n",
      "Epoch 10/10 Iteration: 144200 Avg. Training loss: 4.0147 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 144300 Avg. Training loss: 4.0264 0.2495 sec/batch\n",
      "Epoch 10/10 Iteration: 144400 Avg. Training loss: 4.0498 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 144500 Avg. Training loss: 4.0529 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 144600 Avg. Training loss: 4.0078 0.2494 sec/batch\n",
      "Epoch 10/10 Iteration: 144700 Avg. Training loss: 3.9385 0.2489 sec/batch\n",
      "Epoch 10/10 Iteration: 144800 Avg. Training loss: 4.0232 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 144900 Avg. Training loss: 4.0383 0.2477 sec/batch\n",
      "Epoch 10/10 Iteration: 145000 Avg. Training loss: 4.0172 0.2493 sec/batch\n",
      "Nearest to that: note, suggested, fact, argued, noted, argue, believe, suggest,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, external, called, references, links, below,\n",
      "Nearest to he: himself, where, montpellier, popularize, mulhouse, nihonshoki, also, transmutation,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, everywhere, assorted,\n",
      "Nearest to used: describe, extensively, widely, refer, term, denote, commonly, frequently,\n",
      "Nearest to i: ii, e, am, think, world, my, war, iii,\n",
      "Nearest to when: asked, discussing, referring, especially, arrived, occurs, guessing, happens,\n",
      "Nearest to woman: nationality, man, children, defensemen, fertility, noun, sonnets, women,\n",
      "Nearest to dr: blok, physician, positioned, acc, kovacs, rcn, zaman, caucuses,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, analogical,\n",
      "Nearest to scale: large, scales, richter, casualties, tornadoes, rt, neu, yuga,\n",
      "Nearest to test: ban, tests, cricket, hazardous, matches, dumping, obsoletes, globes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, monks, living, cannibal, worked,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, routes, accident, nestled, waterbury, extractive,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, intersputnik, castilla, ost, intelsat,\n",
      "Epoch 10/10 Iteration: 145100 Avg. Training loss: 3.9899 0.2508 sec/batch\n",
      "Epoch 10/10 Iteration: 145200 Avg. Training loss: 4.0324 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 145300 Avg. Training loss: 4.0280 0.2492 sec/batch\n",
      "Epoch 10/10 Iteration: 145400 Avg. Training loss: 4.0095 0.2484 sec/batch\n",
      "Epoch 10/10 Iteration: 145500 Avg. Training loss: 4.0383 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 145600 Avg. Training loss: 4.0179 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 145700 Avg. Training loss: 3.9866 0.2495 sec/batch\n",
      "Epoch 10/10 Iteration: 145800 Avg. Training loss: 3.9190 0.2490 sec/batch\n",
      "Epoch 10/10 Iteration: 145900 Avg. Training loss: 4.0093 0.2472 sec/batch\n",
      "Epoch 10/10 Iteration: 146000 Avg. Training loss: 3.9993 0.2488 sec/batch\n",
      "Nearest to that: note, suggested, argued, argue, fact, believe, noted, suggest,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, external, called, references, below, links,\n",
      "Nearest to he: himself, popularize, where, nihonshoki, acknowledges, thoroughbred, montpellier, cheltenham,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, things, except, assorted,\n",
      "Nearest to used: describe, extensively, widely, refer, denote, commonly, term, frequently,\n",
      "Nearest to i: ii, e, am, my, world, think, war, iii,\n",
      "Nearest to when: asked, discussing, referring, occurs, guessing, especially, happens, arrived,\n",
      "Nearest to woman: nationality, man, fertility, defensemen, children, sonnets, noun, women,\n",
      "Nearest to dr: blok, physician, positioned, acc, kovacs, rcn, zaman, mummy,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, democratisation, due,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, fifths,\n",
      "Nearest to test: ban, tests, obsoletes, cricket, hazardous, dumping, matches, globes,\n",
      "Nearest to lived: short, eunuch, long, monks, living, inhabited, rabindranath, worked,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, accident, routes, vehicles, nestled, extractive,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, ost, castilla, intersputnik, intelsat,\n",
      "Epoch 10/10 Iteration: 146100 Avg. Training loss: 3.9494 0.2508 sec/batch\n",
      "Epoch 10/10 Iteration: 146200 Avg. Training loss: 4.0098 0.2496 sec/batch\n",
      "Epoch 10/10 Iteration: 146300 Avg. Training loss: 3.9759 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 146400 Avg. Training loss: 3.9752 0.2503 sec/batch\n",
      "Epoch 10/10 Iteration: 146500 Avg. Training loss: 4.0667 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 146600 Avg. Training loss: 3.7494 0.2484 sec/batch\n",
      "Epoch 10/10 Iteration: 146700 Avg. Training loss: 3.6626 0.2471 sec/batch\n",
      "Epoch 10/10 Iteration: 146800 Avg. Training loss: 3.8922 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 146900 Avg. Training loss: 3.7985 0.2477 sec/batch\n",
      "Epoch 10/10 Iteration: 147000 Avg. Training loss: 3.9195 0.2484 sec/batch\n",
