Skip-gram word2vec

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 translations.

Readings

Here are the resources I used to build this notebook. I suggest reading these either beforehand or while you're working on this material.

A really good conceptual overview of word2vec from Chris McCormick
First word2vec paper from Mikolov et al.
NIPS paper with improvements for word2vec also from Mikolov et al.
An implementation of word2vec from Thushan Ganegedara
TensorFlow word2vec tutorial

Word embeddings

When you're dealing with language and words, 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 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.

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.

First up, importing packages.



In [1]:

    
import time

import numpy as np
import tensorflow as tf

import utils

Load the text8 dataset, 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.



In [2]:

    
from urllib.request import urlretrieve
from os.path import isfile, isdir
from tqdm import tqdm
import zipfile

dataset_folder_path = 'data'
dataset_filename = 'text8.zip'
dataset_name = 'Text8 Dataset'

class DLProgress(tqdm):
    last_block = 0

    def hook(self, block_num=1, block_size=1, total_size=None):
        self.total = total_size
        self.update((block_num - self.last_block) * block_size)
        self.last_block = block_num

if not isfile(dataset_filename):
    with DLProgress(unit='B', unit_scale=True, miniters=1, desc=dataset_name) as pbar:
        urlretrieve(
            'http://mattmahoney.net/dc/text8.zip',
            dataset_filename,
            pbar.hook)

if not isdir(dataset_folder_path):
    with zipfile.ZipFile(dataset_filename) as zip_ref:
        zip_ref.extractall(dataset_folder_path)
        
with open('data/text8') as f:
    text = f.read()

Preprocessing

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.



In [3]:

    
words = utils.preprocess(text)
print(words[:30])









    



['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']



In [4]:

    
print("Total words: {}".format(len(words)))
print("Unique words: {}".format(len(set(words))))









    



Total words: 16680599
Unique words: 63641

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.



In [5]:

    
vocab_to_int, int_to_vocab = utils.create_lookup_tables(words)
int_words = [vocab_to_int[word] for word in words]

Subsampling

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

$$ P(w_i) = 1 - \sqrt{\frac{t}{f(w_i)}} $$

where $t$ is a threshold parameter and $f(w_i)$ is the frequency of word $w_i$ in the total dataset.

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.

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.



In [6]:

    
## Your code here
import collections
import math
import random
count = collections.Counter()
for i in int_words:
    count[i] += 1
total_words = len(words)
sample_rate_t = 0.001
p_w = {key: (1-math.sqrt(sample_rate_t/(value/total_words))) for key, value in count.items()}
print(len(int_words))
train_words = []# The final subsampled word list
for i in int_words:
    if random.random() > p_w[i]:
        train_words.append(i)
print(len(train_words))

Making batches

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 $C$.

From Mikolov et al.:

"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."

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.



In [7]:

    
def get_target(words, idx, window_size=5):
    ''' Get a list of words in a window around an index. '''
    r = random.randint(1, window_size)
    lower = max(0, idx - r)
    upper = min(len(words), idx + r + 1)
    return list(set(words[lower:idx] + words[idx+1:upper]))

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.



In [8]:

    
def get_batches(words, batch_size, window_size=5):
    ''' Create a generator of word batches as a tuple (inputs, targets) '''
    
    n_batches = len(words)//batch_size
    
    # only full batches
    words = words[:n_batches*batch_size]
    
    for idx in range(0, len(words), batch_size):
        x, y = [], []
        batch = words[idx:idx+batch_size]
        for ii in range(len(batch)):
            batch_x = batch[ii]
            batch_y = get_target(batch, ii, window_size)
            y.extend(batch_y)
            x.extend([batch_x]*len(batch_y))
        yield x, y

Building the graph

From Chris McCormick's blog, we can see the general structure of our network.

The input words are passed in as one-hot encoded vectors. 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.

The idea here is to train the hidden layer weight matrix to find efficient representations for our words. This weight matrix is usually called the embedding matrix or embedding look-up table. 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.

I'm going to have you build the graph in stages now. First off, creating the inputs and labels placeholders like normal.

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.



In [9]:

    
train_graph = tf.Graph()
with train_graph.as_default():
    inputs = tf.placeholder(tf.int32,  shape=[None], name="inputs")
    labels = tf.placeholder(tf.float32, shape=[None, None], name="labels")

Embedding

The embedding matrix has a size of the number of words by the number of neurons 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 one-hot encoded vectors for our inputs. When you do the matrix multiplication of the one-hot vector with the embedding matrix, you end up selecting only one row out of the entire matrix:

You don't actually need to do the matrix multiplication, you just need to select the row in the embedding matrix that corresponds to the input word. Then, the embedding matrix becomes a lookup table, you're looking up a vector the size of the hidden layer that represents the input word.

Exercise: Tensorflow provides a convenient function 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. This TensorFlow tutorial will help if you get stuck.



In [11]:

    
n_vocab = len(int_to_vocab)
n_embedding = 300 # Number of embedding features 
with train_graph.as_default():
    embedding = tf.Variable(tf.random_uniform((n_vocab,n_embedding), minval=-1, maxval=1)) # create embedding weight matrix here
    embed = tf.nn.embedding_lookup(embedding, inputs)# use tf.nn.embedding_lookup to get the hidden layer output

Negative sampling

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". Tensorflow has a convenient function to do this, tf.nn.sampled_softmax_loss.

Exercise: Below, create weights and biases for the softmax layer. Then, use tf.nn.sampled_softmax_loss to calculate the loss. Be sure to read the documentation to figure out how it works.



In [12]:

    
# Number of negative labels to sample
n_sampled = 100
with train_graph.as_default():
    softmax_w = tf.Variable(tf.truncated_normal((n_vocab, n_embedding), stddev=0.1))
    softmax_b = tf.Variable(tf.zeros(n_vocab))# create softmax biases here
    
    # Calculate the loss using negative sampling
    loss = tf.nn.sampled_softmax_loss(softmax_w, softmax_b, labels, embed, n_sampled, n_vocab)
    
    cost = tf.reduce_mean(loss)
    optimizer = tf.train.AdamOptimizer().minimize(cost)

Validation

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.



In [13]:

    
with train_graph.as_default():
    ## From Thushan Ganegedara's implementation
    valid_size = 16 # Random set of words to evaluate similarity on.
    valid_window = 100
    # pick 8 samples from (0,100) and (1000,1100) each ranges. lower id implies more frequent 
    valid_examples = np.array(random.sample(range(valid_window), valid_size//2))
    valid_examples = np.append(valid_examples, 
                               random.sample(range(1000,1000+valid_window), valid_size//2))

    valid_dataset = tf.constant(valid_examples, dtype=tf.int32)
    
    # We use the cosine distance:
    norm = tf.sqrt(tf.reduce_sum(tf.square(embedding), 1, keep_dims=True))
    normalized_embedding = embedding / norm
    valid_embedding = tf.nn.embedding_lookup(normalized_embedding, valid_dataset)
    similarity = tf.matmul(valid_embedding, tf.transpose(normalized_embedding))



In [14]:

    
# If the checkpoints directory doesn't exist:
!mkdir checkpoints









    



A subdirectory or file checkpoints already exists.

Training

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.



In [15]:

    
epochs = 10
batch_size = 1000
window_size = 10

with train_graph.as_default():
    saver = tf.train.Saver()

with tf.Session(graph=train_graph) as sess:
    iteration = 1
    loss = 0
    sess.run(tf.global_variables_initializer())

    for e in range(1, epochs+1):
        batches = get_batches(train_words, batch_size, window_size)
        start = time.time()
        for x, y in batches:
            
            feed = {inputs: x,
                    labels: np.array(y)[:, None]}
            train_loss, _ = sess.run([cost, optimizer], feed_dict=feed)
            
            loss += train_loss
            
            if iteration % 100 == 0: 
                end = time.time()
                print("Epoch {}/{}".format(e, epochs),
                      "Iteration: {}".format(iteration),
                      "Avg. Training loss: {:.4f}".format(loss/100),
                      "{:.4f} sec/batch".format((end-start)/100))
                loss = 0
                start = time.time()
            
            if iteration % 1000 == 0:
                ## From Thushan Ganegedara's implementation
                # note that this is expensive (~20% slowdown if computed every 500 steps)
                sim = similarity.eval()
                for i in range(valid_size):
                    valid_word = int_to_vocab[valid_examples[i]]
                    top_k = 8 # number of nearest neighbors
                    nearest = (-sim[i, :]).argsort()[1:top_k+1]
                    log = 'Nearest to %s:' % valid_word
                    for k in range(top_k):
                        close_word = int_to_vocab[nearest[k]]
                        log = '%s %s,' % (log, close_word)
                    print(log)
            
            iteration += 1
    save_path = saver.save(sess, "checkpoints/text8.ckpt")
    embed_mat = sess.run(normalized_embedding)









    