      "Nearest to that: note, suggested, argued, argue, fact, suggest, believe, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, but, confused, yet,\n",
      "Nearest to also: see, list, known, called, he, external, references, but,\n",
      "Nearest to he: himself, where, also, whom, wrote, popularize, cheltenham, nihonshoki,\n",
      "Nearest to all: listing, nearly, virtually, almost, practically, things, assorted, everywhere,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, frequently, term, denote,\n",
      "Nearest to i: ii, e, am, world, my, think, war, said,\n",
      "Nearest to when: asked, discussing, referring, arrived, especially, occurs, happens, guessing,\n",
      "Nearest to woman: nationality, man, defensemen, children, women, fertility, noun, sonnets,\n",
      "Nearest to dr: physician, blok, positioned, acc, kovacs, zaman, rcn, character,\n",
      "Nearest to mainly: primarily, mostly, exclusively, emulsions, largely, predominantly, analogical, fond,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, numbers,\n",
      "Nearest to test: tests, ban, obsoletes, cricket, matches, hazardous, dumping, globes,\n",
      "Nearest to lived: short, eunuch, long, worked, story, rabindranath, living, return,\n",
      "Nearest to road: traffic, hildebrand, hypocrisy, routes, accident, donovan, waterbury, nestled,\n",
      "Nearest to ocean: atlantic, pacific, indian, posed, ost, castilla, intersputnik, intelsat,\n",
      "Epoch 10/10 Iteration: 147100 Avg. Training loss: 3.9918 0.2525 sec/batch\n",
      "Epoch 10/10 Iteration: 147200 Avg. Training loss: 4.0015 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 147300 Avg. Training loss: 4.0392 0.2488 sec/batch\n",
      "Epoch 10/10 Iteration: 147400 Avg. Training loss: 3.9777 0.2483 sec/batch\n",
      "Epoch 10/10 Iteration: 147500 Avg. Training loss: 4.0518 0.2487 sec/batch\n",
      "Epoch 10/10 Iteration: 147600 Avg. Training loss: 4.0522 0.2491 sec/batch\n",
      "Epoch 10/10 Iteration: 147700 Avg. Training loss: 4.0644 0.2484 sec/batch\n",
      "Epoch 10/10 Iteration: 147800 Avg. Training loss: 4.0236 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 147900 Avg. Training loss: 4.0277 0.2479 sec/batch\n",
      "Epoch 10/10 Iteration: 148000 Avg. Training loss: 3.8616 0.2484 sec/batch\n",
      "Nearest to that: note, argue, fact, suggested, argued, suggest, believe, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, external, called, references, below, but,\n",
      "Nearest to he: himself, popularize, where, wrote, cheltenham, acknowledges, guangzhou, his,\n",
      "Nearest to all: nearly, virtually, listing, almost, practically, things, everywhere, except,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, frequently, term, denote,\n",
      "Nearest to i: ii, e, am, think, world, my, war, said,\n",
      "Nearest to when: asked, discussing, referring, arrived, happens, occurs, kidman, especially,\n",
      "Nearest to woman: nationality, man, defensemen, women, noun, fertility, children, sonnets,\n",
      "Nearest to dr: blok, physician, positioned, rcn, acc, zaman, kovacs, mummy,\n",
      "Nearest to mainly: primarily, mostly, exclusively, largely, emulsions, predominantly, democratisation, fond,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, numbers,\n",
      "Nearest to test: tests, ban, cricket, obsoletes, dumping, matches, hazardous, globes,\n",
      "Nearest to lived: short, eunuch, rabindranath, settled, long, worked, story, coney,\n",
      "Nearest to road: traffic, hildebrand, routes, hypocrisy, accident, nestled, silk, extractive,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, ost, castilla, intersputnik, station,\n",
      "Epoch 10/10 Iteration: 148100 Avg. Training loss: 3.9937 0.2518 sec/batch\n",
      "Epoch 10/10 Iteration: 148200 Avg. Training loss: 4.0130 0.2480 sec/batch\n",
      "Epoch 10/10 Iteration: 148300 Avg. Training loss: 3.9850 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 148400 Avg. Training loss: 3.9568 0.2483 sec/batch\n",
      "Epoch 10/10 Iteration: 148500 Avg. Training loss: 3.9980 0.2482 sec/batch\n",
      "Epoch 10/10 Iteration: 148600 Avg. Training loss: 4.0150 0.2490 sec/batch\n",
      "Epoch 10/10 Iteration: 148700 Avg. Training loss: 4.0551 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 148800 Avg. Training loss: 3.9665 0.2492 sec/batch\n",
      "Epoch 10/10 Iteration: 148900 Avg. Training loss: 4.0074 0.2492 sec/batch\n",
      "Epoch 10/10 Iteration: 149000 Avg. Training loss: 3.8312 0.2481 sec/batch\n",
      "Nearest to that: note, fact, argue, suggested, argued, believe, suggest, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, yet,\n",
      "Nearest to also: see, list, known, external, called, references, below, links,\n",
      "Nearest to he: himself, where, wrote, cheltenham, popularize, whom, mulhouse, acknowledges,\n",
      "Nearest to all: virtually, nearly, almost, listing, practically, everywhere, things, assorted,\n",
      "Nearest to used: describe, extensively, widely, refer, commonly, denote, frequently, term,\n",
      "Nearest to i: ii, e, am, think, world, my, war, said,\n",
      "Nearest to when: asked, discussing, referring, happens, arrived, occurs, especially, kidman,\n",
      "Nearest to woman: nationality, children, man, women, noun, fertility, defensemen, sonnets,\n",
      "Nearest to dr: physician, blok, positioned, acc, kovacs, zaman, rcn, applicability,\n",
      "Nearest to mainly: primarily, mostly, exclusively, largely, emulsions, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, tornadoes, rt, casualties, neu, numbers,\n",
      "Nearest to test: ban, tests, cricket, obsoletes, hazardous, matches, dumping, globes,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, story, worked, settled, cannibal,\n",
      "Nearest to road: traffic, routes, hildebrand, hypocrisy, accident, vehicles, nestled, silk,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, ost, castilla, intersputnik, intelsat,\n",
      "Epoch 10/10 Iteration: 149100 Avg. Training loss: 3.8465 0.2519 sec/batch\n",
      "Epoch 10/10 Iteration: 149200 Avg. Training loss: 4.0042 0.2476 sec/batch\n",
      "Epoch 10/10 Iteration: 149300 Avg. Training loss: 3.9838 0.2488 sec/batch\n",
      "Epoch 10/10 Iteration: 149400 Avg. Training loss: 3.8428 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 149500 Avg. Training loss: 4.0015 0.2507 sec/batch\n",
      "Epoch 10/10 Iteration: 149600 Avg. Training loss: 4.0348 0.2494 sec/batch\n",
      "Epoch 10/10 Iteration: 149700 Avg. Training loss: 3.9824 0.2485 sec/batch\n",
      "Epoch 10/10 Iteration: 149800 Avg. Training loss: 3.9651 0.2494 sec/batch\n",
      "Epoch 10/10 Iteration: 149900 Avg. Training loss: 3.9500 0.2486 sec/batch\n",
      "Epoch 10/10 Iteration: 150000 Avg. Training loss: 3.9622 0.2482 sec/batch\n",
      "Nearest to that: note, fact, argue, suggested, argued, suggest, believe, announced,\n",
      "Nearest to not: did, does, necessarily, do, uncommon, confused, but, until,\n",
      "Nearest to also: see, list, known, external, called, references, but, below,\n",
      "Nearest to he: himself, where, wrote, cheltenham, whom, popularize, his, guangzhou,\n",
      "Nearest to all: nearly, virtually, listing, almost, practically, things, everywhere, assorted,\n",
      "Nearest to used: describe, extensively, widely, refer, term, commonly, denote, frequently,\n",
      "Nearest to i: ii, e, am, think, my, world, war, iii,\n",
      "Nearest to when: asked, discussing, referring, arrived, happens, occurs, especially, guessing,\n",
      "Nearest to woman: nationality, man, children, noun, defensemen, women, fertility, sonnets,\n",
      "Nearest to dr: physician, blok, positioned, acc, rcn, character, kovacs, zaman,\n",
      "Nearest to mainly: primarily, mostly, exclusively, largely, emulsions, predominantly, democratisation, elas,\n",
      "Nearest to scale: large, scales, richter, casualties, rt, tornadoes, neu, numbers,\n",
      "Nearest to test: tests, ban, obsoletes, hazardous, cricket, dumping, matches, ozone,\n",
      "Nearest to lived: short, eunuch, long, rabindranath, worked, story, settled, monks,\n",
      "Nearest to road: traffic, routes, hildebrand, accident, vehicles, racing, hypocrisy, donovan,\n",
      "Nearest to ocean: atlantic, indian, pacific, posed, castilla, ost, intersputnik, intelsat,\n",
      "Epoch 10/10 Iteration: 150100 Avg. Training loss: 3.9899 0.2514 sec/batch\n",
      "Epoch 10/10 Iteration: 150200 Avg. Training loss: 4.0369 0.2487 sec/batch\n",
      "Epoch 10/10 Iteration: 150300 Avg. Training loss: 4.0683 0.2487 sec/batch\n"
     ]
    }
   ],
   "source": [
    "epochs = 10\n",
    "batch_size = 1000\n",
    "window_size = 10\n",
    "\n",
    "with train_graph.as_default():\n",
    "    saver = tf.train.Saver()\n",
    "\n",
    "with tf.Session(graph=train_graph) as sess:\n",
    "    iteration = 1\n",
    "    loss = 0\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "\n",
    "    for e in range(1, epochs+1):\n",
    "        batches = get_batches(train_words, batch_size, window_size)\n",
    "        start = time.time()\n",
    "        for x, y in batches:\n",
    "            \n",
    "            feed = {inputs: x,\n",
    "                    labels: np.array(y)[:, None]}\n",
    "            train_loss, _ = sess.run([cost, optimizer], feed_dict=feed)\n",
    "            \n",
    "            loss += train_loss\n",
    "            \n",
    "            if iteration % 100 == 0: \n",
    "                end = time.time()\n",
    "                print(\"Epoch {}/{}\".format(e, epochs),\n",
    "                      \"Iteration: {}\".format(iteration),\n",
    "                      \"Avg. Training loss: {:.4f}\".format(loss/100),\n",
    "                      \"{:.4f} sec/batch\".format((end-start)/100))\n",
    "                loss = 0\n",
    "                start = time.time()\n",
    "            \n",
    "            if iteration % 1000 == 0:\n",
    "                ## From Thushan Ganegedara's implementation\n",
    "                # note that this is expensive (~20% slowdown if computed every 500 steps)\n",
    "                sim = similarity.eval()\n",
    "                for i in range(valid_size):\n",
    "                    valid_word = int_to_vocab[valid_examples[i]]\n",
    "                    top_k = 8 # number of nearest neighbors\n",
    "                    nearest = (-sim[i, :]).argsort()[1:top_k+1]\n",