Epoch 1/10 Iteration: 100 Avg. Training loss: 7.3208 0.3171 sec/batch
Epoch 1/10 Iteration: 200 Avg. Training loss: 7.1847 0.2999 sec/batch
Epoch 1/10 Iteration: 300 Avg. Training loss: 6.9375 0.2739 sec/batch
Epoch 1/10 Iteration: 400 Avg. Training loss: 6.7145 0.2782 sec/batch
Epoch 1/10 Iteration: 500 Avg. Training loss: 6.3271 0.2722 sec/batch
Epoch 1/10 Iteration: 600 Avg. Training loss: 5.9755 0.2731 sec/batch
Epoch 1/10 Iteration: 700 Avg. Training loss: 5.7079 0.2911 sec/batch
Epoch 1/10 Iteration: 800 Avg. Training loss: 5.5193 0.2840 sec/batch
Epoch 1/10 Iteration: 900 Avg. Training loss: 5.5146 0.2832 sec/batch
Epoch 1/10 Iteration: 1000 Avg. Training loss: 5.4031 0.2788 sec/batch
Nearest to into: judge, crossword, lanka, taisho, saxony, charged, u, reboot,
Nearest to during: charybdis, revolutionary, pedestrian, invades, explain, different, toys, patience,
Nearest to about: based, gorbachev, hippie, romans, students, shortening, principle, defeating,
Nearest to five: biogeography, component, gulags, classis, jutish, jesus, supplied, spoofed,
Nearest to six: b, lenses, becomes, journeys, prospered, bloody, it, plurality,
Nearest to some: graveolens, statistics, too, dialogs, spirit, rich, brother, die,
Nearest to however: hometown, geometric, water, presented, domination, january, queens, your,
Nearest to war: gone, groups, arunachal, informed, impact, sailed, homerun, ruby,
Nearest to experience: twa, according, hellene, nominee, another, moles, abbess, evolve,
Nearest to magazine: macmullen, plains, screech, unthinkable, polynesian, roster, welsh, deliberately,
Nearest to joseph: retrofitted, gdf, provenance, scarecrow, shop, higginbotham, unsuitable, persons,
Nearest to universe: mecca, inventor, summit, gallen, harrington, habsburgs, truck, scipione,
Nearest to defense: iea, erudite, mixers, stata, dyadic, comix, weber, theoretician,
Nearest to square: amyloid, education, rim, follows, propounded, views, ethologist, mumy,
Nearest to egypt: leagues, bikini, labor, treat, focuses, countenance, heavily, solomon,
Nearest to smith: garner, thera, geddes, obsolescent, goldwyn, head, doorstep, heinrich,
Epoch 1/10 Iteration: 1100 Avg. Training loss: 5.2907 0.2745 sec/batch
Epoch 1/10 Iteration: 1200 Avg. Training loss: 5.2739 0.2715 sec/batch
Epoch 1/10 Iteration: 1300 Avg. Training loss: 5.1961 0.2737 sec/batch
Epoch 1/10 Iteration: 1400 Avg. Training loss: 5.1809 0.2766 sec/batch
Epoch 1/10 Iteration: 1500 Avg. Training loss: 5.1134 0.2778 sec/batch
Epoch 1/10 Iteration: 1600 Avg. Training loss: 5.1082 0.2743 sec/batch
Epoch 1/10 Iteration: 1700 Avg. Training loss: 5.0183 0.2723 sec/batch
Epoch 1/10 Iteration: 1800 Avg. Training loss: 5.0808 0.2809 sec/batch
Epoch 1/10 Iteration: 1900 Avg. Training loss: 5.0521 0.2784 sec/batch
Epoch 1/10 Iteration: 2000 Avg. Training loss: 5.0141 0.2906 sec/batch
Nearest to into: crossword, saxony, judge, lanka, fender, administer, charged, kicked,
Nearest to during: toys, pedestrian, charybdis, masterful, invades, scripture, categorized, scheduling,
Nearest to about: hippie, shortening, gorbachev, based, hafez, romans, defeating, stabilize,
Nearest to five: spoofed, biogeography, jutish, classis, dazzling, gulags, sink, supplied,
Nearest to six: lenses, roofs, prospered, b, printer, isbn, journeys, enmity,
Nearest to some: graveolens, statistics, unambiguous, too, spirit, dialogs, tournament, terminates,
Nearest to however: hometown, queens, basically, presented, geometric, domination, ttir, prominently,
Nearest to war: informed, gone, homerun, pistols, sailed, arunachal, knots, hearth,
Nearest to experience: twa, nominee, moles, hellene, relevance, evolve, observations, tze,
Nearest to magazine: macmullen, plains, unthinkable, screech, seine, roster, polynesian, cariou,
Nearest to joseph: retrofitted, gdf, higginbotham, provenance, caithness, unsuitable, scarecrow, shop,
Nearest to universe: mecca, financed, habsburgs, gallen, truck, weighted, summit, bragg,
Nearest to defense: iea, erudite, dyadic, weber, mixers, comix, sewn, stata,
Nearest to square: amyloid, ethologist, rim, eddic, propounded, follows, triumph, dirac,
Nearest to egypt: leagues, bikini, spree, countenance, labor, solomon, aunt, iglesias,
Nearest to smith: garner, obsolescent, thera, geddes, pins, heinrich, goldwyn, doorstep,
Epoch 1/10 Iteration: 2100 Avg. Training loss: 4.9718 0.3068 sec/batch
Epoch 1/10 Iteration: 2200 Avg. Training loss: 4.9133 0.2768 sec/batch
Epoch 1/10 Iteration: 2300 Avg. Training loss: 4.9466 0.2824 sec/batch
Epoch 1/10 Iteration: 2400 Avg. Training loss: 4.9604 0.3397 sec/batch
Epoch 1/10 Iteration: 2500 Avg. Training loss: 4.9223 0.3298 sec/batch
Epoch 1/10 Iteration: 2600 Avg. Training loss: 4.9141 0.3602 sec/batch
Epoch 1/10 Iteration: 2700 Avg. Training loss: 4.8829 0.3720 sec/batch
Epoch 1/10 Iteration: 2800 Avg. Training loss: 4.8360 0.3731 sec/batch
Epoch 1/10 Iteration: 2900 Avg. Training loss: 4.7423 0.3119 sec/batch
Epoch 1/10 Iteration: 3000 Avg. Training loss: 4.8631 0.2789 sec/batch
Nearest to into: saxony, crossword, lanka, judge, wholeheartedly, reboot, administer, charged,
Nearest to during: toys, pedestrian, masterful, categorized, invades, charybdis, kidman, bringing,
Nearest to about: shortening, gorbachev, hippie, hafez, based, romans, stabilize, yassin,
Nearest to five: spoofed, dazzling, tolstoy, jutish, biogeography, gulags, classis, tons,
Nearest to six: prospered, enmity, isbn, lenses, roofs, elijah, journeys, bloody,
Nearest to some: graveolens, unambiguous, spirit, too, statistics, retort, terminates, dialogs,
Nearest to however: hometown, queens, basically, presented, geometric, domination, illustrative, prominently,
Nearest to war: homerun, pistols, arunachal, dramas, informed, gone, raiders, hearth,
Nearest to experience: twa, nominee, hellene, moles, relevance, abbess, bitterly, incoherent,
Nearest to magazine: macmullen, plains, carolingian, unthinkable, screech, seine, cariou, roster,
Nearest to joseph: retrofitted, gdf, scarecrow, higginbotham, provenance, caithness, unsuitable, shop,
Nearest to universe: mecca, financed, weighted, habsburgs, gallen, truck, harrington, bragg,
Nearest to defense: iea, erudite, dyadic, mixers, comix, weber, sewn, stata,
Nearest to square: amyloid, ethologist, dirac, eddic, rim, triumph, propounded, mumy,
Nearest to egypt: leagues, bikini, spree, solomon, aunt, countenance, iglesias, heavily,
Nearest to smith: garner, thera, obsolescent, geddes, doorstep, goldwyn, pins, hips,
Epoch 1/10 Iteration: 3100 Avg. Training loss: 4.8275 0.2792 sec/batch
Epoch 1/10 Iteration: 3200 Avg. Training loss: 4.8073 0.2762 sec/batch
Epoch 1/10 Iteration: 3300 Avg. Training loss: 4.8101 0.2758 sec/batch
Epoch 1/10 Iteration: 3400 Avg. Training loss: 4.7920 0.2818 sec/batch
Epoch 1/10 Iteration: 3500 Avg. Training loss: 4.7605 0.2795 sec/batch
Epoch 1/10 Iteration: 3600 Avg. Training loss: 4.8176 0.2777 sec/batch
Epoch 1/10 Iteration: 3700 Avg. Training loss: 4.8128 0.2789 sec/batch
Epoch 1/10 Iteration: 3800 Avg. Training loss: 4.7717 0.2886 sec/batch
Epoch 1/10 Iteration: 3900 Avg. Training loss: 4.7773 0.2838 sec/batch
Epoch 1/10 Iteration: 4000 Avg. Training loss: 4.7240 0.2763 sec/batch
Nearest to into: wholeheartedly, saxony, reboot, crossword, judge, lanka, mariana, charged,
Nearest to during: charybdis, toys, masterful, pedestrian, kidman, invades, bonnie, categorized,
Nearest to about: shortening, gorbachev, hippie, hafez, cyclopedia, based, durations, defeating,
Nearest to five: spoofed, dazzling, tolstoy, jutish, classis, biogeography, tons, gulags,
Nearest to six: isbn, enmity, roofs, prospered, elijah, lenses, definable, owned,
Nearest to some: unambiguous, graveolens, too, spirit, retort, taoist, terminates, dialogs,
Nearest to however: hometown, queens, illustrative, presented, basically, cheyenne, geometric, universality,
Nearest to war: homerun, hearth, pistols, arunachal, gyula, dramas, raiders, informed,
Nearest to experience: twa, nominee, moles, hellene, another, incoherent, according, hom,
Nearest to magazine: macmullen, screech, carolingian, unthinkable, plains, seine, cariou, roster,
Nearest to joseph: retrofitted, gdf, scarecrow, provenance, higginbotham, shop, caithness, unsuitable,
Nearest to universe: weighted, gallen, habsburgs, financed, mecca, bragg, truck, scipione,
Nearest to defense: iea, erudite, dyadic, comix, sewn, stata, palpitations, cay,
Nearest to square: amyloid, ethologist, dirac, eddic, propounded, follows, mumy, rim,
Nearest to egypt: leagues, bikini, spree, aunt, iglesias, solomon, jericho, labor,
Nearest to smith: garner, thera, geddes, doorstep, obsolescent, purr, hips, heinrich,
Epoch 1/10 Iteration: 4100 Avg. Training loss: 4.7603 0.2767 sec/batch
Epoch 1/10 Iteration: 4200 Avg. Training loss: 4.7861 0.2746 sec/batch
Epoch 1/10 Iteration: 4300 Avg. Training loss: 4.7279 0.2786 sec/batch
Epoch 1/10 Iteration: 4400 Avg. Training loss: 4.7603 0.2763 sec/batch
Epoch 1/10 Iteration: 4500 Avg. Training loss: 4.6851 0.2740 sec/batch
Epoch 1/10 Iteration: 4600 Avg. Training loss: 4.7657 0.2757 sec/batch
Epoch 1/10 Iteration: 4700 Avg. Training loss: 4.7417 0.2706 sec/batch
Epoch 1/10 Iteration: 4800 Avg. Training loss: 4.7431 0.2776 sec/batch
Epoch 1/10 Iteration: 4900 Avg. Training loss: 4.7468 0.2823 sec/batch
Epoch 1/10 Iteration: 5000 Avg. Training loss: 4.6924 0.2820 sec/batch
Nearest to into: crossword, wholeheartedly, saxony, mariana, reboot, judge, lanka, reaper,
Nearest to during: charybdis, toys, kidman, bonnie, invades, pedestrian, imprisoned, revolutionary,
Nearest to about: shortening, gorbachev, hippie, hafez, durations, cyclopedia, thacker, scientifiques,
Nearest to five: dazzling, spoofed, jutish, gulags, biogeography, tolstoy, classis, patrols,
Nearest to six: isbn, b, enmity, definable, roofs, elijah, owned, prospered,
Nearest to some: unambiguous, graveolens, taoist, statistics, retort, spirit, too, timbres,
Nearest to however: hometown, illustrative, presented, queens, cheyenne, basically, sofa, universality,
Nearest to war: homerun, hindenburg, gyula, dramas, peek, pistols, arunachal, gone,
Nearest to experience: twa, nominee, moles, hellene, incoherent, another, according, hom,
Nearest to magazine: macmullen, screech, unthinkable, carolingian, plains, seine, cariou, polynesian,
Nearest to joseph: retrofitted, higginbotham, gdf, scarecrow, provenance, shop, votive, caithness,
Nearest to universe: weighted, gallen, habsburgs, financed, bragg, mecca, hermeticism, truck,
Nearest to defense: iea, erudite, dyadic, palpitations, comix, stata, sewn, mixers,
Nearest to square: amyloid, ethologist, dirac, eddic, follows, mumy, kingdon, wills,
Nearest to egypt: leagues, aunt, bikini, spree, iglesias, jericho, countenance, dismantling,
Nearest to smith: garner, thera, geddes, purr, doorstep, obsolescent, alvin, hips,
Epoch 1/10 Iteration: 5100 Avg. Training loss: 4.6441 0.3082 sec/batch
Epoch 1/10 Iteration: 5200 Avg. Training loss: 4.7217 0.2930 sec/batch
Epoch 1/10 Iteration: 5300 Avg. Training loss: 4.6341 0.2844 sec/batch
Epoch 1/10 Iteration: 5400 Avg. Training loss: 4.6766 0.2995 sec/batch
Epoch 1/10 Iteration: 5500 Avg. Training loss: 4.6863 0.3248 sec/batch
Epoch 1/10 Iteration: 5600 Avg. Training loss: 4.6684 0.3046 sec/batch
Epoch 1/10 Iteration: 5700 Avg. Training loss: 4.6862 0.3204 sec/batch
Epoch 1/10 Iteration: 5800 Avg. Training loss: 4.6309 0.2883 sec/batch
Epoch 1/10 Iteration: 5900 Avg. Training loss: 4.6653 0.3064 sec/batch
Epoch 1/10 Iteration: 6000 Avg. Training loss: 4.6210 0.3041 sec/batch
Nearest to into: wholeheartedly, crossword, reboot, mariana, lanka, saxony, judge, voter,
Nearest to during: kidman, charybdis, bonnie, toys, invades, pedestrian, revolutionary, masterful,
Nearest to about: shortening, gorbachev, hippie, hafez, scientifiques, thacker, cyclopedia, fahlman,
Nearest to five: spoofed, dazzling, biogeography, jutish, gulags, tons, classis, utf,
Nearest to six: b, isbn, satirist, actor, enmity, racine, voyage, est,
Nearest to some: unambiguous, graveolens, taoist, timbres, dialogs, retort, spirit, applicants,
Nearest to however: hometown, cheyenne, illustrative, queens, sofa, domination, ttir, presented,
Nearest to war: homerun, hindenburg, gyula, dramas, peek, indonesia, pistols, armature,
Nearest to experience: twa, nominee, hellene, incoherent, moles, observations, relevance, abbess,
Nearest to magazine: macmullen, screech, unthinkable, carolingian, seine, cariou, lps, subtitled,
Nearest to joseph: retrofitted, scarecrow, higginbotham, gdf, provenance, shop, votive, caithness,
Nearest to universe: weighted, gallen, habsburgs, hermeticism, bragg, financed, mecca, schiavo,
Nearest to defense: iea, erudite, dyadic, palpitations, comix, mooted, stata, sewn,
Nearest to square: amyloid, ethologist, dirac, eddic, follows, wills, mumy, svend,
Nearest to egypt: leagues, aunt, jericho, iglesias, bikini, spree, yen, labor,
Nearest to smith: garner, thera, geddes, purr, doorstep, alvin, obsolescent, felipe,
Epoch 1/10 Iteration: 6100 Avg. Training loss: 4.6569 0.3052 sec/batch
Epoch 1/10 Iteration: 6200 Avg. Training loss: 4.6439 0.2995 sec/batch
Epoch 1/10 Iteration: 6300 Avg. Training loss: 4.6222 0.2848 sec/batch
Epoch 1/10 Iteration: 6400 Avg. Training loss: 4.6524 0.2890 sec/batch
Epoch 1/10 Iteration: 6500 Avg. Training loss: 4.6289 0.2931 sec/batch
Epoch 1/10 Iteration: 6600 Avg. Training loss: 4.6025 0.3028 sec/batch
Epoch 1/10 Iteration: 6700 Avg. Training loss: 4.5827 0.2881 sec/batch
Epoch 1/10 Iteration: 6800 Avg. Training loss: 4.6264 0.2902 sec/batch
Epoch 1/10 Iteration: 6900 Avg. Training loss: 4.6468 0.3047 sec/batch
Epoch 1/10 Iteration: 7000 Avg. Training loss: 4.6538 0.2941 sec/batch
Nearest to into: wholeheartedly, crossword, mariana, lanka, reboot, saxony, core, administer,
Nearest to during: kidman, charybdis, revolutionary, bonnie, toys, imprisoned, pedestrian, warfare,
Nearest to about: shortening, gorbachev, hafez, hippie, scientifiques, fahlman, btu, thacker,
Nearest to five: tons, spoofed, dazzling, gulags, jutish, classis, patristics, haemophilus,
Nearest to six: enmity, elijah, est, isbn, racine, prospered, satirist, b,
Nearest to some: unambiguous, graveolens, taoist, timbres, dialogs, retort, spirit, dealings,
Nearest to however: hometown, cheyenne, illustrative, ttir, queens, basically, universality, domination,
Nearest to war: homerun, hindenburg, gyula, dramas, indonesia, gone, peek, pistols,
Nearest to experience: hellene, twa, incoherent, nominee, moles, observations, relevance, abbess,
Nearest to magazine: macmullen, screech, unthinkable, carolingian, cariou, seine, weighting, lps,
Nearest to joseph: retrofitted, provenance, scarecrow, higginbotham, votive, caithness, shop, gdf,
Nearest to universe: weighted, gallen, habsburgs, hermeticism, skid, christiern, bragg, financed,
Nearest to defense: iea, erudite, vladivostok, dyadic, tanna, stata, comix, palpitations,
Nearest to square: amyloid, ethologist, dirac, follows, eddic, wills, mumy, rim,
Nearest to egypt: leagues, aunt, iglesias, jericho, spree, bikini, andrews, yen,
Nearest to smith: garner, thera, purr, geddes, alvin, doorstep, obsolescent, felipe,
Epoch 1/10 Iteration: 7100 Avg. Training loss: 4.6666 0.2992 sec/batch
Epoch 1/10 Iteration: 7200 Avg. Training loss: 4.6398 0.3001 sec/batch
Epoch 1/10 Iteration: 7300 Avg. Training loss: 4.5640 0.3597 sec/batch
Epoch 1/10 Iteration: 7400 Avg. Training loss: 4.6315 0.3580 sec/batch
Epoch 1/10 Iteration: 7500 Avg. Training loss: 4.6311 0.3057 sec/batch
Epoch 1/10 Iteration: 7600 Avg. Training loss: 4.6112 0.3076 sec/batch
Epoch 1/10 Iteration: 7700 Avg. Training loss: 4.6512 0.3105 sec/batch
Epoch 1/10 Iteration: 7800 Avg. Training loss: 4.6074 0.2978 sec/batch
Epoch 1/10 Iteration: 7900 Avg. Training loss: 4.6186 0.3224 sec/batch
Epoch 1/10 Iteration: 8000 Avg. Training loss: 4.6265 0.3187 sec/batch
Nearest to into: wholeheartedly, lanka, saxony, mariana, reboot, crossword, core, ariel,
Nearest to during: kidman, charybdis, revolutionary, toys, warfare, bonnie, decades, imprisoned,
Nearest to about: shortening, hippie, gorbachev, hafez, scientifiques, thacker, durations, fahlman,
Nearest to five: spoofed, tons, jutish, classis, dazzling, patristics, gulags, haemophilus,
Nearest to six: enmity, elijah, racine, prospered, isbn, landon, usd, sparteca,
Nearest to some: unambiguous, graveolens, taoist, timbres, dialogs, spirit, retort, terminates,
Nearest to however: hometown, cheyenne, illustrative, january, queens, basically, universality, ttir,
Nearest to war: homerun, hindenburg, gyula, indonesia, raiders, erupted, informed, peek,
Nearest to experience: hellene, twa, incoherent, abbess, nominee, moles, observations, relevance,
Nearest to magazine: macmullen, screech, unthinkable, carolingian, lps, cariou, seine, weighting,
Nearest to joseph: retrofitted, higginbotham, provenance, scarecrow, votive, shop, gdf, caithness,
Nearest to universe: weighted, skid, hermeticism, christiern, gallen, civilisation, inventor, habsburgs,
Nearest to defense: iea, erudite, tanna, bgn, comix, dyadic, undergraduate, vladivostok,
Nearest to square: amyloid, ethologist, eddic, dirac, follows, capitulated, mumy, wills,
Nearest to egypt: jericho, leagues, iglesias, spree, colonised, aunt, bikini, tema,
Nearest to smith: garner, thera, purr, geddes, doorstep, obsolescent, alvin, felipe,
Epoch 1/10 Iteration: 8100 Avg. Training loss: 4.5765 0.3019 sec/batch
Epoch 1/10 Iteration: 8200 Avg. Training loss: 4.5201 0.3127 sec/batch
Epoch 1/10 Iteration: 8300 Avg. Training loss: 4.5469 0.2871 sec/batch
Epoch 1/10 Iteration: 8400 Avg. Training loss: 4.5609 0.2863 sec/batch
Epoch 1/10 Iteration: 8500 Avg. Training loss: 4.5734 0.2854 sec/batch
Epoch 1/10 Iteration: 8600 Avg. Training loss: 4.5565 0.2798 sec/batch
Epoch 1/10 Iteration: 8700 Avg. Training loss: 4.6259 0.2856 sec/batch
Epoch 1/10 Iteration: 8800 Avg. Training loss: 4.4333 0.2838 sec/batch
Epoch 1/10 Iteration: 8900 Avg. Training loss: 4.4337 0.2923 sec/batch
Epoch 1/10 Iteration: 9000 Avg. Training loss: 4.4212 0.2850 sec/batch
Nearest to into: wholeheartedly, lanka, mariana, crossword, ariel, saxony, reboot, lila,
Nearest to during: kidman, charybdis, revolutionary, bonnie, warfare, invades, imprisoned, toys,
Nearest to about: shortening, gorbachev, hafez, thacker, hippie, fahlman, speedily, scientifiques,
Nearest to five: spoofed, tons, classis, jutish, dazzling, one, patristics, biogeography,
Nearest to six: b, becomes, enmity, landon, racine, sprinter, elijah, satirist,
Nearest to some: unambiguous, graveolens, taoist, dialogs, timbres, tournament, retort, settles,
Nearest to however: hometown, cheyenne, january, illustrative, sofa, queens, presented, domination,
Nearest to war: homerun, hindenburg, erupted, campaign, gyula, raiders, saxophonist, arunachal,
Nearest to experience: hellene, twa, incoherent, abbess, nominee, moles, observations, crates,
Nearest to magazine: macmullen, screech, unthinkable, carolingian, lps, cariou, weighting, seine,
Nearest to joseph: retrofitted, higginbotham, provenance, votive, scarecrow, caithness, injures, naturae,
Nearest to universe: weighted, skid, christiern, disconnected, hermeticism, gallen, civilisation, habsburgs,
Nearest to defense: iea, erudite, bgn, tanna, attrition, euhemerus, undergraduate, comix,
Nearest to square: ethologist, amyloid, follows, dirac, eddic, basins, capitulated, mumy,
Nearest to egypt: jericho, leagues, iglesias, spree, aunt, colonised, bikini, solomon,
Nearest to smith: garner, purr, geddes, thera, alvin, biologist, felipe, doorstep,
Epoch 1/10 Iteration: 9100 Avg. Training loss: 4.5138 0.2852 sec/batch
Epoch 1/10 Iteration: 9200 Avg. Training loss: 4.5803 0.2832 sec/batch
Epoch 1/10 Iteration: 9300 Avg. Training loss: 4.5893 0.2755 sec/batch
Epoch 1/10 Iteration: 9400 Avg. Training loss: 4.5721 0.2783 sec/batch
Epoch 1/10 Iteration: 9500 Avg. Training loss: 4.6491 0.2831 sec/batch
Epoch 1/10 Iteration: 9600 Avg. Training loss: 4.6399 0.2840 sec/batch
Epoch 1/10 Iteration: 9700 Avg. Training loss: 4.6074 0.2876 sec/batch
Epoch 1/10 Iteration: 9800 Avg. Training loss: 4.5525 0.2833 sec/batch
Epoch 1/10 Iteration: 9900 Avg. Training loss: 4.4127 0.2811 sec/batch
Epoch 1/10 Iteration: 10000 Avg. Training loss: 4.5667 0.2779 sec/batch
Nearest to into: wholeheartedly, lanka, crossword, saxony, mariana, core, lila, ariel,
Nearest to during: kidman, charybdis, revolutionary, warfare, imprisoned, bonnie, invades, toys,
Nearest to about: shortening, gorbachev, thacker, hafez, hippie, scientifiques, btu, speedily,
Nearest to five: spoofed, tons, dazzling, patristics, gulags, classis, utf, satyagraha,
Nearest to six: racine, sprinter, enmity, landon, sparteca, est, inversional, cale,
Nearest to some: dialogs, unambiguous, graveolens, taoist, timbres, statistics, tournament, settles,
Nearest to however: hometown, january, domination, cheyenne, illustrative, sofa, universality, quite,
Nearest to war: homerun, erupted, campaign, arunachal, peek, hindenburg, raiders, armature,
Nearest to experience: hellene, twa, incoherent, observations, moles, integrase, nominee, abbess,
Nearest to magazine: macmullen, screech, unthinkable, carolingian, cariou, lps, zeppelin, accessories,
Nearest to joseph: retrofitted, provenance, votive, scholarships, higginbotham, scarecrow, caithness, gdf,
Nearest to universe: skid, disconnected, christiern, weighted, civilisation, inventor, gallen, hermeticism,
Nearest to defense: iea, erudite, bgn, tanna, comix, euhemerus, undergraduate, paulist,
Nearest to square: ethologist, amyloid, dirac, follows, eddic, basins, rim, svend,
Nearest to egypt: jericho, spree, iglesias, leagues, bikini, aunt, colonised, lutterworth,
Nearest to smith: garner, purr, thera, geddes, doorstep, alvin, biologist, felipe,
Epoch 1/10 Iteration: 10100 Avg. Training loss: 4.5708 0.2808 sec/batch
Epoch 1/10 Iteration: 10200 Avg. Training loss: 4.6104 0.2747 sec/batch
Epoch 1/10 Iteration: 10300 Avg. Training loss: 4.5635 0.2835 sec/batch
Epoch 1/10 Iteration: 10400 Avg. Training loss: 4.6029 0.2822 sec/batch
Epoch 1/10 Iteration: 10500 Avg. Training loss: 4.5461 0.2751 sec/batch
Epoch 1/10 Iteration: 10600 Avg. Training loss: 4.4558 0.2814 sec/batch
Epoch 1/10 Iteration: 10700 Avg. Training loss: 4.3648 0.2815 sec/batch
Epoch 1/10 Iteration: 10800 Avg. Training loss: 4.5329 0.2837 sec/batch
Epoch 1/10 Iteration: 10900 Avg. Training loss: 4.5099 0.2808 sec/batch
Epoch 1/10 Iteration: 11000 Avg. Training loss: 4.5041 0.2827 sec/batch
Nearest to into: wholeheartedly, crossword, core, knobs, lanka, saxony, reboot, back,
Nearest to during: charybdis, kidman, revolutionary, warfare, tvs, pedestrian, diffuses, bonnie,