    "                    log = 'Nearest to %s:' % valid_word\n",
    "                    for k in range(top_k):\n",
    "                        close_word = int_to_vocab[nearest[k]]\n",
    "                        log = '%s %s,' % (log, close_word)\n",
    "                    print(log)\n",
    "            \n",
    "            iteration += 1\n",
    "    save_path = saver.save(sess, \"checkpoints/text8.ckpt\")\n",
    "    embed_mat = sess.run(normalized_embedding)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "Restore the trained network if you need to:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "with train_graph.as_default():\n",
    "    saver = tf.train.Saver()\n",
    "\n",
    "with tf.Session(graph=train_graph) as sess:\n",
    "    saver.restore(sess, tf.train.latest_checkpoint('checkpoints'))\n",
    "    embed_mat = sess.run(embedding)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "## Visualizing the word vectors\n",
    "\n",
    "Below we'll use T-SNE to visualize how our high-dimensional word vectors cluster together. T-SNE is used to project these vectors into two dimensions while preserving local stucture. Check out [this post from Christopher Olah](http://colah.github.io/posts/2014-10-Visualizing-MNIST/) to learn more about T-SNE and other ways to visualize high-dimensional data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "%config InlineBackend.figure_format = 'retina'\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.manifold import TSNE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "array must not contain infs or NaNs",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-20-a4718e25bc97>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mviz_words\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m500\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mtsne\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTSNE\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0membed_tsne\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtsne\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit_transform\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0membed_mat\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mviz_words\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/Users/dave/anaconda3/lib/python3.5/site-packages/sklearn/manifold/t_sne.py\u001b[0m in \u001b[0;36mfit_transform\u001b[0;34m(self, X, y)\u001b[0m\n\u001b[1;32m    882\u001b[0m             \u001b[0mEmbedding\u001b[0m \u001b[0mof\u001b[0m \u001b[0mthe\u001b[0m \u001b[0mtraining\u001b[0m \u001b[0mdata\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mlow\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mdimensional\u001b[0m \u001b[0mspace\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    883\u001b[0m         \"\"\"\n\u001b[0;32m--> 884\u001b[0;31m         \u001b[0membedding\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    885\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0membedding_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0membedding\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    886\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0membedding_\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dave/anaconda3/lib/python3.5/site-packages/sklearn/manifold/t_sne.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, skip_num_points)\u001b[0m\n\u001b[1;32m    787\u001b[0m                           \u001b[0mX_embedded\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mX_embedded\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    788\u001b[0m                           \u001b[0mneighbors\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mneighbors_nn\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 789\u001b[0;31m                           skip_num_points=skip_num_points)\n\u001b[0m\u001b[1;32m    790\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    791\u001b[0m     def _tsne(self, P, degrees_of_freedom, n_samples, random_state,\n",
      "\u001b[0;32m/Users/dave/anaconda3/lib/python3.5/site-packages/sklearn/manifold/t_sne.py\u001b[0m in \u001b[0;36m_tsne\u001b[0;34m(self, P, degrees_of_freedom, n_samples, random_state, X_embedded, neighbors, skip_num_points)\u001b[0m\n\u001b[1;32m    845\u001b[0m         \u001b[0mopt_args\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'it'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mit\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    846\u001b[0m         params, kl_divergence, it = _gradient_descent(obj_func, params,\n\u001b[0;32m--> 847\u001b[0;31m                                                       **opt_args)\n\u001b[0m\u001b[1;32m    848\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    849\u001b[0m             print(\"[t-SNE] KL divergence after %d iterations with early \"\n",
      "\u001b[0;32m/Users/dave/anaconda3/lib/python3.5/site-packages/sklearn/manifold/t_sne.py\u001b[0m in \u001b[0;36m_gradient_descent\u001b[0;34m(objective, p0, it, n_iter, objective_error, n_iter_check, n_iter_without_progress, momentum, learning_rate, min_gain, min_grad_norm, min_error_diff, verbose, args, kwargs)\u001b[0m\n\u001b[1;32m    386\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mit\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_iter\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    387\u001b[0m         \u001b[0mnew_error\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mgrad\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mobjective\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mp\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 388\u001b[0;31m         \u001b[0mgrad_norm\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlinalg\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnorm\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mgrad\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    389\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    390\u001b[0m         \u001b[0minc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mupdate\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0mgrad\u001b[0m \u001b[0;34m>=\u001b[0m \u001b[0;36m0.0\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dave/anaconda3/lib/python3.5/site-packages/scipy/linalg/misc.py\u001b[0m in \u001b[0;36mnorm\u001b[0;34m(a, ord, axis, keepdims)\u001b[0m\n\u001b[1;32m    127\u001b[0m     \"\"\"\n\u001b[1;32m    128\u001b[0m     \u001b[0;31m# Differs from numpy only in non-finite handling and the use of blas.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 129\u001b[0;31m     \u001b[0ma\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masarray_chkfinite\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    130\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    131\u001b[0m     \u001b[0;31m# Only use optimized norms if axis and keepdims are not specified.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/dave/anaconda3/lib/python3.5/site-packages/numpy/lib/function_base.py\u001b[0m in \u001b[0;36masarray_chkfinite\u001b[0;34m(a, dtype, order)\u001b[0m\n\u001b[1;32m   1213\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0ma\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdtype\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mchar\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mtypecodes\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'AllFloat'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0misfinite\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mall\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1214\u001b[0m         raise ValueError(\n\u001b[0;32m-> 1215\u001b[0;31m             \"array must not contain infs or NaNs\")\n\u001b[0m\u001b[1;32m   1216\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0ma\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1217\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: array must not contain infs or NaNs"
     ]
    }
   ],
   "source": [
    "viz_words = 500\n",
    "tsne = TSNE()\n",
    "embed_tsne = tsne.fit_transform(embed_mat[:viz_words, :])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'embed_tsne' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-21-f98e3593a546>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mfig\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0max\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msubplots\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfigsize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m14\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m14\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0midx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mviz_words\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m     \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscatter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0membed_tsne\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0midx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m:\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcolor\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'steelblue'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m     \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mannotate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mint_to_vocab\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0midx\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0membed_tsne\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0midx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0membed_tsne\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0midx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0malpha\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0.7\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'embed_tsne' is not defined"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABnQAAAY3CAYAAABWM5GcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nOUEHAAAAAACguf8DyUJWk5egZtIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x121acdeb8>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 795,
       "width": 826
      }
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(14, 14))\n",
    "for idx in range(viz_words):\n",
    "    plt.scatter(*embed_tsne[idx, :], color='steelblue')\n",
    "    plt.annotate(int_to_vocab[idx], (embed_tsne[idx, 0], embed_tsne[idx, 1]), alpha=0.7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [default]",
   "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}