Nearest to about: shortening, gorbachev, hafez, thacker, hippie, scientifiques, btu, unavailable,
Nearest to five: spoofed, tons, classis, dazzling, gulags, one, utf, patristics,
Nearest to six: racine, sprinter, landon, becomes, elijah, rocha, b, pople,
Nearest to some: dialogs, taoist, unambiguous, graveolens, timbres, tournament, settles, statistics,
Nearest to however: hometown, illustrative, quite, presented, universality, sofa, ttir, heimskringla,
Nearest to war: homerun, erupted, arunachal, campaign, ii, gyula, wars, raiders,
Nearest to experience: hellene, observations, integrase, twa, incoherent, abbess, nexgen, indication,
Nearest to magazine: macmullen, screech, unthinkable, accessories, cariou, lps, zeppelin, carolingian,
Nearest to joseph: retrofitted, provenance, higginbotham, scarecrow, votive, scholarships, naturae, caithness,
Nearest to universe: skid, disconnected, weighted, civilisation, christiern, gallen, mecca, philosophy,
Nearest to defense: iea, erudite, tanna, bgn, godse, stata, comix, guard,
Nearest to square: ethologist, amyloid, follows, dirac, rim, eddic, mumy, svend,
Nearest to egypt: jericho, iglesias, aunt, bce, spree, colonised, bikini, solomon,
Nearest to smith: garner, purr, thera, geddes, alvin, doorstep, biologist, mcm,
Epoch 1/10 Iteration: 11100 Avg. Training loss: 4.5553 0.2881 sec/batch
Epoch 1/10 Iteration: 11200 Avg. Training loss: 4.5285 0.2815 sec/batch
Epoch 1/10 Iteration: 11300 Avg. Training loss: 4.5001 0.2865 sec/batch
Epoch 1/10 Iteration: 11400 Avg. Training loss: 4.5146 0.2853 sec/batch
Epoch 1/10 Iteration: 11500 Avg. Training loss: 4.5610 0.2816 sec/batch
Epoch 1/10 Iteration: 11600 Avg. Training loss: 4.5864 0.2837 sec/batch
Epoch 2/10 Iteration: 11700 Avg. Training loss: 4.5520 0.1124 sec/batch
Epoch 2/10 Iteration: 11800 Avg. Training loss: 4.5153 0.2796 sec/batch
Epoch 2/10 Iteration: 11900 Avg. Training loss: 4.4746 0.2864 sec/batch
Epoch 2/10 Iteration: 12000 Avg. Training loss: 4.5223 0.2834 sec/batch
Nearest to into: crossword, wholeheartedly, core, lanka, back, reboot, knobs, saxony,
Nearest to during: charybdis, kidman, warfare, revolutionary, carangi, diffuses, pedestrian, tvs,
Nearest to about: shortening, hippie, thacker, hafez, gorbachev, btu, scientifiques, unavailable,
Nearest to five: tons, spoofed, classis, gulags, satyagraha, gibb, patristics, utf,
Nearest to six: racine, sprinter, ironclad, isbn, coluim, sparteca, pople, enmity,
Nearest to some: dialogs, taoist, unambiguous, graveolens, timbres, settles, tournament, statistics,
Nearest to however: hometown, quite, ttir, basically, sofa, illustrative, frosty, heimskringla,
Nearest to war: homerun, wars, campaign, erupted, arunachal, hindenburg, peek, gyula,
Nearest to experience: hellene, incoherent, observations, twa, integrase, abbess, moles, nexgen,
Nearest to magazine: macmullen, screech, unthinkable, accessories, cariou, lps, henze, carolingian,
Nearest to joseph: retrofitted, higginbotham, votive, scarecrow, provenance, scholarships, gdf, naturae,
Nearest to universe: disconnected, skid, weighted, christiern, civilisation, philosophy, gallen, inventor,
Nearest to defense: iea, guard, erudite, bgn, comix, tanna, godse, stata,
Nearest to square: ethologist, amyloid, follows, rim, basins, eddic, dirac, mumy,
Nearest to egypt: jericho, bce, iglesias, aunt, colonised, spree, admixtures, lutterworth,
Nearest to smith: garner, purr, thera, geddes, doorstep, alvin, biologist, obsolescent,
Epoch 2/10 Iteration: 12100 Avg. Training loss: 4.4548 0.2835 sec/batch
Epoch 2/10 Iteration: 12200 Avg. Training loss: 4.4850 0.2929 sec/batch
Epoch 2/10 Iteration: 12300 Avg. Training loss: 4.4818 0.2810 sec/batch
Epoch 2/10 Iteration: 12400 Avg. Training loss: 4.4161 0.2836 sec/batch
Epoch 2/10 Iteration: 12500 Avg. Training loss: 4.4610 0.2796 sec/batch
Epoch 2/10 Iteration: 12600 Avg. Training loss: 4.4809 0.2799 sec/batch
Epoch 2/10 Iteration: 12700 Avg. Training loss: 4.4681 0.2714 sec/batch
Epoch 2/10 Iteration: 12800 Avg. Training loss: 4.4831 0.2721 sec/batch
Epoch 2/10 Iteration: 12900 Avg. Training loss: 4.4906 0.2689 sec/batch
Epoch 2/10 Iteration: 13000 Avg. Training loss: 4.4730 0.2734 sec/batch
Nearest to into: wholeheartedly, core, knobs, back, crossword, reboot, saxony, lanka,
Nearest to during: charybdis, kidman, warfare, revolutionary, carangi, tvs, diffuses, invades,
Nearest to about: shortening, hafez, thacker, hippie, btu, unavailable, pertain, fahlman,
Nearest to five: tons, classis, spoofed, gibb, gulags, utf, patristics, satyagraha,
Nearest to six: coluim, ironclad, enmity, pople, sprinter, rocha, ammonite, becomes,
Nearest to some: dialogs, taoist, unambiguous, settles, graveolens, timbres, toothpaste, statistics,
Nearest to however: hometown, quite, illustrative, universality, ttir, basically, perfecti, january,
Nearest to war: homerun, campaign, erupted, wars, raiders, peek, arunachal, gyula,
Nearest to experience: hellene, incoherent, nexgen, observations, integrase, abbess, moles, crates,
Nearest to magazine: macmullen, screech, accessories, unthinkable, cariou, lps, graven, carolingian,
Nearest to joseph: retrofitted, higginbotham, votive, provenance, scarecrow, scholarships, gdf, stu,
Nearest to universe: disconnected, weighted, skid, civilisation, christiern, gallen, nabokov, plutarch,
Nearest to defense: iea, bgn, erudite, guard, godse, tanna, stata, ignition,
Nearest to square: ethologist, amyloid, rim, follows, basins, dirac, mumy, eddic,
Nearest to egypt: bce, jericho, iglesias, aunt, bikini, colonised, spree, dynasty,
Nearest to smith: garner, purr, thera, alvin, geddes, doorstep, biologist, mcm,
Epoch 2/10 Iteration: 13100 Avg. Training loss: 4.4895 0.2650 sec/batch
Epoch 2/10 Iteration: 13200 Avg. Training loss: 4.4914 0.2728 sec/batch
Epoch 2/10 Iteration: 13300 Avg. Training loss: 4.4819 0.2762 sec/batch
Epoch 2/10 Iteration: 13400 Avg. Training loss: 4.3532 0.2652 sec/batch
Epoch 2/10 Iteration: 13500 Avg. Training loss: 4.5026 0.2711 sec/batch
Epoch 2/10 Iteration: 13600 Avg. Training loss: 4.5367 0.2727 sec/batch
Epoch 2/10 Iteration: 13700 Avg. Training loss: 4.4772 0.2689 sec/batch
Epoch 2/10 Iteration: 13800 Avg. Training loss: 4.4963 0.2715 sec/batch
Epoch 2/10 Iteration: 13900 Avg. Training loss: 4.4825 0.2691 sec/batch
Epoch 2/10 Iteration: 14000 Avg. Training loss: 4.4878 0.2688 sec/batch
Nearest to into: wholeheartedly, core, ariel, saxony, back, reboot, crossword, lanka,
Nearest to during: charybdis, kidman, revolutionary, tvs, warfare, antiproton, carangi, antiderivatives,
Nearest to about: shortening, thacker, hippie, hafez, btu, lalor, gorbachev, pertain,
Nearest to five: tons, classis, spoofed, satyagraha, patristics, gulags, utf, gibb,
Nearest to six: sprinter, ironclad, coluim, enmity, sparteca, racine, pople, rocha,
Nearest to some: dialogs, taoist, graveolens, tournament, timbres, toothpaste, unambiguous, many,
Nearest to however: hometown, quite, illustrative, perfecti, universality, basically, frosty, january,
Nearest to war: homerun, wars, hindenburg, erupted, gyula, campaign, raiders, arunachal,
Nearest to experience: hellene, incoherent, observations, integrase, nexgen, abbess, isopropanol, maybe,
Nearest to magazine: macmullen, screech, cariou, accessories, lps, graven, unthinkable, impending,
Nearest to joseph: retrofitted, votive, scholarships, higginbotham, provenance, scarecrow, gdf, stu,
Nearest to universe: weighted, disconnected, skid, christiern, gallen, civilisation, jingle, nabokov,
Nearest to defense: iea, guard, bgn, stata, erudite, godse, ignition, rodentia,
Nearest to square: ethologist, amyloid, follows, rim, basins, specimen, knee, mumy,
Nearest to egypt: bce, jericho, iglesias, spree, aunt, colonised, dynasty, petrel,
Nearest to smith: garner, purr, thera, alvin, doorstep, geddes, mcm, biologist,
Epoch 2/10 Iteration: 14100 Avg. Training loss: 4.4971 0.2697 sec/batch
Epoch 2/10 Iteration: 14200 Avg. Training loss: 4.4938 0.2684 sec/batch
Epoch 2/10 Iteration: 14300 Avg. Training loss: 4.4558 0.2653 sec/batch
Epoch 2/10 Iteration: 14400 Avg. Training loss: 4.4394 0.2687 sec/batch
Epoch 2/10 Iteration: 14500 Avg. Training loss: 4.3530 0.2725 sec/batch
Epoch 2/10 Iteration: 14600 Avg. Training loss: 4.3769 0.2701 sec/batch
Epoch 2/10 Iteration: 14700 Avg. Training loss: 4.4701 0.2715 sec/batch
Epoch 2/10 Iteration: 14800 Avg. Training loss: 4.4154 0.2707 sec/batch
Epoch 2/10 Iteration: 14900 Avg. Training loss: 4.4446 0.2692 sec/batch
Epoch 2/10 Iteration: 15000 Avg. Training loss: 4.4365 0.2761 sec/batch
Nearest to into: core, crossword, wholeheartedly, reboot, ariel, knobs, saxony, back,
Nearest to during: charybdis, kidman, revolutionary, antiproton, warfare, lamanites, tvs, bonnie,
Nearest to about: shortening, pertain, thacker, fahlman, durations, gorbachev, btu, hippie,
Nearest to five: tons, classis, spoofed, patristics, gulags, utf, satyagraha, one,
Nearest to six: katsuhiro, racine, sprinter, isbn, ironclad, coluim, rocha, three,
Nearest to some: dialogs, taoist, unambiguous, many, graveolens, timbres, toothpaste, terminates,
Nearest to however: hometown, quite, universality, although, perfecti, illustrative, january, domination,
Nearest to war: wars, campaign, homerun, erupted, hindenburg, nationalists, gyula, locate,
Nearest to experience: hellene, observations, incoherent, integrase, maybe, isopropanol, nexgen, abbess,
Nearest to magazine: macmullen, screech, cariou, lps, accessories, graven, unthinkable, impending,
Nearest to joseph: retrofitted, votive, scholarships, higginbotham, scarecrow, stu, provenance, gdf,
Nearest to universe: disconnected, weighted, skid, christiern, civilisation, gallen, jingle, mash,
Nearest to defense: iea, guard, bgn, godse, stata, erudite, ignition, rodentia,
Nearest to square: ethologist, follows, rim, amyloid, specimen, basins, mumy, dirac,
Nearest to egypt: bce, jericho, iglesias, spree, colonised, aunt, invading, petrel,
Nearest to smith: garner, purr, thera, doorstep, geddes, alvin, biologist, kreis,
Epoch 2/10 Iteration: 15100 Avg. Training loss: 4.4483 0.2704 sec/batch
Epoch 2/10 Iteration: 15200 Avg. Training loss: 4.3944 0.2724 sec/batch
Epoch 2/10 Iteration: 15300 Avg. Training loss: 4.4667 0.2685 sec/batch
Epoch 2/10 Iteration: 15400 Avg. Training loss: 4.4645 0.2674 sec/batch
Epoch 2/10 Iteration: 15500 Avg. Training loss: 4.4827 0.2809 sec/batch
Epoch 2/10 Iteration: 15600 Avg. Training loss: 4.4562 0.2745 sec/batch
Epoch 2/10 Iteration: 15700 Avg. Training loss: 4.4387 0.2669 sec/batch
Epoch 2/10 Iteration: 15800 Avg. Training loss: 4.4822 0.2694 sec/batch
Epoch 2/10 Iteration: 15900 Avg. Training loss: 4.4719 0.2652 sec/batch
Epoch 2/10 Iteration: 16000 Avg. Training loss: 4.3945 0.2674 sec/batch
Nearest to into: core, crossword, knobs, back, reboot, lanka, ariel, diverted,
Nearest to during: charybdis, kidman, antiproton, revolutionary, lamanites, bonnie, carangi, diffuses,
Nearest to about: shortening, pertain, thacker, hafez, hippie, durations, fahlman, btu,
Nearest to five: spoofed, tons, classis, one, utf, gibb, satyagraha, patristics,
Nearest to six: isbn, katsuhiro, three, sprinter, coluim, ironclad, landon, inordinate,
Nearest to some: taoist, dialogs, many, unambiguous, graveolens, toothpaste, climate, timbres,
Nearest to however: hometown, although, universality, quite, january, perfecti, illustrative, but,
Nearest to war: wars, campaign, nationalists, erupted, homerun, communists, hindenburg, sped,
Nearest to experience: hellene, observations, incoherent, maybe, integrase, nexgen, isopropanol, abbess,
Nearest to magazine: macmullen, cariou, accessories, screech, lps, graven, unthinkable, impending,
Nearest to joseph: retrofitted, higginbotham, scholarships, votive, scarecrow, stu, provenance, gdf,
Nearest to universe: disconnected, weighted, christiern, skid, philosophy, civilisation, gallen, plutarch,
Nearest to defense: iea, guard, bgn, erudite, godse, stata, ignition, rodentia,
Nearest to square: ethologist, follows, amyloid, rim, dirac, mumy, specimen, basins,
Nearest to egypt: bce, jericho, colonised, iglesias, aunt, spree, dynasty, invading,
Nearest to smith: garner, purr, alvin, thera, geddes, doorstep, biologist, mcm,
Epoch 2/10 Iteration: 16100 Avg. Training loss: 4.4445 0.2711 sec/batch
Epoch 2/10 Iteration: 16200 Avg. Training loss: 4.4179 0.2708 sec/batch
Epoch 2/10 Iteration: 16300 Avg. Training loss: 4.4546 0.2666 sec/batch
Epoch 2/10 Iteration: 16400 Avg. Training loss: 4.4784 0.2689 sec/batch
Epoch 2/10 Iteration: 16500 Avg. Training loss: 4.4915 0.2782 sec/batch
Epoch 2/10 Iteration: 16600 Avg. Training loss: 4.4782 0.2748 sec/batch
Epoch 2/10 Iteration: 16700 Avg. Training loss: 4.4218 0.2675 sec/batch
Epoch 2/10 Iteration: 16800 Avg. Training loss: 4.3948 0.2693 sec/batch
Epoch 2/10 Iteration: 16900 Avg. Training loss: 4.4069 0.2688 sec/batch
Epoch 2/10 Iteration: 17000 Avg. Training loss: 4.4592 0.2673 sec/batch
Nearest to into: crossword, core, back, knobs, diverted, reboot, mariana, lanka,
Nearest to during: charybdis, kidman, revolutionary, antiproton, lamanites, bonnie, carangi, tvs,
Nearest to about: shortening, pertain, thacker, lalor, hafez, durations, fahlman, hippie,
Nearest to five: spoofed, classis, tons, patristics, utf, one, satyagraha, jutish,
Nearest to six: sprinter, katsuhiro, coluim, ironclad, rocha, isbn, pople, crompton,
Nearest to some: many, taoist, dialogs, unambiguous, these, statistics, toothpaste, graveolens,
Nearest to however: hometown, although, universality, quite, but, illustrative, january, perfecti,
Nearest to war: wars, campaign, nationalists, erupted, hindenburg, world, communists, sped,
Nearest to experience: hellene, observations, incoherent, integrase, maybe, nexgen, abbess, isopropanol,
Nearest to magazine: macmullen, cariou, graven, lps, accessories, impending, screech, donald,
Nearest to joseph: retrofitted, higginbotham, scarecrow, scholarships, stu, votive, gdf, shop,
Nearest to universe: disconnected, civilisation, christiern, skid, weighted, philosophy, gallen, plutarch,
Nearest to defense: iea, guard, bgn, godse, attrition, ignition, command, stata,
Nearest to square: ethologist, amyloid, follows, rim, specimen, dirac, mumy, knee,
Nearest to egypt: jericho, bce, spree, petrel, aunt, colonised, iglesias, dynasty,
Nearest to smith: garner, purr, alvin, thera, geddes, doorstep, biologist, mcm,
Epoch 2/10 Iteration: 17100 Avg. Training loss: 4.4320 0.2670 sec/batch
Epoch 2/10 Iteration: 17200 Avg. Training loss: 4.4212 0.2698 sec/batch
Epoch 2/10 Iteration: 17300 Avg. Training loss: 4.4655 0.2694 sec/batch
Epoch 2/10 Iteration: 17400 Avg. Training loss: 4.4435 0.2682 sec/batch
Epoch 2/10 Iteration: 17500 Avg. Training loss: 4.3879 0.2704 sec/batch
Epoch 2/10 Iteration: 17600 Avg. Training loss: 4.4435 0.2692 sec/batch
Epoch 2/10 Iteration: 17700 Avg. Training loss: 4.3970 0.2715 sec/batch
Epoch 2/10 Iteration: 17800 Avg. Training loss: 4.4633 0.2732 sec/batch
Epoch 2/10 Iteration: 17900 Avg. Training loss: 4.3571 0.2687 sec/batch
Epoch 2/10 Iteration: 18000 Avg. Training loss: 4.4395 0.2684 sec/batch
Nearest to into: back, knobs, crossword, core, diverted, wholeheartedly, reboot, lanka,
Nearest to during: charybdis, kidman, antiproton, revolutionary, lamanites, carangi, bonnie, thirty,
Nearest to about: shortening, pertain, thacker, hippie, erbia, lalor, fabricius, hollingworth,
Nearest to five: one, classis, spoofed, patristics, tons, coligny, utf, gibb,
Nearest to six: sprinter, three, rocha, racine, katsuhiro, stahl, coluim, becomes,
Nearest to some: many, dialogs, these, taoist, banned, unambiguous, graveolens, both,
Nearest to however: hometown, although, but, quite, universality, january, illustrative, perfecti,
Nearest to war: world, wars, campaign, nationalists, erupted, abyssinia, ii, locate,
Nearest to experience: hellene, incoherent, observations, integrase, maybe, abbess, isopropanol, nexgen,
Nearest to magazine: macmullen, graven, cariou, accessories, impending, lps, zeppelin, screech,
Nearest to joseph: retrofitted, higginbotham, stu, scarecrow, scholarships, votive, shop, provenance,
Nearest to universe: christiern, disconnected, civilisation, weighted, philosophy, skid, gallen, plutarch,
Nearest to defense: iea, bgn, guard, erudite, stata, ignition, godse, pitman,
Nearest to square: ethologist, knee, amyloid, specimen, mumy, follows, rim, dirac,
Nearest to egypt: jericho, bce, aunt, spree, petrel, iglesias, colonised, admixtures,
Nearest to smith: garner, purr, thera, alvin, geddes, doorstep, spartacus, mcm,
Epoch 2/10 Iteration: 18100 Avg. Training loss: 4.4785 0.2696 sec/batch
Epoch 2/10 Iteration: 18200 Avg. Training loss: 4.4035 0.2693 sec/batch
Epoch 2/10 Iteration: 18300 Avg. Training loss: 4.3394 0.2685 sec/batch
Epoch 2/10 Iteration: 18400 Avg. Training loss: 4.4588 0.2669 sec/batch
Epoch 2/10 Iteration: 18500 Avg. Training loss: 4.4516 0.2847 sec/batch
Epoch 2/10 Iteration: 18600 Avg. Training loss: 4.4435 0.2720 sec/batch
Epoch 2/10 Iteration: 18700 Avg. Training loss: 4.4410 0.2661 sec/batch
Epoch 2/10 Iteration: 18800 Avg. Training loss: 4.4986 0.2701 sec/batch
Epoch 2/10 Iteration: 18900 Avg. Training loss: 4.4340 0.2797 sec/batch
Epoch 2/10 Iteration: 19000 Avg. Training loss: 4.3849 0.2701 sec/batch
Nearest to into: back, core, diverted, crossword, ariel, straddles, knobs, saxony,
Nearest to during: charybdis, antiproton, kidman, lamanites, revolutionary, warfare, carangi, fascists,
Nearest to about: shortening, pertain, thacker, lalor, hafez, hollingworth, fahlman, speedily,
Nearest to five: tons, classis, one, patristics, satyagraha, coligny, spoofed, gulags,
Nearest to six: sprinter, crompton, three, coluim, rocha, sparteca, inordinate, katsuhiro,
Nearest to some: many, dialogs, taoist, these, settles, graveolens, banned, unambiguous,
Nearest to however: but, although, hometown, quite, january, universality, basically, ttir,
Nearest to war: wars, erupted, campaign, nationalists, world, abyssinia, ii, homerun,
Nearest to experience: hellene, integrase, incoherent, observations, maybe, abbess, isopropanol, crates,
Nearest to magazine: macmullen, cariou, graven, accessories, russcol, zeppelin, lps, screech,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, scholarships, provenance, shop, stu,
Nearest to universe: christiern, disconnected, civilisation, weighted, jingle, philosophy, skid, gallen,
Nearest to defense: iea, guard, bgn, attrition, erudite, comix, stata, ignition,
Nearest to square: ethologist, specimen, mumy, rim, amyloid, knee, follows, basins,
Nearest to egypt: jericho, bce, spree, colonised, aunt, petrel, iglesias, bikini,
Nearest to smith: garner, purr, thera, geddes, alvin, doorstep, biologist, spartacus,
Epoch 2/10 Iteration: 19100 Avg. Training loss: 4.4788 0.2701 sec/batch
Epoch 2/10 Iteration: 19200 Avg. Training loss: 4.4389 0.2691 sec/batch
Epoch 2/10 Iteration: 19300 Avg. Training loss: 4.4133 0.2678 sec/batch
Epoch 2/10 Iteration: 19400 Avg. Training loss: 4.4511 0.2672 sec/batch
Epoch 2/10 Iteration: 19500 Avg. Training loss: 4.4653 0.2681 sec/batch
Epoch 2/10 Iteration: 19600 Avg. Training loss: 4.4566 0.2695 sec/batch
Epoch 2/10 Iteration: 19700 Avg. Training loss: 4.4302 0.2674 sec/batch
Epoch 2/10 Iteration: 19800 Avg. Training loss: 4.3816 0.2669 sec/batch
Epoch 2/10 Iteration: 19900 Avg. Training loss: 4.3770 0.2756 sec/batch
Epoch 2/10 Iteration: 20000 Avg. Training loss: 4.3915 0.2694 sec/batch
Nearest to into: back, core, crossword, diverted, knobs, lanka, straddles, wholeheartedly,
Nearest to during: charybdis, antiproton, kidman, lamanites, carangi, warfare, revolutionary, beheaded,
Nearest to about: shortening, pertain, thacker, hafez, scientifiques, lalor, fahlman, hollingworth,
Nearest to five: one, classis, tons, patristics, dazzling, spoofed, jutish, utf,
Nearest to six: three, sprinter, rocha, coluim, sparteca, ironclad, inordinate, enmity,
Nearest to some: many, dialogs, these, taoist, other, both, toothpaste, most,
Nearest to however: although, but, quite, hometown, basically, january, sativa, perfecti,
Nearest to war: wars, erupted, nationalists, campaign, world, homerun, abyssinia, knot,
Nearest to experience: hellene, integrase, observations, incoherent, crates, isopropanol, maybe, abbess,
Nearest to magazine: macmullen, cariou, graven, lps, accessories, impending, screech, kashima,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, scholarships, stu, shop, provenance,
Nearest to universe: christiern, civilisation, disconnected, philosophy, fictional, weighted, skid, gallen,
Nearest to defense: iea, bgn, guard, ignition, rodentia, attrition, stata, erudite,
Nearest to square: ethologist, mumy, knee, specimen, follows, amyloid, rim, basins,
Nearest to egypt: jericho, bce, spree, colonised, aunt, petrel, invading, dynasty,
Nearest to smith: garner, purr, thera, geddes, alvin, doorstep, biologist, huck,
Epoch 2/10 Iteration: 20100 Avg. Training loss: 4.3998 0.2680 sec/batch
Epoch 2/10 Iteration: 20200 Avg. Training loss: 4.4122 0.2665 sec/batch
Epoch 2/10 Iteration: 20300 Avg. Training loss: 4.3775 0.2712 sec/batch
Epoch 2/10 Iteration: 20400 Avg. Training loss: 4.4408 0.2698 sec/batch
Epoch 2/10 Iteration: 20500 Avg. Training loss: 4.1438 0.2691 sec/batch
Epoch 2/10 Iteration: 20600 Avg. Training loss: 4.3570 0.2752 sec/batch
Epoch 2/10 Iteration: 20700 Avg. Training loss: 4.2932 0.2744 sec/batch
Epoch 2/10 Iteration: 20800 Avg. Training loss: 4.4151 0.2689 sec/batch
Epoch 2/10 Iteration: 20900 Avg. Training loss: 4.4312 0.2699 sec/batch
Epoch 2/10 Iteration: 21000 Avg. Training loss: 4.4409 0.2666 sec/batch
Nearest to into: back, crossword, diverted, core, ariel, lanka, knobs, reboot,
Nearest to during: charybdis, kidman, carangi, antiproton, lamanites, revolutionary, synching, fascists,
Nearest to about: shortening, pertain, thacker, hippie, fabricius, lalor, discography, hollingworth,
Nearest to five: classis, spoofed, one, patristics, dazzling, eminem, coligny, steinbeck,
Nearest to six: rocha, racine, sprinter, ironclad, becomes, enmity, farr, katsuhiro,
Nearest to some: many, taoist, these, dialogs, most, graveolens, toothpaste, settles,
Nearest to however: although, but, hometown, quite, though, january, universality, heimskringla,
Nearest to war: erupted, campaign, wars, ii, knot, world, nationalists, gone,
Nearest to experience: hellene, observations, integrase, incoherent, crates, maybe, twa, detectives,
Nearest to magazine: macmullen, cariou, graven, lps, impending, screech, accessories, kashima,
Nearest to joseph: retrofitted, higginbotham, votive, scarecrow, provenance, shop, scholarships, caithness,
Nearest to universe: christiern, disconnected, civilisation, skid, philosophy, fictional, gallen, weighted,
Nearest to defense: iea, guard, bgn, rodentia, comix, erudite, attrition, ignition,
Nearest to square: ethologist, mumy, knee, specimen, rim, follows, amyloid, basins,
Nearest to egypt: jericho, bce, spree, petrel, colonised, aunt, solomon, heydrich,
Nearest to smith: garner, purr, thera, geddes, alvin, mcm, doorstep, huck,
Epoch 2/10 Iteration: 21100 Avg. Training loss: 4.4594 0.2702 sec/batch
Epoch 2/10 Iteration: 21200 Avg. Training loss: 4.5007 0.2665 sec/batch
Epoch 2/10 Iteration: 21300 Avg. Training loss: 4.4947 0.2694 sec/batch
Epoch 2/10 Iteration: 21400 Avg. Training loss: 4.4466 0.2669 sec/batch
Epoch 2/10 Iteration: 21500 Avg. Training loss: 4.3742 0.2661 sec/batch
Epoch 2/10 Iteration: 21600 Avg. Training loss: 4.3568 0.2678 sec/batch
Epoch 2/10 Iteration: 21700 Avg. Training loss: 4.4319 0.2657 sec/batch
Epoch 2/10 Iteration: 21800 Avg. Training loss: 4.3960 0.2709 sec/batch
Epoch 2/10 Iteration: 21900 Avg. Training loss: 4.4315 0.2678 sec/batch
Epoch 2/10 Iteration: 22000 Avg. Training loss: 4.4369 0.2702 sec/batch
Nearest to into: back, knobs, core, crossword, diverted, ariel, wholeheartedly, straddles,
Nearest to during: charybdis, kidman, antiproton, carangi, lamanites, tvs, bonnie, fascists,
Nearest to about: shortening, pertain, thacker, hippie, lalor, discography, hollingworth, fahlman,
Nearest to five: one, classis, tons, spoofed, eminem, average, patristics, dazzling,
Nearest to six: katsuhiro, sprinter, rocha, three, racine, ironclad, equipage, farr,
Nearest to some: many, dialogs, these, taoist, other, toothpaste, graveolens, most,
Nearest to however: although, but, quite, hometown, universality, perfecti, reaffirm, illustrative,
Nearest to war: erupted, wars, campaign, nationalists, abyssinia, ii, homerun, knot,
Nearest to experience: hellene, observations, incoherent, integrase, varnish, maybe, twa, abbess,
Nearest to magazine: macmullen, cariou, graven, lps, screech, impending, accessories, russcol,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, provenance, scholarships, stu, shop,
Nearest to universe: christiern, disconnected, civilisation, fictional, skid, mwh, mash, jingle,
Nearest to defense: guard, iea, rodentia, bgn, comix, stata, ignition, erudite,
Nearest to square: ethologist, mumy, specimen, knee, amyloid, rim, follows, svend,
Nearest to egypt: jericho, bce, spree, petrel, colonised, invading, iglesias, aunt,
Nearest to smith: garner, purr, thera, geddes, alvin, mcm, huck, spartacus,
Epoch 2/10 Iteration: 22100 Avg. Training loss: 4.4233 0.2681 sec/batch
Epoch 2/10 Iteration: 22200 Avg. Training loss: 4.4285 0.2737 sec/batch
Epoch 2/10 Iteration: 22300 Avg. Training loss: 4.2491 0.2769 sec/batch
Epoch 2/10 Iteration: 22400 Avg. Training loss: 4.2751 0.2681 sec/batch
Epoch 2/10 Iteration: 22500 Avg. Training loss: 4.4483 0.2650 sec/batch
Epoch 2/10 Iteration: 22600 Avg. Training loss: 4.2984 0.2713 sec/batch
Epoch 2/10 Iteration: 22700 Avg. Training loss: 4.4214 0.2650 sec/batch
Epoch 2/10 Iteration: 22800 Avg. Training loss: 4.4216 0.2692 sec/batch
Epoch 2/10 Iteration: 22900 Avg. Training loss: 4.4123 0.2728 sec/batch
Epoch 2/10 Iteration: 23000 Avg. Training loss: 4.3722 0.2695 sec/batch
Nearest to into: back, knobs, core, crossword, diverted, ariel, mbps, lanka,
Nearest to during: charybdis, antiproton, kidman, carangi, lamanites, tvs, angora, diffuses,
Nearest to about: shortening, pertain, hippie, thacker, lalor, discography, fahlman, scientifiques,
Nearest to five: one, classis, tons, spoofed, utf, eminem, satyagraha, gibb,
Nearest to six: rocha, three, katsuhiro, ironclad, pople, sprinter, stahl, onboard,
Nearest to some: many, these, dialogs, other, taoist, both, most, several,
Nearest to however: although, but, quite, universality, though, hometown, perfecti, basically,
Nearest to war: ii, wars, erupted, abyssinia, knot, nationalists, campaign, homerun,
Nearest to experience: hellene, observations, incoherent, integrase, varnish, maybe, abbess, isopropanol,
Nearest to magazine: macmullen, cariou, lps, graven, accessories, impending, screech, russcol,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, stu, provenance, caithness, scholarships,
Nearest to universe: christiern, fictional, disconnected, civilisation, astronomical, skid, weighted, jingle,
Nearest to defense: guard, iea, bgn, stata, comix, ignition, rodentia, erudite,
Nearest to square: ethologist, knee, mumy, specimen, rim, differentiability, amyloid, follows,
Nearest to egypt: jericho, bce, spree, petrel, countenance, colonised, invading, iglesias,
Nearest to smith: garner, purr, geddes, thera, alvin, mcm, spartacus, huck,
Epoch 2/10 Iteration: 23100 Avg. Training loss: 4.4310 0.2685 sec/batch
Epoch 2/10 Iteration: 23200 Avg. Training loss: 4.4110 0.2646 sec/batch
Epoch 2/10 Iteration: 23300 Avg. Training loss: 4.4853 0.2676 sec/batch
Epoch 3/10 Iteration: 23400 Avg. Training loss: 4.4287 0.2121 sec/batch
Epoch 3/10 Iteration: 23500 Avg. Training loss: 4.3659 0.2708 sec/batch
Epoch 3/10 Iteration: 23600 Avg. Training loss: 4.4086 0.2708 sec/batch
Epoch 3/10 Iteration: 23700 Avg. Training loss: 4.3819 0.2675 sec/batch
Epoch 3/10 Iteration: 23800 Avg. Training loss: 4.3793 0.2694 sec/batch
Epoch 3/10 Iteration: 23900 Avg. Training loss: 4.3905 0.2708 sec/batch
Epoch 3/10 Iteration: 24000 Avg. Training loss: 4.3303 0.2729 sec/batch
Nearest to into: back, ariel, core, knobs, diverted, crossword, straddles, lanka,
Nearest to during: charybdis, antiproton, carangi, kidman, lamanites, ago, diffuses, beheaded,
Nearest to about: shortening, pertain, lalor, thacker, hafez, hippie, scientifiques, fabricius,
Nearest to five: one, classis, seven, tons, jan, spoofed, cir, zero,
Nearest to six: three, rocha, sprinter, ironclad, katsuhiro, one, pople, elmwood,
Nearest to some: many, these, dialogs, taoist, other, both, most, several,
Nearest to however: although, but, quite, hometown, though, universality, because, perfecti,
Nearest to war: wars, knot, abyssinia, nationalists, erupted, ii, homerun, sailed,
Nearest to experience: hellene, observations, incoherent, integrase, maybe, abbess, twa, varnish,
Nearest to magazine: macmullen, cariou, graven, lps, accessories, impending, russcol, screech,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, stu, sarah, dynes, caithness,
Nearest to universe: fictional, christiern, disconnected, civilisation, gallen, skid, philosophy, mash,
Nearest to defense: guard, iea, comix, stata, ignition, rodentia, cardigan, bgn,
Nearest to square: ethologist, specimen, knee, mumy, rim, differentiability, amyloid, follows,
Nearest to egypt: jericho, bce, spree, petrel, countenance, dynasty, colonised, admixtures,
Nearest to smith: garner, purr, geddes, alvin, thera, mcm, huck, spartacus,
Epoch 3/10 Iteration: 24100 Avg. Training loss: 4.3211 0.2705 sec/batch
Epoch 3/10 Iteration: 24200 Avg. Training loss: 4.3972 0.2672 sec/batch
Epoch 3/10 Iteration: 24300 Avg. Training loss: 4.3798 0.2691 sec/batch
Epoch 3/10 Iteration: 24400 Avg. Training loss: 4.3809 0.2714 sec/batch
Epoch 3/10 Iteration: 24500 Avg. Training loss: 4.3887 0.2677 sec/batch
Epoch 3/10 Iteration: 24600 Avg. Training loss: 4.4172 0.2688 sec/batch
Epoch 3/10 Iteration: 24700 Avg. Training loss: 4.4091 0.2708 sec/batch
Epoch 3/10 Iteration: 24800 Avg. Training loss: 4.3911 0.2673 sec/batch
Epoch 3/10 Iteration: 24900 Avg. Training loss: 4.3980 0.2718 sec/batch
Epoch 3/10 Iteration: 25000 Avg. Training loss: 4.2676 0.2712 sec/batch
Nearest to into: back, core, diverted, ariel, crossword, knobs, straddles, lanka,
Nearest to during: charybdis, antiproton, carangi, lamanites, kidman, tvs, prior, beheaded,
Nearest to about: shortening, thacker, lalor, pertain, fahlman, hippie, discography, zero,
Nearest to five: one, zero, tons, classis, seven, jutish, average, four,
Nearest to six: three, ironclad, equipage, rocha, pople, katsuhiro, sparteca, takers,
Nearest to some: many, these, other, both, dialogs, most, several, taoist,
Nearest to however: although, but, quite, though, hometown, universality, perfecti, because,
Nearest to war: wars, erupted, nationalists, abyssinia, knot, homerun, world, sailed,
Nearest to experience: hellene, maybe, observations, incoherent, integrase, abbess, isopropanol, crates,
Nearest to magazine: macmullen, graven, cariou, lps, accessories, impending, russcol, screech,
Nearest to joseph: retrofitted, higginbotham, votive, scarecrow, caithness, stu, provenance, dynes,
Nearest to universe: fictional, christiern, disconnected, civilisation, gallen, philosophy, skid, prot,
Nearest to defense: guard, iea, ignition, bgn, stata, attrition, rodentia, comix,
Nearest to square: ethologist, specimen, rim, mumy, knee, differentiability, amyloid, follows,
Nearest to egypt: jericho, bce, petrel, spree, countenance, invading, colonised, bikini,
Nearest to smith: garner, purr, thera, geddes, alvin, mcm, huck, pup,
Epoch 3/10 Iteration: 25100 Avg. Training loss: 4.4049 0.2676 sec/batch
Epoch 3/10 Iteration: 25200 Avg. Training loss: 4.3900 0.2720 sec/batch
Epoch 3/10 Iteration: 25300 Avg. Training loss: 4.4335 0.2718 sec/batch
Epoch 3/10 Iteration: 25400 Avg. Training loss: 4.3949 0.2659 sec/batch
Epoch 3/10 Iteration: 25500 Avg. Training loss: 4.3991 0.2661 sec/batch
Epoch 3/10 Iteration: 25600 Avg. Training loss: 4.3937 0.2663 sec/batch
Epoch 3/10 Iteration: 25700 Avg. Training loss: 4.4246 0.2767 sec/batch
Epoch 3/10 Iteration: 25800 Avg. Training loss: 4.4094 0.2693 sec/batch
Epoch 3/10 Iteration: 25900 Avg. Training loss: 4.4282 0.2643 sec/batch
Epoch 3/10 Iteration: 26000 Avg. Training loss: 4.3530 0.2659 sec/batch
Nearest to into: back, knobs, ariel, core, crossword, diverted, straddles, mbps,
Nearest to during: charybdis, antiproton, carangi, lamanites, kidman, beheaded, fascists, tvs,
Nearest to about: shortening, thacker, pertain, lalor, hippie, discography, fabricius, fahlman,
Nearest to five: one, classis, tons, spoofed, zero, jutish, cir, seven,
Nearest to six: three, rocha, katsuhiro, ironclad, equipage, coluim, elijah, racine,
Nearest to some: many, these, other, most, several, both, dialogs, taoist,
Nearest to however: although, but, quite, though, basically, perfecti, sativa, hometown,
Nearest to war: wars, nationalists, wwii, world, wwi, ii, erupted, knot,
Nearest to experience: hellene, maybe, observations, incoherent, integrase, isopropanol, varnish, abbess,
Nearest to magazine: macmullen, graven, cariou, lps, impending, screech, accessories, russcol,
Nearest to joseph: retrofitted, higginbotham, votive, scarecrow, stu, shop, caithness, provenance,
Nearest to universe: disconnected, christiern, fictional, civilisation, gallen, skid, philosophy, mwh,
Nearest to defense: guard, iea, stata, ignition, bgn, comix, erudite, attrition,
Nearest to square: ethologist, specimen, knee, mumy, rim, differentiability, follows, kilmister,
Nearest to egypt: jericho, petrel, spree, bce, invading, countenance, colonised, receded,
Nearest to smith: garner, purr, geddes, thera, mcm, alvin, huck, pup,
Epoch 3/10 Iteration: 26100 Avg. Training loss: 4.3287 0.2664 sec/batch
Epoch 3/10 Iteration: 26200 Avg. Training loss: 4.2116 0.2686 sec/batch
Epoch 3/10 Iteration: 26300 Avg. Training loss: 4.3661 0.2660 sec/batch
Epoch 3/10 Iteration: 26400 Avg. Training loss: 4.3592 0.2638 sec/batch
Epoch 3/10 Iteration: 26500 Avg. Training loss: 4.3793 0.2664 sec/batch
Epoch 3/10 Iteration: 26600 Avg. Training loss: 4.3446 0.2674 sec/batch
Epoch 3/10 Iteration: 26700 Avg. Training loss: 4.3468 0.2752 sec/batch
Epoch 3/10 Iteration: 26800 Avg. Training loss: 4.2674 0.2660 sec/batch
Epoch 3/10 Iteration: 26900 Avg. Training loss: 4.3758 0.2688 sec/batch
Epoch 3/10 Iteration: 27000 Avg. Training loss: 4.3999 0.2685 sec/batch
Nearest to into: back, knobs, ariel, crossword, diverted, core, straddles, hazzan,
Nearest to during: antiproton, charybdis, carangi, lamanites, tvs, kidman, first, fascists,
Nearest to about: shortening, pertain, thacker, lalor, fabricius, discography, erbia, hippie,
Nearest to five: one, zero, seven, classis, tons, spoofed, four, two,
Nearest to six: three, katsuhiro, equipage, rocha, takers, nine, cycles, coluim,
Nearest to some: many, these, most, dialogs, several, other, taoist, critics,
Nearest to however: although, but, quite, though, hometown, basically, sativa, universality,
Nearest to war: wars, wwii, ii, nationalists, wwi, knot, world, campaign,
Nearest to experience: hellene, maybe, observations, incoherent, isopropanol, integrase, varnish, abbess,
Nearest to magazine: macmullen, graven, cariou, lps, impending, russcol, screech, donald,
Nearest to joseph: retrofitted, higginbotham, votive, scarecrow, stu, shop, gdf, caithness,
Nearest to universe: disconnected, christiern, fictional, civilisation, philosophy, mwh, skid, gallen,
Nearest to defense: guard, iea, bgn, comix, stata, attrition, ignition, erudite,
Nearest to square: specimen, ethologist, knee, mumy, differentiability, rim, follows, kilmister,
Nearest to egypt: jericho, bce, spree, petrel, invading, colonised, receded, countenance,
Nearest to smith: garner, purr, geddes, thera, mcm, alvin, pup, huck,
Epoch 3/10 Iteration: 27100 Avg. Training loss: 4.3912 0.2682 sec/batch
Epoch 3/10 Iteration: 27200 Avg. Training loss: 4.3737 0.2672 sec/batch
Epoch 3/10 Iteration: 27300 Avg. Training loss: 4.3741 0.2657 sec/batch
Epoch 3/10 Iteration: 27400 Avg. Training loss: 4.3644 0.2650 sec/batch
Epoch 3/10 Iteration: 27500 Avg. Training loss: 4.4209 0.2688 sec/batch
Epoch 3/10 Iteration: 27600 Avg. Training loss: 4.3360 0.2689 sec/batch
Epoch 3/10 Iteration: 27700 Avg. Training loss: 4.3766 0.2676 sec/batch
Epoch 3/10 Iteration: 27800 Avg. Training loss: 4.3480 0.2686 sec/batch
Epoch 3/10 Iteration: 27900 Avg. Training loss: 4.3796 0.2657 sec/batch
Epoch 3/10 Iteration: 28000 Avg. Training loss: 4.3847 0.2689 sec/batch
Nearest to into: back, knobs, crossword, mbps, ariel, core, diverted, hazzan,
Nearest to during: charybdis, antiproton, carangi, lamanites, kidman, beheaded, fascists, following,
Nearest to about: shortening, pertain, lalor, thacker, fabricius, regarding, erbia, discography,
Nearest to five: one, zero, spoofed, cir, classis, seven, utf, tons,
Nearest to six: three, rocha, katsuhiro, takers, crompton, one, nine, coluim,
Nearest to some: many, these, several, most, dialogs, other, taoist, critics,
Nearest to however: but, although, quite, though, hometown, basically, sativa, universality,
Nearest to war: wars, abyssinia, nationalists, wwii, world, ii, knot, sailed,
Nearest to experience: hellene, maybe, incoherent, observations, integrase, moles, isopropanol, abbess,
Nearest to magazine: macmullen, graven, cariou, lps, donald, impending, russcol, accessories,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, stu, shop, gdf, dynes,
Nearest to universe: disconnected, christiern, fictional, gallen, civilisation, philosophy, mash, mwh,
Nearest to defense: guard, iea, bgn, attrition, comix, erudite, stata, pitman,
Nearest to square: differentiability, specimen, ethologist, knee, mumy, matrices, rim, follows,
Nearest to egypt: jericho, bce, petrel, spree, invading, countenance, colonised, receded,
Nearest to smith: purr, garner, geddes, mcm, alvin, thera, pup, roberts,
Epoch 3/10 Iteration: 28100 Avg. Training loss: 4.4221 0.2674 sec/batch
Epoch 3/10 Iteration: 28200 Avg. Training loss: 4.3999 0.2669 sec/batch
Epoch 3/10 Iteration: 28300 Avg. Training loss: 4.3820 0.2679 sec/batch
Epoch 3/10 Iteration: 28400 Avg. Training loss: 4.2819 0.2653 sec/batch
Epoch 3/10 Iteration: 28500 Avg. Training loss: 4.3869 0.2720 sec/batch
Epoch 3/10 Iteration: 28600 Avg. Training loss: 4.3121 0.2706 sec/batch
Epoch 3/10 Iteration: 28700 Avg. Training loss: 4.3983 0.2700 sec/batch
Epoch 3/10 Iteration: 28800 Avg. Training loss: 4.3990 0.2700 sec/batch
Epoch 3/10 Iteration: 28900 Avg. Training loss: 4.3513 0.2735 sec/batch
Epoch 3/10 Iteration: 29000 Avg. Training loss: 4.4103 0.2658 sec/batch
Nearest to into: back, knobs, straddles, diverted, core, mbps, crossword, destry,
Nearest to during: charybdis, antiproton, lamanites, carangi, kidman, beheaded, first, following,
Nearest to about: shortening, pertain, thacker, discography, erbia, fabricius, regarding, lalor,
Nearest to five: one, zero, classis, patristics, seven, spoofed, cir, jutish,
Nearest to six: three, katsuhiro, rocha, sprinter, crompton, coluim, nine, takers,
Nearest to some: many, these, several, most, dialogs, other, taoist, critics,
Nearest to however: but, although, though, quite, universality, sativa, when, hometown,
Nearest to war: wars, ii, wwii, world, abyssinia, knot, campaign, nationalists,
Nearest to experience: hellene, maybe, incoherent, observations, integrase, isopropanol, varnish, abbess,
Nearest to magazine: macmullen, graven, cariou, lps, russcol, donald, screech, impending,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, shop, stu, dynes, caithness,
Nearest to universe: disconnected, christiern, fictional, gallen, nigerian, civilisation, mash, scipione,
Nearest to defense: guard, iea, bgn, attrition, erudite, stata, comix, pitman,
Nearest to square: differentiability, mumy, ethologist, specimen, matrices, knee, rim, follows,
Nearest to egypt: jericho, bce, petrel, spree, invading, countenance, colonised, instytut,
Nearest to smith: garner, thera, purr, geddes, mcm, alvin, pup, roberts,
Epoch 3/10 Iteration: 29100 Avg. Training loss: 4.3062 0.2771 sec/batch
Epoch 3/10 Iteration: 29200 Avg. Training loss: 4.3816 0.2666 sec/batch
Epoch 3/10 Iteration: 29300 Avg. Training loss: 4.3288 0.2702 sec/batch
Epoch 3/10 Iteration: 29400 Avg. Training loss: 4.3646 0.2691 sec/batch
Epoch 3/10 Iteration: 29500 Avg. Training loss: 4.3800 0.2681 sec/batch
Epoch 3/10 Iteration: 29600 Avg. Training loss: 4.2961 0.2706 sec/batch
Epoch 3/10 Iteration: 29700 Avg. Training loss: 4.3972 0.2671 sec/batch
Epoch 3/10 Iteration: 29800 Avg. Training loss: 4.3690 0.2723 sec/batch
Epoch 3/10 Iteration: 29900 Avg. Training loss: 4.3591 0.2649 sec/batch
Epoch 3/10 Iteration: 30000 Avg. Training loss: 4.2616 0.2665 sec/batch
Nearest to into: back, straddles, knobs, diverted, mariana, core, mbps, ariel,
Nearest to during: antiproton, charybdis, following, carangi, lamanites, kidman, first, followed,
Nearest to about: shortening, pertain, thacker, regarding, rugians, discography, hollingworth, erbia,
Nearest to five: one, zero, seven, four, two, classis, tons, patristics,
Nearest to six: three, one, nine, katsuhiro, zero, equipage, five, eight,
Nearest to some: many, these, several, most, other, dialogs, both, such,
Nearest to however: but, although, though, quite, because, hometown, january, when,
Nearest to war: ii, wwii, world, abyssinia, wars, outbreak, wwi, nationalists,
Nearest to experience: hellene, maybe, incoherent, integrase, isopropanol, observations, nexgen, varnish,
Nearest to magazine: macmullen, graven, cariou, russcol, lps, donald, impending, zeppelin,
Nearest to joseph: retrofitted, higginbotham, shop, votive, scarecrow, stu, dynes, provenance,
Nearest to universe: christiern, disconnected, fictional, civilisation, gallen, mash, nigerian, lvares,
Nearest to defense: guard, iea, bgn, erudite, comix, attrition, stata, pitman,
Nearest to square: differentiability, specimen, knee, ethologist, mumy, matrices, rim, kilmister,
Nearest to egypt: jericho, bce, petrel, spree, aunt, colonised, invading, andrews,
Nearest to smith: garner, thera, geddes, purr, mcm, alvin, roberts, spartacus,
Epoch 3/10 Iteration: 30100 Avg. Training loss: 4.3924 0.2711 sec/batch
Epoch 3/10 Iteration: 30200 Avg. Training loss: 4.3965 0.2688 sec/batch
Epoch 3/10 Iteration: 30300 Avg. Training loss: 4.3768 0.2687 sec/batch
Epoch 3/10 Iteration: 30400 Avg. Training loss: 4.4062 0.2678 sec/batch
Epoch 3/10 Iteration: 30500 Avg. Training loss: 4.4165 0.2673 sec/batch
Epoch 3/10 Iteration: 30600 Avg. Training loss: 4.3024 0.2672 sec/batch
Epoch 3/10 Iteration: 30700 Avg. Training loss: 4.3894 0.2700 sec/batch
Epoch 3/10 Iteration: 30800 Avg. Training loss: 4.3971 0.2666 sec/batch
Epoch 3/10 Iteration: 30900 Avg. Training loss: 4.3959 0.2660 sec/batch
Epoch 3/10 Iteration: 31000 Avg. Training loss: 4.3564 0.2703 sec/batch
Nearest to into: back, straddles, diverted, knobs, from, ariel, core, destry,
Nearest to during: antiproton, carangi, following, lamanites, charybdis, beheaded, followed, before,
Nearest to about: shortening, pertain, regarding, albanians, thacker, discography, hollingworth, fabricius,
Nearest to five: one, zero, seven, patristics, four, classis, cir, six,
Nearest to six: three, one, katsuhiro, five, rocha, nine, cycles, crompton,
Nearest to some: many, these, several, other, most, dialogs, taoist, such,
Nearest to however: but, although, though, quite, when, because, hometown, sativa,
Nearest to war: ii, abyssinia, wars, world, wwii, outbreak, campaign, nationalists,
Nearest to experience: hellene, maybe, incoherent, integrase, isopropanol, observations, abbess, nexgen,
Nearest to magazine: macmullen, graven, cariou, russcol, lps, donald, impending, zeppelin,
Nearest to joseph: retrofitted, higginbotham, votive, shop, scarecrow, dynes, provenance, sarah,
Nearest to universe: christiern, disconnected, fictional, gallen, lvares, civilisation, scipione, nigerian,
Nearest to defense: guard, iea, bgn, attrition, comix, pitman, erudite, isothermal,
Nearest to square: specimen, differentiability, mumy, ethologist, knee, matrices, rim, follows,
Nearest to egypt: jericho, petrel, bce, spree, invading, colonised, andrews, dynasty,
Nearest to smith: garner, thera, mcm, geddes, purr, alvin, pup, roberts,
Epoch 3/10 Iteration: 31100 Avg. Training loss: 4.3984 0.2650 sec/batch
Epoch 3/10 Iteration: 31200 Avg. Training loss: 4.3658 0.2714 sec/batch
Epoch 3/10 Iteration: 31300 Avg. Training loss: 4.3930 0.2687 sec/batch
Epoch 3/10 Iteration: 31400 Avg. Training loss: 4.3576 0.2671 sec/batch
Epoch 3/10 Iteration: 31500 Avg. Training loss: 4.2771 0.2679 sec/batch
Epoch 3/10 Iteration: 31600 Avg. Training loss: 4.3391 0.2666 sec/batch
Epoch 3/10 Iteration: 31700 Avg. Training loss: 4.3488 0.2688 sec/batch
Epoch 3/10 Iteration: 31800 Avg. Training loss: 4.3600 0.2665 sec/batch
Epoch 3/10 Iteration: 31900 Avg. Training loss: 4.3144 0.2729 sec/batch
Epoch 3/10 Iteration: 32000 Avg. Training loss: 4.4163 0.2710 sec/batch
Nearest to into: back, straddles, knobs, diverted, mbps, core, destry, hazzan,
Nearest to during: antiproton, carangi, charybdis, lamanites, kidman, beheaded, before, following,
Nearest to about: shortening, regarding, pertain, thacker, albanians, yassin, rugians, discography,
Nearest to five: one, zero, seven, four, two, six, utf, classis,
Nearest to six: three, one, eight, five, nine, katsuhiro, rocha, chiller,
Nearest to some: many, these, several, other, such, most, both, taoist,
Nearest to however: although, but, though, quite, because, sativa, when, hometown,
Nearest to war: wars, abyssinia, wwii, ii, campaign, world, outbreak, sailed,
Nearest to experience: maybe, hellene, observations, incoherent, integrase, varnish, nexgen, subdivide,
Nearest to magazine: macmullen, graven, cariou, russcol, impending, donald, lps, screech,
Nearest to joseph: retrofitted, higginbotham, scarecrow, dynes, votive, shop, provenance, sarah,
Nearest to universe: christiern, disconnected, fictional, gallen, mash, lvares, civilisation, nigerian,
Nearest to defense: guard, iea, attrition, bgn, pitman, erudite, isothermal, ignition,
Nearest to square: specimen, mumy, knee, ethologist, differentiability, rim, matrices, follows,
Nearest to egypt: jericho, bce, petrel, spree, colonised, invading, egyptian, aunt,
Nearest to smith: garner, mcm, thera, geddes, purr, alvin, pup, roberts,
Epoch 3/10 Iteration: 32100 Avg. Training loss: 4.2673 0.2723 sec/batch
Epoch 3/10 Iteration: 32200 Avg. Training loss: 4.1218 0.2701 sec/batch
Epoch 3/10 Iteration: 32300 Avg. Training loss: 4.2334 0.2691 sec/batch
Epoch 3/10 Iteration: 32400 Avg. Training loss: 4.3366 0.2692 sec/batch
Epoch 3/10 Iteration: 32500 Avg. Training loss: 4.3507 0.2808 sec/batch
Epoch 3/10 Iteration: 32600 Avg. Training loss: 4.4049 0.2689 sec/batch
Epoch 3/10 Iteration: 32700 Avg. Training loss: 4.3707 0.2657 sec/batch
Epoch 3/10 Iteration: 32800 Avg. Training loss: 4.4333 0.2663 sec/batch
Epoch 3/10 Iteration: 32900 Avg. Training loss: 4.4415 0.2681 sec/batch
Epoch 3/10 Iteration: 33000 Avg. Training loss: 4.4390 0.2703 sec/batch
Nearest to into: back, straddles, diverted, knobs, ariel, mbps, destry, moruroa,
Nearest to during: antiproton, carangi, following, kidman, lamanites, charybdis, first, barak,
Nearest to about: shortening, discography, pertain, regarding, lalor, thacker, albanians, hippie,
Nearest to five: one, zero, satyagraha, seven, utf, classis, average, four,
Nearest to six: three, racine, crompton, equipage, katsuhiro, rocha, chiller, elijah,
Nearest to some: many, these, several, most, other, such, taoist, both,
Nearest to however: although, but, though, quite, because, sativa, when, hometown,
Nearest to war: ii, veterans, campaign, world, outbreak, abyssinia, knot, wwii,
Nearest to experience: hellene, maybe, incoherent, observations, varnish, beni, integrase, abbess,
Nearest to magazine: macmullen, cariou, graven, lps, russcol, impending, coil, donald,
Nearest to joseph: retrofitted, higginbotham, votive, provenance, caithness, scarecrow, dynes, shop,
Nearest to universe: disconnected, christiern, gallen, mash, fictional, lvares, constellations, mwh,
Nearest to defense: guard, attrition, erudite, comix, iea, bgn, pitman, rodentia,
Nearest to square: specimen, mumy, ethologist, knee, differentiability, kilmister, burton, rim,
Nearest to egypt: jericho, petrel, bce, spree, colonised, invading, countenance, andrews,
Nearest to smith: garner, mcm, thera, geddes, purr, alvin, pup, electing,
Epoch 3/10 Iteration: 33100 Avg. Training loss: 4.3787 0.2707 sec/batch
Epoch 3/10 Iteration: 33200 Avg. Training loss: 4.2065 0.2701 sec/batch
Epoch 3/10 Iteration: 33300 Avg. Training loss: 4.3840 0.2680 sec/batch
Epoch 3/10 Iteration: 33400 Avg. Training loss: 4.3688 0.2762 sec/batch
Epoch 3/10 Iteration: 33500 Avg. Training loss: 4.3777 0.2684 sec/batch
Epoch 3/10 Iteration: 33600 Avg. Training loss: 4.3413 0.2719 sec/batch
Epoch 3/10 Iteration: 33700 Avg. Training loss: 4.4462 0.2729 sec/batch
Epoch 3/10 Iteration: 33800 Avg. Training loss: 4.3306 0.2703 sec/batch
Epoch 3/10 Iteration: 33900 Avg. Training loss: 4.3339 0.2683 sec/batch
Epoch 3/10 Iteration: 34000 Avg. Training loss: 4.1267 0.2702 sec/batch
Nearest to into: back, straddles, knobs, ariel, diverted, mbps, mariana, destry,
Nearest to during: antiproton, following, carangi, first, war, lamanites, charybdis, followed,
Nearest to about: zero, shortening, thacker, pertain, regarding, depleted, lalor, yassin,
Nearest to five: one, zero, two, four, seven, six, three, and,
Nearest to six: three, one, five, nine, zero, two, eight, four,
Nearest to some: many, these, several, most, other, both, such, various,
Nearest to however: but, although, though, quite, because, that, since, sativa,
Nearest to war: ii, outbreak, nationalists, campaign, wwii, abyssinia, wars, world,
Nearest to experience: maybe, hellene, observations, varnish, integrase, incoherent, isopropanol, glycogen,
Nearest to magazine: macmullen, graven, cariou, russcol, lps, monitors, coil, impending,
Nearest to joseph: retrofitted, higginbotham, scarecrow, caithness, votive, provenance, dynes, shop,
Nearest to universe: christiern, disconnected, gallen, fictional, lvares, mash, mwh, constellations,
Nearest to defense: guard, attrition, erudite, iea, comix, bgn, ignition, isothermal,
Nearest to square: specimen, knee, ethologist, mumy, differentiability, rim, burton, dock,
Nearest to egypt: jericho, petrel, bce, spree, invading, colonised, admixtures, countenance,
Nearest to smith: mcm, garner, thera, purr, geddes, alvin, pup, huck,
Epoch 3/10 Iteration: 34100 Avg. Training loss: 4.3816 0.2703 sec/batch
Epoch 3/10 Iteration: 34200 Avg. Training loss: 4.3596 0.2676 sec/batch
Epoch 3/10 Iteration: 34300 Avg. Training loss: 4.2594 0.2709 sec/batch
Epoch 3/10 Iteration: 34400 Avg. Training loss: 4.3684 0.2696 sec/batch
Epoch 3/10 Iteration: 34500 Avg. Training loss: 4.3590 0.2701 sec/batch
Epoch 3/10 Iteration: 34600 Avg. Training loss: 4.3426 0.2673 sec/batch
Epoch 3/10 Iteration: 34700 Avg. Training loss: 4.3008 0.2695 sec/batch
Epoch 3/10 Iteration: 34800 Avg. Training loss: 4.3893 0.2699 sec/batch
Epoch 3/10 Iteration: 34900 Avg. Training loss: 4.4012 0.2690 sec/batch
Epoch 4/10 Iteration: 35000 Avg. Training loss: 4.4016 0.0547 sec/batch
Nearest to into: back, straddles, knobs, ariel, destry, mbps, moruroa, hazzan,
Nearest to during: antiproton, carangi, lamanites, charybdis, before, first, kidman, following,
Nearest to about: shortening, pertain, regarding, thacker, hippie, discography, depleted, fabricius,
Nearest to five: one, zero, four, two, seven, classis, six, utf,
Nearest to six: three, eight, rocha, onboard, five, one, enmity, nine,
Nearest to some: many, these, several, other, such, various, both, most,
Nearest to however: although, but, though, quite, because, that, sativa, since,
Nearest to war: ii, outbreak, campaign, abyssinia, knot, veterans, nationalists, wwii,
Nearest to experience: maybe, hellene, incoherent, observations, varnish, integrase, isopropanol, beni,
Nearest to magazine: macmullen, cariou, graven, lps, coil, russcol, impending, fetus,
Nearest to joseph: retrofitted, higginbotham, caithness, scarecrow, dynes, votive, moglen, provenance,
Nearest to universe: fictional, disconnected, gallen, christiern, mash, mwh, constellations, monkey,
Nearest to defense: guard, attrition, comix, erudite, iea, bgn, ignition, command,
Nearest to square: specimen, mumy, knee, burton, ethologist, differentiability, rim, dock,
Nearest to egypt: jericho, petrel, bce, spree, egyptian, invading, countenance, admixtures,
Nearest to smith: mcm, garner, purr, geddes, thera, alvin, pup, huck,
Epoch 4/10 Iteration: 35100 Avg. Training loss: 4.3791 0.2742 sec/batch
Epoch 4/10 Iteration: 35200 Avg. Training loss: 4.2965 0.2775 sec/batch
Epoch 4/10 Iteration: 35300 Avg. Training loss: 4.3933 0.2710 sec/batch
Epoch 4/10 Iteration: 35400 Avg. Training loss: 4.3116 0.2703 sec/batch
Epoch 4/10 Iteration: 35500 Avg. Training loss: 4.3200 0.2713 sec/batch
Epoch 4/10 Iteration: 35600 Avg. Training loss: 4.3664 0.2726 sec/batch
Epoch 4/10 Iteration: 35700 Avg. Training loss: 4.2852 0.2690 sec/batch
Epoch 4/10 Iteration: 35800 Avg. Training loss: 4.3041 0.2653 sec/batch
Epoch 4/10 Iteration: 35900 Avg. Training loss: 4.3651 0.2806 sec/batch
Epoch 4/10 Iteration: 36000 Avg. Training loss: 4.3735 0.2685 sec/batch
Nearest to into: back, knobs, hazzan, straddles, destry, mbps, diverted, ariel,
Nearest to during: antiproton, carangi, first, charybdis, before, lamanites, following, barak,
Nearest to about: shortening, regarding, pertain, fabricius, discography, thacker, depleted, approximately,
Nearest to five: one, zero, four, two, seven, six, and, classis,
Nearest to six: three, one, five, eight, nine, rocha, onboard, enmity,
Nearest to some: many, these, several, other, most, various, both, such,
Nearest to however: although, but, though, because, quite, that, when, since,
Nearest to war: ii, nationalists, outbreak, campaign, knot, veterans, wwii, abyssinia,
Nearest to experience: maybe, hellene, incoherent, integrase, beni, observations, varnish, isopropanol,
Nearest to magazine: macmullen, graven, cariou, russcol, lps, coil, handguns, slogans,
Nearest to joseph: retrofitted, higginbotham, scarecrow, caithness, votive, sarah, stu, shop,
Nearest to universe: fictional, gallen, christiern, disconnected, mash, mwh, monkey, lvares,
Nearest to defense: guard, comix, ignition, attrition, iea, rodentia, pitman, bgn,
Nearest to square: specimen, knee, ethologist, rim, mumy, differentiability, dock, burton,
Nearest to egypt: jericho, petrel, bce, spree, egyptian, countenance, invading, circa,
Nearest to smith: garner, mcm, thera, geddes, purr, alvin, pup, axioms,
Epoch 4/10 Iteration: 36100 Avg. Training loss: 4.3109 0.2692 sec/batch
Epoch 4/10 Iteration: 36200 Avg. Training loss: 4.3399 0.2747 sec/batch
Epoch 4/10 Iteration: 36300 Avg. Training loss: 4.3626 0.2689 sec/batch
Epoch 4/10 Iteration: 36400 Avg. Training loss: 4.3716 0.2680 sec/batch
Epoch 4/10 Iteration: 36500 Avg. Training loss: 4.3482 0.2723 sec/batch
Epoch 4/10 Iteration: 36600 Avg. Training loss: 4.3484 0.2673 sec/batch
Epoch 4/10 Iteration: 36700 Avg. Training loss: 4.1943 0.2707 sec/batch
Epoch 4/10 Iteration: 36800 Avg. Training loss: 4.3741 0.2668 sec/batch
Epoch 4/10 Iteration: 36900 Avg. Training loss: 4.3873 0.2687 sec/batch
Epoch 4/10 Iteration: 37000 Avg. Training loss: 4.3703 0.2699 sec/batch
Nearest to into: back, straddles, mbps, knobs, ariel, destry, through, suffocating,
Nearest to during: antiproton, carangi, first, before, charybdis, lamanites, beginning, kidman,
Nearest to about: shortening, regarding, thacker, pertain, discography, wired, deg, depleted,
Nearest to five: one, zero, four, seven, two, six, and, eight,
Nearest to six: three, five, one, eight, nine, four, rocha, racine,
Nearest to some: many, these, several, other, various, most, such, both,
Nearest to however: but, although, though, because, quite, that, since, sativa,
Nearest to war: ii, veterans, world, wwi, knot, wars, wwii, outbreak,
Nearest to experience: maybe, hellene, incoherent, integrase, varnish, isopropanol, beni, observations,
Nearest to magazine: macmullen, graven, cariou, lps, russcol, impending, coil, handguns,
Nearest to joseph: retrofitted, higginbotham, votive, shop, sarah, stu, moglen, scarecrow,
Nearest to universe: gallen, fictional, disconnected, mwh, mash, christiern, constellations, lvares,
Nearest to defense: guard, attrition, iea, ignition, pitman, bgn, rodentia, comix,
Nearest to square: specimen, knee, ethologist, rim, mumy, differentiability, burton, dock,
Nearest to egypt: jericho, petrel, egyptian, bce, spree, invading, andrews, countenance,
Nearest to smith: garner, mcm, thera, purr, geddes, alvin, pup, capitalism,
Epoch 4/10 Iteration: 37100 Avg. Training loss: 4.3673 0.2695 sec/batch
Epoch 4/10 Iteration: 37200 Avg. Training loss: 4.3701 0.2676 sec/batch
Epoch 4/10 Iteration: 37300 Avg. Training loss: 4.3576 0.2686 sec/batch
Epoch 4/10 Iteration: 37400 Avg. Training loss: 4.3932 0.2707 sec/batch
Epoch 4/10 Iteration: 37500 Avg. Training loss: 4.3773 0.2700 sec/batch
Epoch 4/10 Iteration: 37600 Avg. Training loss: 4.3332 0.2717 sec/batch
Epoch 4/10 Iteration: 37700 Avg. Training loss: 4.3197 0.2716 sec/batch
Epoch 4/10 Iteration: 37800 Avg. Training loss: 4.2214 0.2690 sec/batch
Epoch 4/10 Iteration: 37900 Avg. Training loss: 4.2181 0.2740 sec/batch
Epoch 4/10 Iteration: 38000 Avg. Training loss: 4.3224 0.2693 sec/batch
Nearest to into: back, straddles, mbps, through, mariana, ariel, diverted, knobs,
Nearest to during: antiproton, carangi, before, first, following, beginning, late, fascists,
Nearest to about: shortening, regarding, thacker, depleted, rugians, pertain, albanians, discography,
Nearest to five: one, zero, four, two, six, seven, three, eight,
Nearest to six: three, five, one, nine, eight, two, zero, four,
Nearest to some: many, these, several, other, various, most, such, both,
Nearest to however: but, although, though, because, quite, that, since, sativa,
Nearest to war: ii, wwii, campaign, nationalists, wars, knot, wwi, outbreak,
Nearest to experience: maybe, hellene, integrase, isopropanol, incoherent, observations, subdivide, varnish,
Nearest to magazine: macmullen, graven, cariou, lps, impending, coil, handguns, russcol,
Nearest to joseph: retrofitted, higginbotham, votive, scarecrow, shop, caithness, sarah, stu,
Nearest to universe: gallen, disconnected, mwh, fictional, christiern, mash, lvares, constellations,
Nearest to defense: guard, attrition, iea, pitman, ignition, bgn, comix, rodentia,
Nearest to square: specimen, knee, rim, ethologist, dock, mumy, westport, differentiability,
Nearest to egypt: jericho, petrel, invading, egyptian, colonised, countenance, bce, circa,
Nearest to smith: garner, mcm, geddes, thera, purr, alvin, pup, axioms,
Epoch 4/10 Iteration: 38100 Avg. Training loss: 4.3216 0.2733 sec/batch
Epoch 4/10 Iteration: 38200 Avg. Training loss: 4.2923 0.2712 sec/batch
Epoch 4/10 Iteration: 38300 Avg. Training loss: 4.3557 0.2690 sec/batch
Epoch 4/10 Iteration: 38400 Avg. Training loss: 4.3275 0.2717 sec/batch
Epoch 4/10 Iteration: 38500 Avg. Training loss: 4.2199 0.2742 sec/batch
Epoch 4/10 Iteration: 38600 Avg. Training loss: 4.3674 0.2693 sec/batch
Epoch 4/10 Iteration: 38700 Avg. Training loss: 4.3548 0.2708 sec/batch
Epoch 4/10 Iteration: 38800 Avg. Training loss: 4.3446 0.2698 sec/batch
Epoch 4/10 Iteration: 38900 Avg. Training loss: 4.3652 0.2658 sec/batch
Epoch 4/10 Iteration: 39000 Avg. Training loss: 4.3327 0.2685 sec/batch
Nearest to into: back, mbps, ariel, hazzan, knobs, through, straddles, destry,
Nearest to during: antiproton, first, before, carangi, lamanites, late, beginning, following,
Nearest to about: shortening, regarding, pertain, thacker, discography, gssps, rugians, fabricius,
Nearest to five: one, zero, four, two, six, seven, eight, three,
Nearest to six: three, one, nine, five, eight, katsuhiro, four, two,
Nearest to some: many, these, several, other, various, such, most, both,
Nearest to however: but, although, though, because, quite, that, since, when,
Nearest to war: ii, wwii, wars, nationalists, veterans, campaign, outbreak, wwi,
Nearest to experience: maybe, hellene, integrase, incoherent, isopropanol, observations, varnish, gabrieli,
Nearest to magazine: macmullen, graven, cariou, lps, coil, handguns, fetus, russcol,
Nearest to joseph: retrofitted, higginbotham, scarecrow, votive, shop, sarah, stu, caithness,
Nearest to universe: disconnected, gallen, mwh, lvares, christiern, constellations, mash, fictional,
Nearest to defense: guard, comix, attrition, pitman, bgn, ignition, erudite, iea,
Nearest to square: knee, specimen, differentiability, rim, ethologist, burton, mumy, westport,
Nearest to egypt: jericho, petrel, invading, egyptian, spree, receded, andrews, countenance,
Nearest to smith: garner, mcm, geddes, thera, purr, alvin, pup, huck,
Epoch 4/10 Iteration: 39100 Avg. Training loss: 4.3674 0.2678 sec/batch
Epoch 4/10 Iteration: 39200 Avg. Training loss: 4.3842 0.2689 sec/batch
Epoch 4/10 Iteration: 39300 Avg. Training loss: 4.2880 0.2794 sec/batch
Epoch 4/10 Iteration: 39400 Avg. Training loss: 4.3483 0.2694 sec/batch
Epoch 4/10 Iteration: 39500 Avg. Training loss: 4.2940 0.2682 sec/batch
Epoch 4/10 Iteration: 39600 Avg. Training loss: 4.3323 0.2693 sec/batch
Epoch 4/10 Iteration: 39700 Avg. Training loss: 4.3629 0.2724 sec/batch
Epoch 4/10 Iteration: 39800 Avg. Training loss: 4.3768 0.2702 sec/batch
Epoch 4/10 Iteration: 39900 Avg. Training loss: 4.3682 0.2695 sec/batch
Epoch 4/10 Iteration: 40000 Avg. Training loss: 4.3017 0.2768 sec/batch
Nearest to into: back, straddles, mbps, destry, mariana, ariel, knobs, distancing,
Nearest to during: antiproton, carangi, before, first, lamanites, following, late, early,
Nearest to about: shortening, pertain, regarding, thacker, discography, wired, rugians, approximately,
Nearest to five: one, zero, four, six, two, seven, cir, utf,
Nearest to six: three, one, nine, five, katsuhiro, eight, rocha, crompton,
Nearest to some: many, several, these, other, various, such, most, both,
Nearest to however: but, although, though, because, quite, that, since, sativa,
Nearest to war: ii, wwii, campaign, veterans, knot, wwi, outbreak, wars,
Nearest to experience: maybe, integrase, hellene, observations, incoherent, moles, gabrieli, subdivide,
Nearest to magazine: macmullen, graven, cariou, lps, coil, impending, zeppelin, published,
Nearest to joseph: retrofitted, higginbotham, scarecrow, shop, moglen, sarah, caithness, injures,
Nearest to universe: gallen, fictional, disconnected, christiern, monkey, lvares, scipione, constellations,
Nearest to defense: guard, attrition, iea, comix, pitman, bgn, erudite, ignition,
Nearest to square: knee, differentiability, specimen, rim, burton, mumy, ethologist, westport,
Nearest to egypt: jericho, petrel, invading, countenance, spree, andrews, egyptian, colonised,
Nearest to smith: garner, mcm, geddes, thera, purr, alvin, pup, huck,
Epoch 4/10 Iteration: 40100 Avg. Training loss: 4.2612 0.2764 sec/batch
Epoch 4/10 Iteration: 40200 Avg. Training loss: 4.3374 0.2716 sec/batch
Epoch 4/10 Iteration: 40300 Avg. Training loss: 4.2993 0.2731 sec/batch
Epoch 4/10 Iteration: 40400 Avg. Training loss: 4.3663 0.2693 sec/batch
Epoch 4/10 Iteration: 40500 Avg. Training loss: 4.3537 0.2722 sec/batch
Epoch 4/10 Iteration: 40600 Avg. Training loss: 4.3492 0.2684 sec/batch
Epoch 4/10 Iteration: 40700 Avg. Training loss: 4.3479 0.2689 sec/batch
Epoch 4/10 Iteration: 40800 Avg. Training loss: 4.3015 0.2686 sec/batch
Epoch 4/10 Iteration: 40900 Avg. Training loss: 4.3434 0.2720 sec/batch
Epoch 4/10 Iteration: 41000 Avg. Training loss: 4.2992 0.2685 sec/batch
Nearest to into: back, mbps, straddles, knobs, destry, distancing, diverted, ariel,
Nearest to during: antiproton, carangi, before, first, late, lamanites, beginning, following,
Nearest to about: shortening, regarding, pertain, thacker, discography, rugians, depleted, wired,
Nearest to five: one, four, zero, seven, six, two, eight, nine,
Nearest to six: three, one, nine, five, eight, four, two, rocha,
Nearest to some: many, these, several, such, most, other, various, both,
Nearest to however: although, but, though, because, quite, that, since, sativa,
Nearest to war: ii, wwii, veterans, outbreak, campaign, world, nationalists, wwi,
Nearest to experience: maybe, integrase, hellene, incoherent, gabrieli, subdivide, isopropanol, observations,
Nearest to magazine: macmullen, graven, cariou, impending, zeppelin, coil, lps, histoire,
Nearest to joseph: retrofitted, higginbotham, scarecrow, shop, stu, caithness, sarah, moglen,
Nearest to universe: gallen, disconnected, fictional, christiern, lvares, monkey, scipione, constellations,
Nearest to defense: guard, attrition, comix, pitman, bgn, iea, ignition, erudite,
Nearest to square: knee, differentiability, specimen, burton, mumy, ethologist, dock, shaper,
Nearest to egypt: jericho, petrel, egyptian, spree, colonised, instytut, countenance, garuda,
Nearest to smith: garner, thera, geddes, mcm, alvin, purr, pup, spartacus,
Epoch 4/10 Iteration: 41100 Avg. Training loss: 4.3475 0.2725 sec/batch
Epoch 4/10 Iteration: 41200 Avg. Training loss: 4.2302 0.2679 sec/batch
Epoch 4/10 Iteration: 41300 Avg. Training loss: 4.3809 0.2682 sec/batch
Epoch 4/10 Iteration: 41400 Avg. Training loss: 4.3697 0.2706 sec/batch
Epoch 4/10 Iteration: 41500 Avg. Training loss: 4.3011 0.2688 sec/batch
Epoch 4/10 Iteration: 41600 Avg. Training loss: 4.2869 0.2688 sec/batch
Epoch 4/10 Iteration: 41700 Avg. Training loss: 4.2892 0.2700 sec/batch
Epoch 4/10 Iteration: 41800 Avg. Training loss: 4.3398 0.2678 sec/batch
Epoch 4/10 Iteration: 41900 Avg. Training loss: 4.3397 0.2744 sec/batch
Epoch 4/10 Iteration: 42000 Avg. Training loss: 4.3641 0.2694 sec/batch
Nearest to into: back, straddles, destry, mbps, furrows, knobs, diverted, divisions,
Nearest to during: antiproton, before, first, carangi, lamanites, throughout, late, prior,
Nearest to about: shortening, regarding, pertain, thacker, approximately, lalor, discography, depleted,
Nearest to five: one, zero, four, two, six, seven, eight, three,
Nearest to six: three, one, five, eight, nine, rocha, four, katsuhiro,
Nearest to some: many, these, several, such, various, other, most, both,
Nearest to however: but, although, though, because, quite, that, while, since,
Nearest to war: ii, wwii, veterans, nationalists, outbreak, wars, campaign, occupation,
Nearest to experience: maybe, hellene, integrase, incoherent, isopropanol, gabrieli, observations, varnish,
Nearest to magazine: macmullen, graven, cariou, coil, impending, zeppelin, magazines, handguns,
Nearest to joseph: retrofitted, higginbotham, shop, scarecrow, caithness, sarah, moglen, votive,
Nearest to universe: fictional, gallen, disconnected, constellations, christiern, lvares, mash, monkey,
Nearest to defense: guard, pitman, comix, attrition, bgn, iea, erudite, ignition,
Nearest to square: knee, differentiability, specimen, burton, mumy, ethologist, rim, shaper,
Nearest to egypt: jericho, petrel, egyptian, colonised, spree, andrews, circa, receded,
Nearest to smith: garner, geddes, thera, mcm, alvin, purr, spartacus, pup,
Epoch 4/10 Iteration: 42100 Avg. Training loss: 4.3853 0.2690 sec/batch
Epoch 4/10 Iteration: 42200 Avg. Training loss: 4.3633 0.2688 sec/batch
Epoch 4/10 Iteration: 42300 Avg. Training loss: 4.2952 0.2751 sec/batch
Epoch 4/10 Iteration: 42400 Avg. Training loss: 4.3690 0.2682 sec/batch
Epoch 4/10 Iteration: 42500 Avg. Training loss: 4.3679 0.2723 sec/batch
Epoch 4/10 Iteration: 42600 Avg. Training loss: 4.3273 0.2716 sec/batch
Epoch 4/10 Iteration: 42700 Avg. Training loss: 4.3569 0.2783 sec/batch
Epoch 4/10 Iteration: 42800 Avg. Training loss: 4.3659 0.2705 sec/batch
Epoch 4/10 Iteration: 42900 Avg. Training loss: 4.3652 0.2706 sec/batch
Epoch 4/10 Iteration: 43000 Avg. Training loss: 4.3689 0.2660 sec/batch
Nearest to into: back, straddles, destry, knobs, throats, diverted, furrows, distancing,
Nearest to during: antiproton, before, lamanites, throughout, carangi, first, late, prior,
Nearest to about: shortening, regarding, pertain, thacker, approximately, albanians, lalor, discography,
Nearest to five: one, four, zero, two, six, seven, eight, patristics,
Nearest to six: three, five, one, eight, nine, four, katsuhiro, rocha,
Nearest to some: many, these, several, such, other, various, most, both,
Nearest to however: but, although, though, because, quite, that, while, since,
Nearest to war: ii, wwii, outbreak, campaign, wars, nationalists, knot, veterans,
Nearest to experience: maybe, hellene, integrase, incoherent, isopropanol, gabrieli, auditory, beni,
Nearest to magazine: macmullen, graven, cariou, coil, impending, magazines, handguns, zeppelin,
Nearest to joseph: retrofitted, higginbotham, scarecrow, shop, votive, moglen, sarah, caithness,
Nearest to universe: fictional, gallen, disconnected, christiern, lvares, constellations, monkey, scipione,
Nearest to defense: guard, bgn, comix, pitman, attrition, iea, rodentia, erudite,
Nearest to square: knee, differentiability, specimen, burton, mumy, upland, rim, ethologist,
Nearest to egypt: jericho, petrel, colonised, egyptian, receded, spree, circa, andrews,
Nearest to smith: garner, thera, mcm, geddes, purr, alvin, pup, commence,
Epoch 4/10 Iteration: 43100 Avg. Training loss: 4.3166 0.2753 sec/batch
Epoch 4/10 Iteration: 43200 Avg. Training loss: 4.2575 0.2705 sec/batch
Epoch 4/10 Iteration: 43300 Avg. Training loss: 4.2980 0.2699 sec/batch
Epoch 4/10 Iteration: 43400 Avg. Training loss: 4.3353 0.2698 sec/batch
Epoch 4/10 Iteration: 43500 Avg. Training loss: 4.3303 0.2700 sec/batch
Epoch 4/10 Iteration: 43600 Avg. Training loss: 4.2806 0.2690 sec/batch
Epoch 4/10 Iteration: 43700 Avg. Training loss: 4.3779 0.2683 sec/batch
Epoch 4/10 Iteration: 43800 Avg. Training loss: 4.1125 0.2690 sec/batch
Epoch 4/10 Iteration: 43900 Avg. Training loss: 4.2178 0.2656 sec/batch
Epoch 4/10 Iteration: 44000 Avg. Training loss: 4.1840 0.2706 sec/batch
Nearest to into: back, destry, straddles, mbps, mariana, from, raping, throats,
Nearest to during: antiproton, before, carangi, first, following, throughout, early, lamanites,
Nearest to about: shortening, regarding, discography, pertain, thacker, depleted, lalor, gssps,
Nearest to five: one, four, six, zero, two, seven, and, three,
Nearest to six: three, five, one, eight, nine, four, rocha, two,
Nearest to some: many, these, several, other, such, various, most, both,
Nearest to however: although, but, though, because, that, quite, since, while,
Nearest to war: ii, campaign, wars, veterans, outbreak, wwii, knot, nationalists,
Nearest to experience: maybe, hellene, beni, incoherent, varnish, gabrieli, lawyer, isopropanol,
Nearest to magazine: macmullen, graven, cariou, coil, magazines, monitors, published, zeppelin,
Nearest to joseph: retrofitted, higginbotham, moglen, sarah, caithness, scarecrow, dynes, votive,
Nearest to universe: fictional, gallen, disconnected, christiern, constellations, lvares, monkey, prot,
Nearest to defense: guard, attrition, pitman, comix, bgn, iea, rodentia, erudite,
Nearest to square: knee, specimen, mumy, burton, differentiability, shaper, upland, westport,
Nearest to egypt: jericho, petrel, egyptian, colonised, spree, andrews, countenance, garuda,
Nearest to smith: garner, mcm, geddes, thera, purr, pup, alvin, spartacus,
Epoch 4/10 Iteration: 44100 Avg. Training loss: 4.3253 0.2686 sec/batch
Epoch 4/10 Iteration: 44200 Avg. Training loss: 4.3652 0.2687 sec/batch
Epoch 4/10 Iteration: 44300 Avg. Training loss: 4.4057 0.2695 sec/batch
Epoch 4/10 Iteration: 44400 Avg. Training loss: 4.3441 0.2737 sec/batch
Epoch 4/10 Iteration: 44500 Avg. Training loss: 4.4162 0.2727 sec/batch
Epoch 4/10 Iteration: 44600 Avg. Training loss: 4.4226 0.2733 sec/batch
Epoch 4/10 Iteration: 44700 Avg. Training loss: 4.3719 0.2703 sec/batch
Epoch 4/10 Iteration: 44800 Avg. Training loss: 4.3203 0.2721 sec/batch
Epoch 4/10 Iteration: 44900 Avg. Training loss: 4.2240 0.2698 sec/batch
Epoch 4/10 Iteration: 45000 Avg. Training loss: 4.3406 0.2660 sec/batch
Nearest to into: back, straddles, destry, divisions, mariana, mbps, distancing, diverted,
Nearest to during: antiproton, carangi, before, lamanites, first, beginning, throughout, barak,
Nearest to about: approximately, lalor, shortening, regarding, pertain, discography, thacker, depleted,
Nearest to five: one, zero, four, two, six, seven, three, eight,
Nearest to six: three, five, one, nine, eight, four, zero, two,
Nearest to some: many, these, several, such, other, various, most, both,
Nearest to however: although, but, though, because, quite, that, since, while,
Nearest to war: ii, wwii, outbreak, campaign, veterans, wars, nationalists, knot,
Nearest to experience: maybe, varnish, beni, hellene, lawyer, incoherent, isopropanol, integrase,
Nearest to magazine: macmullen, graven, cariou, coil, lps, skeptics, monitors, fetus,
Nearest to joseph: retrofitted, higginbotham, moglen, scarecrow, votive, caithness, ssen, sarah,
Nearest to universe: gallen, fictional, christiern, monkey, lvares, disconnected, constellations, harrington,
Nearest to defense: guard, comix, attrition, rodentia, bgn, iea, ignition, pitman,
Nearest to square: knee, specimen, upland, differentiability, mumy, dock, burton, shaper,
Nearest to egypt: jericho, petrel, colonised, egyptian, andrews, spree, receded, countenance,
Nearest to smith: garner, mcm, thera, geddes, purr, electing, pup, commence,
Epoch 4/10 Iteration: 45100 Avg. Training loss: 4.3025 0.2697 sec/batch
Epoch 4/10 Iteration: 45200 Avg. Training loss: 4.3762 0.2748 sec/batch
Epoch 4/10 Iteration: 45300 Avg. Training loss: 4.3108 0.2678 sec/batch
Epoch 4/10 Iteration: 45400 Avg. Training loss: 4.3714 0.2686 sec/batch
Epoch 4/10 Iteration: 45500 Avg. Training loss: 4.3228 0.2680 sec/batch
Epoch 4/10 Iteration: 45600 Avg. Training loss: 4.2080 0.2647 sec/batch
Epoch 4/10 Iteration: 45700 Avg. Training loss: 4.1549 0.2660 sec/batch
Epoch 4/10 Iteration: 45800 Avg. Training loss: 4.3641 0.2685 sec/batch
Epoch 4/10 Iteration: 45900 Avg. Training loss: 4.2581 0.2683 sec/batch
Epoch 4/10 Iteration: 46000 Avg. Training loss: 4.3027 0.2736 sec/batch
Nearest to into: back, straddles, mbps, destry, throats, through, hazzan, knobs,
Nearest to during: antiproton, carangi, first, before, lamanites, beginning, prior, esdras,
Nearest to about: shortening, regarding, discography, pertain, lalor, gssps, depleted, approximately,
Nearest to five: one, zero, six, four, two, seven, eight, nine,
Nearest to six: five, three, one, eight, nine, four, zero, two,
Nearest to some: many, these, several, various, other, such, most, both,
Nearest to however: although, but, though, because, that, quite, while, since,
Nearest to war: ii, outbreak, wwii, veterans, campaign, knot, wars, nationalists,
Nearest to experience: maybe, isopropanol, integrase, varnish, beni, hellene, gabrieli, incoherent,
Nearest to magazine: macmullen, graven, cariou, coil, lps, magazines, monitors, fetus,
Nearest to joseph: retrofitted, higginbotham, moglen, sarah, scarecrow, dynes, ssen, caithness,
Nearest to universe: fictional, gallen, lvares, christiern, constellations, disconnected, monkey, mainstream,
Nearest to defense: guard, comix, attrition, rodentia, ignition, iea, bgn, command,
Nearest to square: knee, burton, specimen, upland, differentiability, mumy, shaper, dock,
Nearest to egypt: jericho, petrel, egyptian, colonised, countenance, receded, andrews, spree,
Nearest to smith: garner, mcm, geddes, thera, purr, electing, commence, pup,
Epoch 4/10 Iteration: 46100 Avg. Training loss: 4.3542 0.2741 sec/batch
Epoch 4/10 Iteration: 46200 Avg. Training loss: 4.3419 0.2660 sec/batch
Epoch 4/10 Iteration: 46300 Avg. Training loss: 4.2801 0.2687 sec/batch
Epoch 4/10 Iteration: 46400 Avg. Training loss: 4.3541 0.2673 sec/batch
Epoch 4/10 Iteration: 46500 Avg. Training loss: 4.3342 0.2714 sec/batch
Epoch 4/10 Iteration: 46600 Avg. Training loss: 4.3982 0.2663 sec/batch
Epoch 5/10 Iteration: 46700 Avg. Training loss: 4.3817 0.1625 sec/batch
Epoch 5/10 Iteration: 46800 Avg. Training loss: 4.3004 0.2732 sec/batch
Epoch 5/10 Iteration: 46900 Avg. Training loss: 4.3225 0.2698 sec/batch
Epoch 5/10 Iteration: 47000 Avg. Training loss: 4.3481 0.2715 sec/batch
Nearest to into: back, straddles, destry, mbps, mariana, through, throats, distancing,
Nearest to during: antiproton, carangi, before, first, beginning, lamanites, following, early,
Nearest to about: regarding, shortening, approximately, depleted, discography, fabricius, pertain, gssps,
Nearest to five: one, zero, four, two, six, seven, three, eight,
Nearest to six: three, five, eight, one, nine, four, zero, obrenovi,
Nearest to some: many, these, several, various, such, other, most, both,
Nearest to however: although, but, though, because, that, quite, while, since,
Nearest to war: ii, veterans, wwii, campaign, outbreak, knot, wars, abyssinia,
Nearest to experience: maybe, isopropanol, hellene, lawyer, beni, integrase, glycogen, moles,
Nearest to magazine: macmullen, graven, cariou, coil, lps, magazines, impending, skeptics,
Nearest to joseph: retrofitted, higginbotham, moglen, sarah, scarecrow, ssen, caithness, votive,
Nearest to universe: fictional, gallen, disconnected, constellations, christiern, lvares, monkey, nigerian,
Nearest to defense: guard, attrition, comix, rodentia, iea, command, ignition, isothermal,
Nearest to square: knee, burton, specimen, dock, mumy, shaper, upland, rim,
Nearest to egypt: jericho, petrel, egyptian, colonised, circa, countenance, receded, andrews,
Nearest to smith: garner, mcm, geddes, thera, purr, commence, pup, electing,
Epoch 5/10 Iteration: 47100 Avg. Training loss: 4.2821 0.2710 sec/batch
Epoch 5/10 Iteration: 47200 Avg. Training loss: 4.3170 0.2672 sec/batch
Epoch 5/10 Iteration: 47300 Avg. Training loss: 4.2924 0.2669 sec/batch
Epoch 5/10 Iteration: 47400 Avg. Training loss: 4.2989 0.2672 sec/batch
Epoch 5/10 Iteration: 47500 Avg. Training loss: 4.3215 0.2661 sec/batch
Epoch 5/10 Iteration: 47600 Avg. Training loss: 4.3477 0.2699 sec/batch
Epoch 5/10 Iteration: 47700 Avg. Training loss: 4.3153 0.2664 sec/batch
Epoch 5/10 Iteration: 47800 Avg. Training loss: 4.3047 0.2673 sec/batch
Epoch 5/10 Iteration: 47900 Avg. Training loss: 4.3816 0.2676 sec/batch
Epoch 5/10 Iteration: 48000 Avg. Training loss: 4.3191 0.2683 sec/batch
Nearest to into: back, straddles, destry, hazzan, mbps, through, distancing, throats,
Nearest to during: antiproton, first, carangi, before, lamanites, beginning, throughout, prior,
Nearest to about: shortening, regarding, approximately, depleted, fabricius, gssps, rugians, iqs,
Nearest to five: one, four, zero, six, two, three, seven, eight,
Nearest to six: three, five, one, eight, four, nine, zero, two,
Nearest to some: many, several, these, various, other, most, such, both,
Nearest to however: but, although, though, because, that, while, quite, since,
Nearest to war: veterans, ii, wwii, outbreak, campaign, manassas, wars, knot,
Nearest to experience: maybe, isopropanol, moles, integrase, lawyer, glycogen, beni, hellene,
Nearest to magazine: macmullen, graven, cariou, coil, lps, histoire, fetus, handguns,
Nearest to joseph: retrofitted, higginbotham, moglen, sarah, caithness, votive, scarecrow, ssen,
Nearest to universe: fictional, gallen, disconnected, monkey, constellations, christiern, lvares, mwh,
Nearest to defense: guard, attrition, comix, rodentia, iea, ignition, erudite, pitman,
Nearest to square: knee, specimen, rim, burton, dock, upland, shaper, mumy,
Nearest to egypt: jericho, petrel, egyptian, colonised, countenance, receded, andrews, circa,
Nearest to smith: garner, mcm, geddes, thera, purr, electing, pup, spartacus,
Epoch 5/10 Iteration: 48100 Avg. Training loss: 4.3215 0.2684 sec/batch
Epoch 5/10 Iteration: 48200 Avg. Training loss: 4.3160 0.2647 sec/batch
Epoch 5/10 Iteration: 48300 Avg. Training loss: 4.2586 0.2707 sec/batch
Epoch 5/10 Iteration: 48400 Avg. Training loss: 4.2436 0.2705 sec/batch
Epoch 5/10 Iteration: 48500 Avg. Training loss: 4.3402 0.2726 sec/batch
Epoch 5/10 Iteration: 48600 Avg. Training loss: 4.3595 0.2716 sec/batch
Epoch 5/10 Iteration: 48700 Avg. Training loss: 4.3256 0.2672 sec/batch
Epoch 5/10 Iteration: 48800 Avg. Training loss: 4.3430 0.2710 sec/batch
Epoch 5/10 Iteration: 48900 Avg. Training loss: 4.3498 0.2745 sec/batch
Epoch 5/10 Iteration: 49000 Avg. Training loss: 4.3521 0.2727 sec/batch
Nearest to into: back, then, destry, through, straddles, hazzan, suffocating, divisions,
Nearest to during: antiproton, first, before, carangi, lamanites, esdras, late, beginning,
Nearest to about: shortening, regarding, approximately, fabricius, depleted, rugians, discography, thacker,
Nearest to five: one, four, zero, six, two, three, seven, eight,
Nearest to six: three, five, one, eight, nine, four, zero, two,
Nearest to some: many, these, several, most, various, such, other, both,
Nearest to however: but, although, though, because, that, while, quite, since,
Nearest to war: wwii, ii, outbreak, manassas, veterans, wars, knot, campaign,
Nearest to experience: maybe, isopropanol, hellene, integrase, auditory, gabrieli, moles, lawyer,
Nearest to magazine: macmullen, cariou, coil, graven, handguns, lps, fetus, impending,
Nearest to joseph: retrofitted, higginbotham, moglen, votive, scarecrow, sarah, ssen, phylum,
Nearest to universe: fictional, gallen, constellations, lvares, disconnected, mwh, christiern, monkey,
Nearest to defense: guard, rodentia, attrition, comix, iea, pitman, ignition, erudite,
Nearest to square: knee, specimen, rim, dock, upland, shaper, burton, ethologist,
Nearest to egypt: jericho, petrel, egyptian, colonised, countenance, receded, andrews, spree,
Nearest to smith: garner, geddes, mcm, thera, purr, joseph, pup, sisley,
Epoch 5/10 Iteration: 49100 Avg. Training loss: 4.3741 0.2732 sec/batch
Epoch 5/10 Iteration: 49200 Avg. Training loss: 4.3514 0.2704 sec/batch
Epoch 5/10 Iteration: 49300 Avg. Training loss: 4.2997 0.2695 sec/batch
Epoch 5/10 Iteration: 49400 Avg. Training loss: 4.2844 0.2760 sec/batch
Epoch 5/10 Iteration: 49500 Avg. Training loss: 4.1232 0.2723 sec/batch
Epoch 5/10 Iteration: 49600 Avg. Training loss: 4.2620 0.2704 sec/batch
Epoch 5/10 Iteration: 49700 Avg. Training loss: 4.3434 0.2661 sec/batch
Epoch 5/10 Iteration: 49800 Avg. Training loss: 4.2628 0.2696 sec/batch
Epoch 5/10 Iteration: 49900 Avg. Training loss: 4.3001 0.2701 sec/batch
Epoch 5/10 Iteration: 50000 Avg. Training loss: 4.2887 0.2680 sec/batch
Nearest to into: back, distancing, destry, straddles, through, hazzan, further, divisions,
Nearest to during: antiproton, before, carangi, lamanites, beginning, late, first, throughout,
Nearest to about: shortening, regarding, approximately, rugians, pertain, depleted, thacker, gssps,
Nearest to five: one, zero, four, six, two, three, seven, eight,
Nearest to six: three, five, eight, one, nine, four, zero, two,
Nearest to some: many, these, several, most, various, such, other, certain,
Nearest to however: but, although, though, because, that, while, since, quite,
Nearest to war: ii, wwii, outbreak, wars, veterans, campaign, manassas, knot,
Nearest to experience: maybe, isopropanol, gabrieli, integrase, lawyer, hellene, auditory, moles,
Nearest to magazine: macmullen, coil, graven, cariou, lps, monitors, handguns, magazines,
Nearest to joseph: smith, higginbotham, scarecrow, retrofitted, votive, phylum, moglen, ssen,
Nearest to universe: fictional, gallen, disconnected, lvares, constellations, mwh, christiern, monkey,
Nearest to defense: guard, rodentia, comix, attrition, pitman, ignition, bgn, iea,
Nearest to square: dock, knee, rim, specimen, upland, shaper, differentiability, westport,
Nearest to egypt: jericho, petrel, colonised, egyptian, countenance, receded, andrews, circa,
Nearest to smith: garner, geddes, thera, joseph, mcm, sisley, purr, spartacus,
Epoch 5/10 Iteration: 50100 Avg. Training loss: 4.2058 0.2710 sec/batch
Epoch 5/10 Iteration: 50200 Avg. Training loss: 4.3218 0.2690 sec/batch
Epoch 5/10 Iteration: 50300 Avg. Training loss: 4.3416 0.2708 sec/batch
Epoch 5/10 Iteration: 50400 Avg. Training loss: 4.3319 0.2674 sec/batch
Epoch 5/10 Iteration: 50500 Avg. Training loss: 4.3326 0.2673 sec/batch
Epoch 5/10 Iteration: 50600 Avg. Training loss: 4.3236 0.2701 sec/batch
Epoch 5/10 Iteration: 50700 Avg. Training loss: 4.3035 0.2687 sec/batch
Epoch 5/10 Iteration: 50800 Avg. Training loss: 4.3661 0.2704 sec/batch
Epoch 5/10 Iteration: 50900 Avg. Training loss: 4.3225 0.2674 sec/batch
Epoch 5/10 Iteration: 51000 Avg. Training loss: 4.2792 0.2687 sec/batch
Nearest to into: back, straddles, mariana, destry, throats, mbps, distancing, through,
Nearest to during: antiproton, before, first, carangi, beginning, lamanites, late, prior,
Nearest to about: shortening, regarding, approximately, rugians, discography, depleted, lalor, thacker,
Nearest to five: one, zero, four, two, six, three, seven, eight,
Nearest to six: three, five, one, eight, nine, four, zero, two,
Nearest to some: many, these, several, most, various, such, other, certain,
Nearest to however: although, but, though, because, that, since, while, quite,
Nearest to war: wwii, ii, veterans, outbreak, manassas, wars, campaign, knot,
Nearest to experience: maybe, isopropanol, gabrieli, auditory, integrase, hellene, lawyer, moles,
Nearest to magazine: macmullen, cariou, graven, coil, lps, handguns, fetus, skeptics,
Nearest to joseph: higginbotham, scarecrow, retrofitted, smith, moglen, votive, sarah, ssen,
Nearest to universe: gallen, fictional, lvares, disconnected, constellations, monkey, big, christiern,
Nearest to defense: guard, comix, rodentia, pitman, attrition, ignition, erudite, iea,
Nearest to square: dock, knee, upland, rim, differentiability, specimen, burton, shaper,
Nearest to egypt: jericho, petrel, colonised, andrews, countenance, receded, egyptian, circa,
Nearest to smith: garner, geddes, mcm, joseph, thera, purr, pup, sisley,
Epoch 5/10 Iteration: 51100 Avg. Training loss: 4.3166 0.2725 sec/batch
Epoch 5/10 Iteration: 51200 Avg. Training loss: 4.2826 0.2681 sec/batch
Epoch 5/10 Iteration: 51300 Avg. Training loss: 4.3551 0.2729 sec/batch
Epoch 5/10 Iteration: 51400 Avg. Training loss: 4.3512 0.2696 sec/batch
Epoch 5/10 Iteration: 51500 Avg. Training loss: 4.3725 0.2711 sec/batch
Epoch 5/10 Iteration: 51600 Avg. Training loss: 4.3402 0.2664 sec/batch
Epoch 5/10 Iteration: 51700 Avg. Training loss: 4.2548 0.2662 sec/batch
Epoch 5/10 Iteration: 51800 Avg. Training loss: 4.3197 0.2687 sec/batch
Epoch 5/10 Iteration: 51900 Avg. Training loss: 4.2550 0.2712 sec/batch
Epoch 5/10 Iteration: 52000 Avg. Training loss: 4.3336 0.2690 sec/batch
Nearest to into: back, straddles, mbps, distancing, throats, destry, further, hazzan,
Nearest to during: antiproton, before, lamanites, carangi, first, late, beginning, early,
Nearest to about: shortening, regarding, approximately, rugians, discography, thacker, depleted, gssps,
Nearest to five: one, four, zero, six, two, seven, three, eight,
Nearest to six: five, three, eight, one, nine, four, two, zero,
Nearest to some: many, several, these, most, various, such, certain, other,
Nearest to however: although, but, though, because, that, since, while, nevertheless,
Nearest to war: wwii, ii, outbreak, veterans, campaign, wars, manassas, knot,
Nearest to experience: maybe, isopropanol, gabrieli, integrase, moles, auditory, hellene, lawyer,
Nearest to magazine: macmullen, graven, cariou, coil, lps, fetus, published, magazines,
Nearest to joseph: higginbotham, retrofitted, scarecrow, moglen, smith, ssen, sarah, shop,
Nearest to universe: gallen, fictional, lvares, disconnected, constellations, monkey, big, mainstream,
Nearest to defense: guard, comix, rodentia, attrition, pitman, ignition, iea, bgn,
Nearest to square: knee, differentiability, rim, dock, upland, burton, ethologist, shaper,
Nearest to egypt: jericho, petrel, egyptian, andrews, colonised, garuda, receded, instytut,
Nearest to smith: garner, geddes, thera, mcm, joseph, purr, pup, huck,
Epoch 5/10 Iteration: 52100 Avg. Training loss: 4.3356 0.2673 sec/batch
Epoch 5/10 Iteration: 52200 Avg. Training loss: 4.2972 0.2649 sec/batch
Epoch 5/10 Iteration: 52300 Avg. Training loss: 4.3596 0.2668 sec/batch
Epoch 5/10 Iteration: 52400 Avg. Training loss: 4.3069 0.2690 sec/batch
Epoch 5/10 Iteration: 52500 Avg. Training loss: 4.2851 0.2674 sec/batch
Epoch 5/10 Iteration: 52600 Avg. Training loss: 4.3031 0.2664 sec/batch
Epoch 5/10 Iteration: 52700 Avg. Training loss: 4.2901 0.2680 sec/batch
Epoch 5/10 Iteration: 52800 Avg. Training loss: 4.3703 0.2717 sec/batch
Epoch 5/10 Iteration: 52900 Avg. Training loss: 4.2185 0.2692 sec/batch
Epoch 5/10 Iteration: 53000 Avg. Training loss: 4.3258 0.2678 sec/batch
Nearest to into: back, distancing, destry, mbps, further, categories, straddles, knobs,
Nearest to during: antiproton, first, carangi, before, lamanites, throughout, late, prior,
Nearest to about: regarding, shortening, rugians, discography, depleted, pertain, thacker, hollingworth,
Nearest to five: one, four, six, zero, two, seven, three, eight,
Nearest to six: five, three, one, eight, nine, four, crompton, zero,
Nearest to some: many, these, several, such, most, various, certain, other,
Nearest to however: although, but, though, because, that, since, while, nevertheless,
Nearest to war: ii, wwii, outbreak, veterans, campaign, manassas, world, wwi,
Nearest to experience: maybe, gabrieli, isopropanol, integrase, incoherent, moles, auditory, observations,
Nearest to magazine: macmullen, graven, cariou, coil, lps, magazines, fetus, histoire,
Nearest to joseph: higginbotham, retrofitted, sarah, smith, scarecrow, moglen, ssen, dynes,
Nearest to universe: gallen, fictional, lvares, constellations, disconnected, mainstream, religions, big,
Nearest to defense: guard, comix, pitman, rodentia, ignition, attrition, iea, erudite,
Nearest to square: knee, dock, burton, differentiability, rim, shaper, specimen, mumy,
Nearest to egypt: jericho, petrel, andrews, egyptian, colonised, receded, garuda, countenance,
Nearest to smith: garner, geddes, mcm, thera, joseph, pup, purr, spartacus,
Epoch 5/10 Iteration: 53100 Avg. Training loss: 4.3418 0.2697 sec/batch
Epoch 5/10 Iteration: 53200 Avg. Training loss: 4.2981 0.2671 sec/batch
Epoch 5/10 Iteration: 53300 Avg. Training loss: 4.1631 0.2696 sec/batch
Epoch 5/10 Iteration: 53400 Avg. Training loss: 4.3601 0.2768 sec/batch
Epoch 5/10 Iteration: 53500 Avg. Training loss: 4.3557 0.2738 sec/batch
Epoch 5/10 Iteration: 53600 Avg. Training loss: 4.3138 0.2682 sec/batch
Epoch 5/10 Iteration: 53700 Avg. Training loss: 4.3414 0.2658 sec/batch
Epoch 5/10 Iteration: 53800 Avg. Training loss: 4.4021 0.2686 sec/batch
Epoch 5/10 Iteration: 53900 Avg. Training loss: 4.2741 0.2702 sec/batch
Epoch 5/10 Iteration: 54000 Avg. Training loss: 4.2765 0.2643 sec/batch
Nearest to into: back, distancing, straddles, eventually, further, raping, then, mariana,
Nearest to during: antiproton, before, in, throughout, prior, late, lamanites, first,
Nearest to about: regarding, shortening, approximately, thacker, rugians, zero, depleted, discography,
Nearest to five: one, four, zero, six, two, seven, three, eight,
Nearest to six: five, three, one, eight, nine, four, two, zero,
Nearest to some: many, these, several, most, various, such, certain, other,
Nearest to however: but, although, though, because, that, while, since, quite,
Nearest to war: ii, wwii, outbreak, during, campaign, wars, manassas, veterans,
Nearest to experience: maybe, isopropanol, gabrieli, integrase, incoherent, beni, hellene, auditory,
Nearest to magazine: macmullen, coil, graven, magazines, cariou, handguns, lps, histoire,
Nearest to joseph: higginbotham, sarah, retrofitted, smith, moglen, scarecrow, injures, unemployment,
Nearest to universe: gallen, fictional, constellations, disconnected, lvares, mainstream, monkey, christiern,
Nearest to defense: guard, comix, pitman, attrition, ignition, rodentia, caustic, iea,
Nearest to square: knee, dock, rim, upland, differentiability, shaper, specimen, interesting,
Nearest to egypt: jericho, petrel, egyptian, andrews, receded, colonised, garuda, spree,
Nearest to smith: garner, geddes, mcm, thera, joseph, purr, pup, sisley,
Epoch 5/10 Iteration: 54100 Avg. Training loss: 4.3487 0.2708 sec/batch
Epoch 5/10 Iteration: 54200 Avg. Training loss: 4.3382 0.2676 sec/batch
Epoch 5/10 Iteration: 54300 Avg. Training loss: 4.3077 0.2677 sec/batch
Epoch 5/10 Iteration: 54400 Avg. Training loss: 4.3323 0.2678 sec/batch
Epoch 5/10 Iteration: 54500 Avg. Training loss: 4.3272 0.2658 sec/batch
Epoch 5/10 Iteration: 54600 Avg. Training loss: 4.3555 0.2712 sec/batch
Epoch 5/10 Iteration: 54700 Avg. Training loss: 4.3164 0.2706 sec/batch
Epoch 5/10 Iteration: 54800 Avg. Training loss: 4.2124 0.2700 sec/batch
Epoch 5/10 Iteration: 54900 Avg. Training loss: 4.2973 0.2692 sec/batch
Epoch 5/10 Iteration: 55000 Avg. Training loss: 4.3062 0.2664 sec/batch
Nearest to into: back, further, straddles, throats, distancing, divisions, mbps, categories,
Nearest to during: antiproton, throughout, lamanites, prior, before, late, carangi, beginning,
Nearest to about: regarding, shortening, approximately, zero, depleted, thacker, rugians, iqs,
Nearest to five: one, four, zero, two, six, three, seven, eight,
Nearest to six: five, three, one, eight, nine, four, zero, two,
Nearest to some: many, these, several, most, such, various, other, certain,
Nearest to however: although, but, though, because, that, while, since, only,
Nearest to war: ii, outbreak, wwii, campaign, wars, manassas, nationalists, conflict,
Nearest to experience: maybe, isopropanol, gabrieli, beni, hellene, auditory, lawyer, moles,
Nearest to magazine: macmullen, coil, magazines, graven, cariou, handguns, lps, fetus,
Nearest to joseph: higginbotham, sarah, retrofitted, moglen, smith, scarecrow, votive, dynes,
Nearest to universe: gallen, fictional, constellations, lvares, disconnected, christiern, monkey, mainstream,
Nearest to defense: guard, pitman, rodentia, comix, attrition, ignition, iea, isothermal,
Nearest to square: knee, upland, dock, differentiability, rim, shaper, interesting, westport,
Nearest to egypt: jericho, egyptian, petrel, colonised, andrews, receded, circa, garuda,
Nearest to smith: garner, geddes, thera, mcm, joseph, purr, pup, sisley,
Epoch 5/10 Iteration: 55100 Avg. Training loss: 4.3048 0.2727 sec/batch
Epoch 5/10 Iteration: 55200 Avg. Training loss: 4.2985 0.2732 sec/batch
Epoch 5/10 Iteration: 55300 Avg. Training loss: 4.3261 0.2692 sec/batch
Epoch 5/10 Iteration: 55400 Avg. Training loss: 4.2989 0.2651 sec/batch
Epoch 5/10 Iteration: 55500 Avg. Training loss: 4.0350 0.2699 sec/batch
Epoch 5/10 Iteration: 55600 Avg. Training loss: 4.1683 0.2646 sec/batch
Epoch 5/10 Iteration: 55700 Avg. Training loss: 4.2469 0.2678 sec/batch
Epoch 5/10 Iteration: 55800 Avg. Training loss: 4.3110 0.2724 sec/batch
Epoch 5/10 Iteration: 55900 Avg. Training loss: 4.3403 0.2687 sec/batch
Epoch 5/10 Iteration: 56000 Avg. Training loss: 4.3204 0.2699 sec/batch
Nearest to into: back, destry, distancing, mariana, straddles, further, categories, then,
Nearest to during: antiproton, throughout, before, lamanites, following, first, carangi, beginning,
Nearest to about: regarding, shortening, thacker, depleted, approximately, discography, rugians, lalor,
Nearest to five: one, four, six, zero, two, three, seven, eight,
Nearest to six: five, three, one, eight, nine, four, two, zero,
Nearest to some: many, these, several, such, most, various, other, certain,
Nearest to however: but, although, though, because, while, that, since, nevertheless,
Nearest to war: ii, campaign, outbreak, wwii, veterans, knot, manassas, wars,
Nearest to experience: maybe, beni, lawyer, hellene, severn, isopropanol, varnish, incoherent,
Nearest to magazine: macmullen, cariou, graven, magazines, coil, lps, published, fetus,
Nearest to joseph: higginbotham, smith, sarah, moglen, ssen, retrofitted, phylum, votive,
Nearest to universe: gallen, fictional, constellations, disconnected, lvares, monkey, prot, christiern,
Nearest to defense: guard, comix, rodentia, pitman, attrition, ignition, paulist, iea,
Nearest to square: knee, upland, dock, shaper, mumy, specimen, rim, westport,
Nearest to egypt: jericho, egyptian, petrel, andrews, receded, countenance, garuda, colonised,
Nearest to smith: garner, geddes, mcm, thera, joseph, sisley, purr, lila,
Epoch 5/10 Iteration: 56100 Avg. Training loss: 4.3899 0.2730 sec/batch
Epoch 5/10 Iteration: 56200 Avg. Training loss: 4.4032 0.2772 sec/batch
Epoch 5/10 Iteration: 56300 Avg. Training loss: 4.4038 0.2709 sec/batch
Epoch 5/10 Iteration: 56400 Avg. Training loss: 4.3434 0.2654 sec/batch
Epoch 5/10 Iteration: 56500 Avg. Training loss: 4.1844 0.2740 sec/batch
Epoch 5/10 Iteration: 56600 Avg. Training loss: 4.3013 0.2716 sec/batch
Epoch 5/10 Iteration: 56700 Avg. Training loss: 4.3337 0.2673 sec/batch
Epoch 5/10 Iteration: 56800 Avg. Training loss: 4.2933 0.2645 sec/batch
Epoch 5/10 Iteration: 56900 Avg. Training loss: 4.3093 0.2739 sec/batch
Epoch 5/10 Iteration: 57000 Avg. Training loss: 4.3747 0.2688 sec/batch
Nearest to into: back, distancing, straddles, destry, categories, mbps, mariana, onto,
Nearest to during: antiproton, lamanites, beginning, before, throughout, carangi, prior, barak,
Nearest to about: regarding, shortening, thacker, discography, depleted, approximately, hollingworth, lalor,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, one, eight, three, nine, four, two, zero,
Nearest to some: many, these, several, such, most, various, other, certain,
Nearest to however: but, although, though, because, that, while, since, nevertheless,
Nearest to war: ii, outbreak, wwii, campaign, veterans, knot, manassas, nationalists,
Nearest to experience: maybe, isopropanol, varnish, beni, lawyer, auditory, incoherent, integrase,
Nearest to magazine: macmullen, coil, cariou, magazines, graven, skeptics, monitors, handguns,
Nearest to joseph: higginbotham, moglen, sarah, retrofitted, smith, ssen, phylum, scarecrow,
Nearest to universe: fictional, gallen, constellations, disconnected, lvares, prot, harrington, christiern,
Nearest to defense: guard, comix, rodentia, ignition, attrition, caustic, pitman, bgn,
Nearest to square: knee, dock, interesting, differentiability, rim, upland, specimen, shaper,
Nearest to egypt: jericho, egyptian, petrel, colonised, andrews, receded, garuda, countenance,
Nearest to smith: garner, geddes, thera, mcm, joseph, purr, lila, sisley,
Epoch 5/10 Iteration: 57100 Avg. Training loss: 4.2930 0.2709 sec/batch
Epoch 5/10 Iteration: 57200 Avg. Training loss: 4.3064 0.2692 sec/batch
Epoch 5/10 Iteration: 57300 Avg. Training loss: 4.1200 0.2676 sec/batch
Epoch 5/10 Iteration: 57400 Avg. Training loss: 4.2782 0.2635 sec/batch
Epoch 5/10 Iteration: 57500 Avg. Training loss: 4.3234 0.2677 sec/batch
Epoch 5/10 Iteration: 57600 Avg. Training loss: 4.2050 0.2705 sec/batch
Epoch 5/10 Iteration: 57700 Avg. Training loss: 4.3171 0.2664 sec/batch
Epoch 5/10 Iteration: 57800 Avg. Training loss: 4.3193 0.2677 sec/batch
Epoch 5/10 Iteration: 57900 Avg. Training loss: 4.2973 0.2660 sec/batch
Epoch 5/10 Iteration: 58000 Avg. Training loss: 4.2953 0.2691 sec/batch
Nearest to into: back, mbps, distancing, mariana, categories, destry, furrows, onto,
Nearest to during: antiproton, before, beginning, throughout, carangi, first, late, lamanites,
Nearest to about: regarding, approximately, discography, shortening, depleted, thacker, wired, iqs,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, eight, one, three, four, nine, two, zero,
Nearest to some: many, several, these, such, various, most, other, certain,
Nearest to however: but, although, though, because, while, that, since, and,
Nearest to war: ii, wwii, outbreak, knot, campaign, veterans, wars, manassas,
Nearest to experience: maybe, isopropanol, subjective, auditory, glycogen, hellene, incoherent, varnish,
Nearest to magazine: macmullen, coil, handguns, cariou, lps, magazines, graven, recoil,
Nearest to joseph: higginbotham, moglen, sarah, smith, retrofitted, ssen, unemployment, scarecrow,
Nearest to universe: fictional, gallen, constellations, disconnected, lvares, monkey, christiern, mainstream,
Nearest to defense: guard, comix, rodentia, ignition, attrition, caustic, erudite, ecclesiastic,
Nearest to square: knee, dock, interesting, burton, rim, differentiability, upland, triangle,
Nearest to egypt: jericho, egyptian, petrel, andrews, countenance, receded, colonised, circa,
Nearest to smith: garner, geddes, thera, mcm, joseph, lila, purr, sisley,
Epoch 5/10 Iteration: 58100 Avg. Training loss: 4.3163 0.2691 sec/batch
Epoch 5/10 Iteration: 58200 Avg. Training loss: 4.3446 0.2633 sec/batch
Epoch 5/10 Iteration: 58300 Avg. Training loss: 4.3853 0.2690 sec/batch
Epoch 6/10 Iteration: 58400 Avg. Training loss: 4.3459 0.2678 sec/batch
Epoch 6/10 Iteration: 58500 Avg. Training loss: 4.2674 0.2675 sec/batch
Epoch 6/10 Iteration: 58600 Avg. Training loss: 4.3246 0.2673 sec/batch
Epoch 6/10 Iteration: 58700 Avg. Training loss: 4.2768 0.2667 sec/batch
Epoch 6/10 Iteration: 58800 Avg. Training loss: 4.3244 0.2703 sec/batch
Epoch 6/10 Iteration: 58900 Avg. Training loss: 4.2969 0.2687 sec/batch
Epoch 6/10 Iteration: 59000 Avg. Training loss: 4.2552 0.2714 sec/batch
Nearest to into: back, then, distancing, furrows, mariana, throats, destry, onto,
Nearest to during: antiproton, before, first, beginning, throughout, late, early, carangi,
Nearest to about: regarding, approximately, zero, shortening, depleted, five, discography, main,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: one, five, eight, three, four, nine, two, seven,
Nearest to some: many, several, these, such, most, various, other, certain,
Nearest to however: but, although, though, because, that, while, since, and,
Nearest to war: ii, wwii, outbreak, veterans, campaign, wars, manassas, knot,
Nearest to experience: maybe, subjective, hellene, incoherent, auditory, gabrieli, lawyer, sensory,
Nearest to magazine: macmullen, coil, cariou, handguns, graven, magazines, histoire, skeptics,
Nearest to joseph: higginbotham, moglen, sarah, phylum, retrofitted, ssen, smith, unemployment,
Nearest to universe: fictional, gallen, disconnected, constellations, monkey, lvares, harrington, prot,
Nearest to defense: guard, comix, ignition, attrition, rodentia, palpitations, pitman, parody,
Nearest to square: knee, dock, interesting, rim, upland, triangle, specimen, burton,
Nearest to egypt: jericho, egyptian, petrel, andrews, receded, countenance, colonised, claw,
Nearest to smith: garner, geddes, thera, mcm, joseph, purr, lila, sisley,
Epoch 6/10 Iteration: 59100 Avg. Training loss: 4.2780 0.2758 sec/batch
Epoch 6/10 Iteration: 59200 Avg. Training loss: 4.3123 0.2725 sec/batch
Epoch 6/10 Iteration: 59300 Avg. Training loss: 4.3387 0.2708 sec/batch
Epoch 6/10 Iteration: 59400 Avg. Training loss: 4.2715 0.2713 sec/batch
Epoch 6/10 Iteration: 59500 Avg. Training loss: 4.3200 0.2697 sec/batch
Epoch 6/10 Iteration: 59600 Avg. Training loss: 4.3441 0.2790 sec/batch
Epoch 6/10 Iteration: 59700 Avg. Training loss: 4.3299 0.2731 sec/batch
Epoch 6/10 Iteration: 59800 Avg. Training loss: 4.3295 0.2683 sec/batch
Epoch 6/10 Iteration: 59900 Avg. Training loss: 4.2944 0.2761 sec/batch
Epoch 6/10 Iteration: 60000 Avg. Training loss: 4.1427 0.2688 sec/batch
Nearest to into: back, further, mariana, sub, mbps, divisions, then, divided,
Nearest to during: antiproton, late, throughout, before, beginning, early, first, carangi,
Nearest to about: regarding, zero, approximately, five, discography, main, shortening, depleted,
Nearest to five: four, one, zero, two, six, three, seven, eight,
Nearest to six: five, three, one, four, eight, two, nine, zero,
Nearest to some: many, these, several, most, such, various, other, certain,
Nearest to however: but, although, though, because, that, since, while, and,
Nearest to war: ii, wwii, wars, outbreak, manassas, veterans, knot, campaign,
Nearest to experience: maybe, auditory, isopropanol, subjective, sensory, gz, glycogen, hellene,
Nearest to magazine: macmullen, coil, handguns, graven, magazines, skeptics, lps, fetus,
Nearest to joseph: higginbotham, moglen, smith, sarah, retrofitted, phylum, unemployment, ssen,
Nearest to universe: fictional, gallen, constellations, disconnected, monkey, lvares, harrington, prot,
Nearest to defense: guard, ignition, attrition, comix, rodentia, iea, pitman, bgn,
Nearest to square: knee, dock, rim, upland, interesting, specimen, wadis, burton,
Nearest to egypt: jericho, egyptian, petrel, countenance, andrews, receded, colonised, claw,
Nearest to smith: garner, geddes, thera, mcm, joseph, sisley, lila, purr,
Epoch 6/10 Iteration: 60100 Avg. Training loss: 4.3624 0.2742 sec/batch
Epoch 6/10 Iteration: 60200 Avg. Training loss: 4.3203 0.2728 sec/batch
Epoch 6/10 Iteration: 60300 Avg. Training loss: 4.3365 0.2684 sec/batch
Epoch 6/10 Iteration: 60400 Avg. Training loss: 4.3308 0.2725 sec/batch
Epoch 6/10 Iteration: 60500 Avg. Training loss: 4.3274 0.2698 sec/batch
Epoch 6/10 Iteration: 60600 Avg. Training loss: 4.3319 0.2698 sec/batch
Epoch 6/10 Iteration: 60700 Avg. Training loss: 4.3592 0.2689 sec/batch
Epoch 6/10 Iteration: 60800 Avg. Training loss: 4.3295 0.2714 sec/batch
Epoch 6/10 Iteration: 60900 Avg. Training loss: 4.3403 0.2696 sec/batch
Epoch 6/10 Iteration: 61000 Avg. Training loss: 4.2460 0.2674 sec/batch
Nearest to into: back, further, distancing, destry, mbps, then, categories, sub,
Nearest to during: before, antiproton, late, throughout, beginning, lamanites, carangi, prior,
Nearest to about: regarding, shortening, approximately, discography, depleted, rugians, thacker, gssps,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, three, eight, one, four, nine, two, seven,
Nearest to some: many, these, several, most, various, such, certain, other,
Nearest to however: but, although, though, because, that, while, since, despite,
Nearest to war: ii, wwii, wars, outbreak, campaign, manassas, nationalists, veterans,
Nearest to experience: maybe, isopropanol, hellene, auditory, lawyer, bearable, incoherent, subjective,
Nearest to magazine: macmullen, coil, handguns, cariou, magazines, graven, fetus, skeptics,
Nearest to joseph: smith, higginbotham, sarah, moglen, phylum, votive, retrofitted, ssen,
Nearest to universe: fictional, gallen, constellations, disconnected, lvares, monkey, harrington, prot,
Nearest to defense: guard, ignition, comix, rodentia, attrition, pitman, parody, iea,
Nearest to square: knee, rim, dock, upland, interesting, shaper, wadis, triangle,
Nearest to egypt: jericho, egyptian, petrel, countenance, colonised, andrews, receded, circa,
Nearest to smith: garner, geddes, joseph, mcm, thera, sisley, purr, huck,
Epoch 6/10 Iteration: 61100 Avg. Training loss: 4.2383 0.2726 sec/batch
Epoch 6/10 Iteration: 61200 Avg. Training loss: 4.1246 0.2706 sec/batch
Epoch 6/10 Iteration: 61300 Avg. Training loss: 4.3056 0.2719 sec/batch
Epoch 6/10 Iteration: 61400 Avg. Training loss: 4.2697 0.2782 sec/batch
Epoch 6/10 Iteration: 61500 Avg. Training loss: 4.2870 0.2701 sec/batch
Epoch 6/10 Iteration: 61600 Avg. Training loss: 4.2855 0.2694 sec/batch
Epoch 6/10 Iteration: 61700 Avg. Training loss: 4.3030 0.2691 sec/batch
Epoch 6/10 Iteration: 61800 Avg. Training loss: 4.1805 0.2715 sec/batch
Epoch 6/10 Iteration: 61900 Avg. Training loss: 4.3117 0.2804 sec/batch
Epoch 6/10 Iteration: 62000 Avg. Training loss: 4.3354 0.2735 sec/batch
Nearest to into: back, further, distancing, furrows, destry, then, divisions, sub,
Nearest to during: before, late, antiproton, beginning, first, throughout, prior, early,
Nearest to about: regarding, shortening, approximately, depleted, thacker, rugians, discography, gssps,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, one, three, eight, four, nine, two, seven,
Nearest to some: many, these, several, most, such, various, certain, other,
Nearest to however: but, although, though, because, that, while, since, despite,
Nearest to war: ii, wwii, campaign, outbreak, manassas, wars, veterans, nationalists,
Nearest to experience: maybe, isopropanol, hellene, gz, gabrieli, subjective, auditory, subdivide,
Nearest to magazine: macmullen, coil, magazines, skeptics, fetus, handguns, graven, cariou,
Nearest to joseph: smith, higginbotham, moglen, scarecrow, sarah, phylum, ssen, votive,
Nearest to universe: fictional, gallen, disconnected, constellations, lvares, monkey, harrington, violet,
Nearest to defense: guard, ignition, attrition, comix, pitman, rodentia, parody, caustic,
Nearest to square: knee, dock, rim, upland, wadis, interesting, differentiability, specimen,
Nearest to egypt: jericho, egyptian, petrel, colonised, countenance, andrews, receded, circa,
Nearest to smith: joseph, geddes, garner, thera, mcm, sisley, lila, dildo,
Epoch 6/10 Iteration: 62100 Avg. Training loss: 4.3086 0.2722 sec/batch
Epoch 6/10 Iteration: 62200 Avg. Training loss: 4.3208 0.2698 sec/batch
Epoch 6/10 Iteration: 62300 Avg. Training loss: 4.3098 0.2709 sec/batch
Epoch 6/10 Iteration: 62400 Avg. Training loss: 4.2929 0.2769 sec/batch
Epoch 6/10 Iteration: 62500 Avg. Training loss: 4.3441 0.2742 sec/batch
Epoch 6/10 Iteration: 62600 Avg. Training loss: 4.2612 0.2772 sec/batch
Epoch 6/10 Iteration: 62700 Avg. Training loss: 4.3057 0.2741 sec/batch
Epoch 6/10 Iteration: 62800 Avg. Training loss: 4.2368 0.2718 sec/batch
Epoch 6/10 Iteration: 62900 Avg. Training loss: 4.3152 0.2763 sec/batch
Epoch 6/10 Iteration: 63000 Avg. Training loss: 4.3390 0.2871 sec/batch
Nearest to into: back, destry, further, then, mbps, distancing, sub, onto,
Nearest to during: late, before, antiproton, beginning, early, lamanites, first, throughout,
Nearest to about: regarding, shortening, approximately, thacker, depleted, rugians, discography, zero,
Nearest to five: four, one, six, two, zero, three, seven, eight,
Nearest to six: five, one, eight, three, four, nine, two, seven,
Nearest to some: many, several, these, most, various, certain, such, other,
Nearest to however: but, although, though, because, that, while, since, quite,
Nearest to war: ii, wwii, manassas, wars, campaign, outbreak, veterans, nationalists,
Nearest to experience: maybe, gz, auditory, isopropanol, gabrieli, hellene, subjective, moles,
Nearest to magazine: macmullen, coil, magazines, fetus, cariou, handguns, skeptics, graven,
Nearest to joseph: smith, higginbotham, moglen, sarah, scarecrow, unemployment, ssen, phylum,
Nearest to universe: gallen, fictional, monkey, disconnected, constellations, lvares, prot, nolan,
Nearest to defense: guard, comix, pitman, attrition, ignition, rodentia, parody, caustic,
Nearest to square: knee, dock, rim, upland, differentiability, interesting, burton, stochastic,
Nearest to egypt: jericho, egyptian, andrews, petrel, receded, colonised, countenance, claw,
Nearest to smith: joseph, garner, geddes, mcm, thera, sisley, lila, purr,
Epoch 6/10 Iteration: 63100 Avg. Training loss: 4.3309 0.2832 sec/batch
Epoch 6/10 Iteration: 63200 Avg. Training loss: 4.3550 0.2793 sec/batch
Epoch 6/10 Iteration: 63300 Avg. Training loss: 4.3003 0.2766 sec/batch
Epoch 6/10 Iteration: 63400 Avg. Training loss: 4.2132 0.2729 sec/batch
Epoch 6/10 Iteration: 63500 Avg. Training loss: 4.3159 0.2715 sec/batch
Epoch 6/10 Iteration: 63600 Avg. Training loss: 4.2498 0.2727 sec/batch
Epoch 6/10 Iteration: 63700 Avg. Training loss: 4.3311 0.2732 sec/batch
Epoch 6/10 Iteration: 63800 Avg. Training loss: 4.3006 0.2786 sec/batch
Epoch 6/10 Iteration: 63900 Avg. Training loss: 4.3135 0.2646 sec/batch
Epoch 6/10 Iteration: 64000 Avg. Training loss: 4.3214 0.2671 sec/batch
Nearest to into: back, distancing, further, destry, categories, sub, then, mbps,
Nearest to during: before, antiproton, late, lamanites, early, beginning, end, after,
Nearest to about: regarding, shortening, approximately, discography, thacker, rugians, zero, depleted,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, eight, one, three, nine, four, seven, two,
Nearest to some: many, these, several, most, such, various, certain, other,
Nearest to however: but, although, though, because, that, while, since, nevertheless,
Nearest to war: ii, wwii, outbreak, campaign, manassas, wars, veterans, world,
Nearest to experience: maybe, gz, subjective, auditory, isopropanol, moles, sensory, hellene,
Nearest to magazine: macmullen, magazines, coil, newspaper, skeptics, graven, fetus, cariou,
Nearest to joseph: higginbotham, smith, moglen, sarah, scarecrow, unemployment, phylum, publication,
Nearest to universe: fictional, gallen, constellations, disconnected, monkey, lvares, prot, mainstream,
Nearest to defense: guard, pitman, comix, ignition, attrition, rodentia, parody, ecclesiastic,
Nearest to square: knee, dock, upland, rim, differentiability, wadis, stochastic, shaper,
Nearest to egypt: jericho, egyptian, petrel, andrews, receded, unspecified, garuda, claw,
Nearest to smith: garner, geddes, joseph, mcm, thera, sisley, lila, purr,
Epoch 6/10 Iteration: 64100 Avg. Training loss: 4.2333 0.2661 sec/batch
Epoch 6/10 Iteration: 64200 Avg. Training loss: 4.3093 0.2660 sec/batch
Epoch 6/10 Iteration: 64300 Avg. Training loss: 4.2632 0.2670 sec/batch
Epoch 6/10 Iteration: 64400 Avg. Training loss: 4.3033 0.2664 sec/batch
Epoch 6/10 Iteration: 64500 Avg. Training loss: 4.2256 0.2673 sec/batch
Epoch 6/10 Iteration: 64600 Avg. Training loss: 4.2971 0.2653 sec/batch
Epoch 6/10 Iteration: 64700 Avg. Training loss: 4.3382 0.2726 sec/batch
Epoch 6/10 Iteration: 64800 Avg. Training loss: 4.2694 0.2633 sec/batch
Epoch 6/10 Iteration: 64900 Avg. Training loss: 4.2757 0.2669 sec/batch
Epoch 6/10 Iteration: 65000 Avg. Training loss: 4.2432 0.2653 sec/batch
Nearest to into: back, distancing, further, divided, categories, divisions, destry, sub,
Nearest to during: antiproton, before, late, lamanites, throughout, after, early, following,
Nearest to about: regarding, approximately, main, shortening, zero, rugians, thacker, discography,
Nearest to five: four, one, six, two, zero, three, seven, eight,
Nearest to six: five, one, eight, three, four, nine, two, seven,
Nearest to some: many, these, several, such, most, various, certain, other,
Nearest to however: but, although, though, because, that, while, since, despite,
Nearest to war: ii, wwii, outbreak, manassas, wars, campaign, world, during,
Nearest to experience: maybe, subjective, hellene, gabrieli, isopropanol, gz, auditory, moles,
Nearest to magazine: macmullen, magazines, coil, handguns, fetus, graven, skeptics, cariou,
Nearest to joseph: higginbotham, smith, sarah, moglen, unemployment, scarecrow, phylum, ssen,
Nearest to universe: fictional, gallen, constellations, disconnected, lvares, mainstream, monkey, prot,
Nearest to defense: guard, pitman, comix, ignition, ecclesiastic, command, parody, attrition,
Nearest to square: knee, dock, upland, wadis, stochastic, rim, shaper, burton,
Nearest to egypt: jericho, egyptian, petrel, andrews, receded, circa, claw, garuda,
Nearest to smith: geddes, garner, joseph, mcm, thera, sisley, lila, spartacus,
Epoch 6/10 Iteration: 65100 Avg. Training loss: 4.3185 0.2656 sec/batch
Epoch 6/10 Iteration: 65200 Avg. Training loss: 4.3219 0.2697 sec/batch
Epoch 6/10 Iteration: 65300 Avg. Training loss: 4.3137 0.2677 sec/batch
Epoch 6/10 Iteration: 65400 Avg. Training loss: 4.3678 0.2738 sec/batch
Epoch 6/10 Iteration: 65500 Avg. Training loss: 4.3119 0.2742 sec/batch
Epoch 6/10 Iteration: 65600 Avg. Training loss: 4.2403 0.2676 sec/batch
Epoch 6/10 Iteration: 65700 Avg. Training loss: 4.3330 0.2667 sec/batch
Epoch 6/10 Iteration: 65800 Avg. Training loss: 4.3223 0.2662 sec/batch
Epoch 6/10 Iteration: 65900 Avg. Training loss: 4.2951 0.2630 sec/batch
Epoch 6/10 Iteration: 66000 Avg. Training loss: 4.3299 0.2686 sec/batch
Nearest to into: back, further, then, divided, distancing, categories, eventually, destry,
Nearest to during: before, late, antiproton, lamanites, beginning, after, in, prior,
Nearest to about: regarding, approximately, main, shortening, depleted, zero, gssps, thacker,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, one, eight, four, three, nine, seven, two,
Nearest to some: many, these, several, such, most, various, other, certain,
Nearest to however: but, although, though, because, that, while, since, despite,
Nearest to war: ii, outbreak, campaign, wwii, wars, manassas, veterans, world,
Nearest to experience: maybe, subjective, incoherent, auditory, gabrieli, bearable, isopropanol, hellene,
Nearest to magazine: macmullen, magazines, coil, fetus, graven, cariou, handguns, newspaper,
Nearest to joseph: sarah, higginbotham, smith, unemployment, moglen, phylum, ssen, mashal,
Nearest to universe: gallen, fictional, constellations, lvares, disconnected, nolan, monkey, violet,
Nearest to defense: guard, pitman, comix, ignition, ecclesiastic, attrition, parody, rodentia,
Nearest to square: knee, dock, upland, wadis, burton, rim, stochastic, shaper,
Nearest to egypt: jericho, egyptian, circa, andrews, petrel, receded, claw, pianist,
Nearest to smith: garner, geddes, mcm, joseph, thera, sisley, lila, huck,
Epoch 6/10 Iteration: 66100 Avg. Training loss: 4.3283 0.2727 sec/batch
Epoch 6/10 Iteration: 66200 Avg. Training loss: 4.3044 0.2689 sec/batch
Epoch 6/10 Iteration: 66300 Avg. Training loss: 4.3351 0.2700 sec/batch
Epoch 6/10 Iteration: 66400 Avg. Training loss: 4.2771 0.2782 sec/batch
Epoch 6/10 Iteration: 66500 Avg. Training loss: 4.2163 0.2686 sec/batch
Epoch 6/10 Iteration: 66600 Avg. Training loss: 4.2829 0.2658 sec/batch
Epoch 6/10 Iteration: 66700 Avg. Training loss: 4.2697 0.2766 sec/batch
Epoch 6/10 Iteration: 66800 Avg. Training loss: 4.3020 0.2685 sec/batch
Epoch 6/10 Iteration: 66900 Avg. Training loss: 4.2643 0.2731 sec/batch
Epoch 6/10 Iteration: 67000 Avg. Training loss: 4.3618 0.2691 sec/batch
Nearest to into: back, categories, further, divided, throats, then, sub, onto,
Nearest to during: late, antiproton, before, early, lamanites, throughout, after, beginning,
Nearest to about: regarding, approximately, main, shortening, thacker, depleted, zero, rugians,
Nearest to five: four, one, six, two, zero, three, seven, eight,
Nearest to six: five, eight, one, four, three, nine, two, seven,
Nearest to some: many, these, several, such, most, certain, other, various,
Nearest to however: although, but, though, because, that, while, since, despite,
Nearest to war: ii, outbreak, wwii, wars, campaign, manassas, nationalists, world,
Nearest to experience: maybe, subjective, auditory, isopropanol, gabrieli, bearable, coincident, sensory,
Nearest to magazine: macmullen, magazines, fetus, coil, graven, skeptics, newspaper, handguns,
Nearest to joseph: sarah, higginbotham, smith, unemployment, phylum, moglen, scarecrow, ssen,
Nearest to universe: gallen, fictional, constellations, lvares, disconnected, mainstream, nolan, violet,
Nearest to defense: guard, pitman, ignition, rodentia, comix, attrition, iea, parody,
Nearest to square: knee, upland, dock, interesting, rim, wadis, stochastic, shaper,
Nearest to egypt: jericho, egyptian, petrel, circa, receded, andrews, garuda, claw,
Nearest to smith: geddes, garner, joseph, mcm, sisley, thera, dildo, huck,
Epoch 6/10 Iteration: 67100 Avg. Training loss: 4.1328 0.2706 sec/batch
Epoch 6/10 Iteration: 67200 Avg. Training loss: 4.0667 0.2653 sec/batch
Epoch 6/10 Iteration: 67300 Avg. Training loss: 4.1598 0.2696 sec/batch
Epoch 6/10 Iteration: 67400 Avg. Training loss: 4.2423 0.2693 sec/batch
Epoch 6/10 Iteration: 67500 Avg. Training loss: 4.3365 0.2703 sec/batch
Epoch 6/10 Iteration: 67600 Avg. Training loss: 4.3503 0.2707 sec/batch
Epoch 6/10 Iteration: 67700 Avg. Training loss: 4.3034 0.2690 sec/batch
Epoch 6/10 Iteration: 67800 Avg. Training loss: 4.3998 0.2679 sec/batch
Epoch 6/10 Iteration: 67900 Avg. Training loss: 4.3783 0.2697 sec/batch
Epoch 6/10 Iteration: 68000 Avg. Training loss: 4.3491 0.2705 sec/batch
Nearest to into: back, categories, onto, through, distancing, destry, further, then,
Nearest to during: before, antiproton, late, after, beginning, early, throughout, lamanites,
Nearest to about: regarding, approximately, thacker, main, shortening, depleted, discography, jail,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, eight, one, three, four, nine, seven, two,
Nearest to some: many, several, these, such, most, other, various, certain,
Nearest to however: but, although, though, because, while, that, since, despite,
Nearest to war: ii, wwii, outbreak, veterans, campaign, manassas, nationalists, wars,
Nearest to experience: maybe, subjective, auditory, experiences, bearable, incoherent, bloodless, varnish,
Nearest to magazine: macmullen, magazines, coil, fetus, graven, skeptics, cariou, handguns,
Nearest to joseph: sarah, higginbotham, phylum, smith, unemployment, ssen, moglen, votive,
Nearest to universe: fictional, gallen, constellations, lvares, comics, disconnected, violet, nolan,
Nearest to defense: guard, pitman, comix, rodentia, ecclesiastic, ignition, parody, attrition,
Nearest to square: knee, dock, upland, interesting, shaper, mumy, wadis, burton,
Nearest to egypt: jericho, egyptian, petrel, andrews, receded, circa, claw, countenance,
Nearest to smith: mcm, garner, geddes, joseph, thera, sisley, avoirdupois, huck,
Epoch 6/10 Iteration: 68100 Avg. Training loss: 4.3202 0.2756 sec/batch
Epoch 6/10 Iteration: 68200 Avg. Training loss: 4.1553 0.2679 sec/batch
Epoch 6/10 Iteration: 68300 Avg. Training loss: 4.3382 0.2663 sec/batch
Epoch 6/10 Iteration: 68400 Avg. Training loss: 4.2926 0.2695 sec/batch
Epoch 6/10 Iteration: 68500 Avg. Training loss: 4.3417 0.2762 sec/batch
Epoch 6/10 Iteration: 68600 Avg. Training loss: 4.2715 0.2662 sec/batch
Epoch 6/10 Iteration: 68700 Avg. Training loss: 4.3601 0.2682 sec/batch
Epoch 6/10 Iteration: 68800 Avg. Training loss: 4.2715 0.2684 sec/batch
Epoch 6/10 Iteration: 68900 Avg. Training loss: 4.2143 0.2680 sec/batch
Epoch 6/10 Iteration: 69000 Avg. Training loss: 4.1076 0.2667 sec/batch
Nearest to into: back, categories, through, divided, further, onto, sub, distancing,
Nearest to during: late, before, early, after, beginning, antiproton, throughout, following,
Nearest to about: approximately, regarding, zero, main, five, estimated, depleted, covering,
Nearest to five: four, one, two, zero, six, three, seven, eight,
Nearest to six: five, one, four, three, eight, nine, two, seven,
Nearest to some: many, several, these, most, such, other, various, certain,
Nearest to however: but, although, though, while, because, that, since, despite,
Nearest to war: ii, outbreak, wwii, campaign, veterans, wars, manassas, conflict,
Nearest to experience: subjective, maybe, auditory, gz, isopropanol, bloodless, experiences, varnish,
Nearest to magazine: macmullen, magazines, coil, fetus, skeptics, graven, cariou, recoil,
Nearest to joseph: higginbotham, sarah, phylum, smith, moglen, unemployment, ssen, votive,
Nearest to universe: fictional, gallen, constellations, lvares, violet, disconnected, comics, monkey,
Nearest to defense: guard, comix, ignition, pitman, command, ecclesiastic, attrition, parody,
Nearest to square: knee, dock, upland, interesting, rim, wadis, triangle, shaper,
Nearest to egypt: egyptian, jericho, petrel, andrews, receded, garuda, claw, countenance,
Nearest to smith: mcm, garner, geddes, joseph, sisley, thera, huck, avoirdupois,
Epoch 6/10 Iteration: 69100 Avg. Training loss: 4.3197 0.2732 sec/batch
Epoch 6/10 Iteration: 69200 Avg. Training loss: 4.2615 0.2711 sec/batch
Epoch 6/10 Iteration: 69300 Avg. Training loss: 4.2483 0.2664 sec/batch
Epoch 6/10 Iteration: 69400 Avg. Training loss: 4.3313 0.2702 sec/batch
Epoch 6/10 Iteration: 69500 Avg. Training loss: 4.3009 0.2688 sec/batch
Epoch 6/10 Iteration: 69600 Avg. Training loss: 4.2727 0.2691 sec/batch
Epoch 6/10 Iteration: 69700 Avg. Training loss: 4.2898 0.2677 sec/batch
Epoch 6/10 Iteration: 69800 Avg. Training loss: 4.3141 0.2846 sec/batch
Epoch 6/10 Iteration: 69900 Avg. Training loss: 4.3708 0.2646 sec/batch
Epoch 7/10 Iteration: 70000 Avg. Training loss: 4.3381 0.1066 sec/batch
Nearest to into: back, categories, onto, through, further, destry, distancing, eastbourne,
Nearest to during: before, antiproton, late, early, beginning, after, throughout, first,
Nearest to about: regarding, approximately, thacker, discography, shortening, depleted, rugians, iqs,
Nearest to five: four, one, six, two, zero, three, seven, eight,
Nearest to six: five, eight, one, four, three, nine, seven, two,
Nearest to some: many, several, these, such, various, most, other, certain,
Nearest to however: but, although, though, because, while, that, since, despite,
Nearest to war: ii, outbreak, wwii, veterans, campaign, wars, manassas, branson,
Nearest to experience: maybe, subjective, auditory, gz, isopropanol, experiences, bearable, bloodless,
Nearest to magazine: coil, macmullen, magazines, fetus, skeptics, graven, recoil, cariou,
Nearest to joseph: higginbotham, sarah, smith, unemployment, phylum, moglen, ssen, votive,
Nearest to universe: fictional, gallen, constellations, disconnected, violet, comics, monkey, nolan,
Nearest to defense: guard, comix, ignition, pitman, rodentia, erudite, attrition, ecclesiastic,
Nearest to square: knee, dock, stochastic, upland, gonzaga, interesting, rim, shaper,
Nearest to egypt: egyptian, jericho, andrews, petrel, receded, pianist, countenance, claw,
Nearest to smith: mcm, geddes, garner, joseph, sisley, huck, thera, lila,
Epoch 7/10 Iteration: 70100 Avg. Training loss: 4.3117 0.2710 sec/batch
Epoch 7/10 Iteration: 70200 Avg. Training loss: 4.2670 0.2689 sec/batch
Epoch 7/10 Iteration: 70300 Avg. Training loss: 4.3133 0.2719 sec/batch
Epoch 7/10 Iteration: 70400 Avg. Training loss: 4.2629 0.2692 sec/batch
Epoch 7/10 Iteration: 70500 Avg. Training loss: 4.2871 0.2663 sec/batch
Epoch 7/10 Iteration: 70600 Avg. Training loss: 4.2848 0.2652 sec/batch
Epoch 7/10 Iteration: 70700 Avg. Training loss: 4.2255 0.2661 sec/batch
Epoch 7/10 Iteration: 70800 Avg. Training loss: 4.2859 0.2683 sec/batch
Epoch 7/10 Iteration: 70900 Avg. Training loss: 4.3150 0.2669 sec/batch
Epoch 7/10 Iteration: 71000 Avg. Training loss: 4.3021 0.2680 sec/batch
Nearest to into: back, onto, then, categories, destry, further, through, separate,
Nearest to during: before, antiproton, late, early, throughout, after, beginning, first,
Nearest to about: regarding, approximately, main, shortening, discography, covering, thacker, zero,
Nearest to five: one, four, six, two, zero, three, seven, eight,
Nearest to six: five, one, eight, four, three, nine, seven, two,
Nearest to some: many, several, these, certain, such, various, most, other,
Nearest to however: but, although, though, because, that, while, since, only,
Nearest to war: ii, outbreak, wwii, veterans, manassas, campaign, wars, branson,
Nearest to experience: maybe, subjective, auditory, gz, bearable, isopropanol, sensory, experiences,
Nearest to magazine: macmullen, magazines, coil, fetus, graven, recoil, lps, cariou,
Nearest to joseph: higginbotham, sarah, phylum, smith, unemployment, moglen, ssen, votive,
Nearest to universe: fictional, gallen, constellations, lvares, comics, monkey, cbe, nolan,
Nearest to defense: guard, comix, ignition, attrition, pitman, rodentia, iea, caustic,
Nearest to square: knee, dock, upland, rim, triangle, specimen, stochastic, gonzaga,
Nearest to egypt: egyptian, jericho, receded, andrews, claw, circa, petrel, countenance,
Nearest to smith: mcm, joseph, geddes, garner, thera, sisley, avoirdupois, huck,
Epoch 7/10 Iteration: 71100 Avg. Training loss: 4.2935 0.2731 sec/batch
Epoch 7/10 Iteration: 71200 Avg. Training loss: 4.3236 0.2672 sec/batch
Epoch 7/10 Iteration: 71300 Avg. Training loss: 4.3177 0.2702 sec/batch
Epoch 7/10 Iteration: 71400 Avg. Training loss: 4.3078 0.2742 sec/batch
Epoch 7/10 Iteration: 71500 Avg. Training loss: 4.2981 0.2698 sec/batch
Epoch 7/10 Iteration: 71600 Avg. Training loss: 4.2905 0.2681 sec/batch
Epoch 7/10 Iteration: 71700 Avg. Training loss: 4.1475 0.2669 sec/batch
Epoch 7/10 Iteration: 71800 Avg. Training loss: 4.3231 0.2704 sec/batch
Epoch 7/10 Iteration: 71900 Avg. Training loss: 4.3484 0.2792 sec/batch
Epoch 7/10 Iteration: 72000 Avg. Training loss: 4.3058 0.2699 sec/batch
Nearest to into: back, categories, onto, destry, separate, eastbourne, through, sub,
Nearest to during: before, throughout, antiproton, late, early, first, after, beginning,
Nearest to about: regarding, approximately, discography, zero, thacker, shortening, covering, depleted,
Nearest to five: four, one, six, two, zero, three, seven, eight,
Nearest to six: five, eight, one, three, four, seven, nine, two,
Nearest to some: many, these, several, certain, most, various, such, other,
Nearest to however: but, although, though, because, while, that, very, despite,
Nearest to war: ii, wwii, outbreak, veterans, manassas, wars, campaign, world,
Nearest to experience: maybe, subjective, auditory, gz, isopropanol, bloodless, sensory, experiences,
Nearest to magazine: magazines, macmullen, coil, fetus, skeptics, recoil, cariou, graven,
Nearest to joseph: smith, sarah, higginbotham, moglen, phylum, votive, unemployment, ssen,
Nearest to universe: gallen, fictional, lvares, monkey, constellations, big, comics, nolan,
Nearest to defense: guard, comix, ignition, pitman, attrition, rodentia, command, parody,
Nearest to square: knee, dock, upland, specimen, triangle, rim, wadis, gonzaga,
Nearest to egypt: egyptian, jericho, andrews, receded, pianist, claw, circa, petrel,
Nearest to smith: joseph, mcm, geddes, garner, thera, sisley, huck, lila,
Epoch 7/10 Iteration: 72100 Avg. Training loss: 4.3221 0.2685 sec/batch
Epoch 7/10 Iteration: 72200 Avg. Training loss: 4.3137 0.2687 sec/batch
Epoch 7/10 Iteration: 72300 Avg. Training loss: 4.3114 0.2686 sec/batch
Epoch 7/10 Iteration: 72400 Avg. Training loss: 4.3400 0.2680 sec/batch
Epoch 7/10 Iteration: 72500 Avg. Training loss: 4.3439 0.2681 sec/batch
Epoch 7/10 Iteration: 72600 Avg. Training loss: 4.2809 0.2696 sec/batch
Epoch 7/10 Iteration: 72700 Avg. Training loss: 4.2670 0.2688 sec/batch
Epoch 7/10 Iteration: 72800 Avg. Training loss: 4.1448 0.2700 sec/batch
Epoch 7/10 Iteration: 72900 Avg. Training loss: 4.1793 0.2709 sec/batch
Epoch 7/10 Iteration: 73000 Avg. Training loss: 4.3013 0.2711 sec/batch
Nearest to into: back, divided, sub, through, categories, divisions, then, distancing,
Nearest to during: late, before, throughout, early, antiproton, beginning, after, lamanites,
Nearest to about: regarding, approximately, zero, depleted, thacker, discography, main, rugians,
Nearest to five: four, one, six, two, zero, three, seven, eight,
Nearest to six: five, one, four, eight, three, nine, seven, two,
Nearest to some: many, these, several, certain, various, such, most, other,
Nearest to however: although, but, though, because, while, that, since, despite,
Nearest to war: ii, wwii, outbreak, veterans, campaign, manassas, wars, nationalists,
Nearest to experience: maybe, subjective, auditory, isopropanol, gz, sensory, our, bearable,
Nearest to magazine: macmullen, magazines, coil, recoil, fetus, skeptics, cariou, graven,
Nearest to joseph: smith, sarah, higginbotham, phylum, unemployment, ssen, moglen, votive,
Nearest to universe: gallen, fictional, lvares, constellations, monkey, terrific, cbe, ok,
Nearest to defense: guard, comix, ignition, rodentia, attrition, pitman, ecclesiastic, command,
Nearest to square: knee, dock, upland, rim, wadis, gonzaga, triangle, stochastic,
Nearest to egypt: egyptian, jericho, circa, andrews, petrel, pianist, receded, claw,
Nearest to smith: mcm, joseph, geddes, garner, thera, sisley, huck, dildo,
Epoch 7/10 Iteration: 73100 Avg. Training loss: 4.2610 0.2742 sec/batch
Epoch 7/10 Iteration: 73200 Avg. Training loss: 4.2796 0.2755 sec/batch
Epoch 7/10 Iteration: 73300 Avg. Training loss: 4.2703 0.2684 sec/batch
Epoch 7/10 Iteration: 73400 Avg. Training loss: 4.2665 0.2674 sec/batch
Epoch 7/10 Iteration: 73500 Avg. Training loss: 4.1961 0.2680 sec/batch
Epoch 7/10 Iteration: 73600 Avg. Training loss: 4.2915 0.2741 sec/batch
Epoch 7/10 Iteration: 73700 Avg. Training loss: 4.3061 0.2655 sec/batch
Epoch 7/10 Iteration: 73800 Avg. Training loss: 4.2937 0.2683 sec/batch
Epoch 7/10 Iteration: 73900 Avg. Training loss: 4.2988 0.2681 sec/batch
Epoch 7/10 Iteration: 74000 Avg. Training loss: 4.3018 0.2704 sec/batch
Nearest to into: back, sub, categories, then, through, further, distancing, destry,
Nearest to during: before, late, throughout, beginning, early, antiproton, lamanites, first,
Nearest to about: regarding, shortening, approximately, thacker, discography, gssps, rugians, depleted,
Nearest to five: four, one, six, two, zero, three, seven, eight,
Nearest to six: five, one, eight, four, three, nine, seven, two,
Nearest to some: many, several, these, certain, various, most, such, other,
Nearest to however: but, although, though, because, while, that, since, despite,
Nearest to war: ii, wwii, outbreak, veterans, wars, manassas, campaign, nationalists,
Nearest to experience: maybe, subjective, isopropanol, gz, auditory, experiences, bearable, incoherent,
Nearest to magazine: magazines, coil, macmullen, recoil, fetus, cariou, skeptics, handguns,
Nearest to joseph: smith, sarah, phylum, ssen, higginbotham, moglen, unemployment, publication,
Nearest to universe: gallen, fictional, lvares, constellations, cbe, big, terrific, monkey,
Nearest to defense: guard, comix, pitman, ecclesiastic, attrition, ignition, rodentia, caustic,
Nearest to square: knee, dock, upland, rim, interesting, triangle, stochastic, wadis,
Nearest to egypt: egyptian, jericho, andrews, circa, claw, pianist, receded, petrel,
Nearest to smith: joseph, mcm, geddes, garner, thera, avoirdupois, huck, sisley,
Epoch 7/10 Iteration: 74100 Avg. Training loss: 4.3259 0.2669 sec/batch
Epoch 7/10 Iteration: 74200 Avg. Training loss: 4.3200 0.2650 sec/batch
Epoch 7/10 Iteration: 74300 Avg. Training loss: 4.2434 0.2697 sec/batch
Epoch 7/10 Iteration: 74400 Avg. Training loss: 4.2993 0.2661 sec/batch
Epoch 7/10 Iteration: 74500 Avg. Training loss: 4.2686 0.2668 sec/batch
Epoch 7/10 Iteration: 74600 Avg. Training loss: 4.3139 0.2660 sec/batch
Epoch 7/10 Iteration: 74700 Avg. Training loss: 4.3286 0.2657 sec/batch
Epoch 7/10 Iteration: 74800 Avg. Training loss: 4.3323 0.2665 sec/batch
Epoch 7/10 Iteration: 74900 Avg. Training loss: 4.3282 0.2692 sec/batch
Epoch 7/10 Iteration: 75000 Avg. Training loss: 4.2495 0.2672 sec/batch
Nearest to into: back, sub, categories, distancing, then, further, onto, divided,
Nearest to during: late, before, antiproton, throughout, lamanites, beginning, early, after,
Nearest to about: regarding, approximately, shortening, thacker, main, rugians, gssps, depleted,
Nearest to five: four, six, one, two, zero, three, seven, eight,
Nearest to six: five, one, eight, four, three, nine, seven, two,
Nearest to some: many, several, these, certain, various, most, such, other,
Nearest to however: but, although, though, because, that, while, since, despite,
Nearest to war: ii, wwii, outbreak, campaign, veterans, manassas, branson, wars,
Nearest to experience: maybe, isopropanol, auditory, subjective, experiences, bearable, gz, incoherent,
Nearest to magazine: magazines, macmullen, coil, cariou, fetus, recoil, graven, skeptics,
Nearest to joseph: smith, sarah, higginbotham, moglen, phylum, ssen, unemployment, shop,
Nearest to universe: gallen, fictional, constellations, monkey, lvares, cbe, terrific, big,
Nearest to defense: guard, comix, pitman, ignition, ecclesiastic, rodentia, attrition, parody,
Nearest to square: knee, dock, upland, gonzaga, triangle, stochastic, interesting, rim,
Nearest to egypt: egyptian, jericho, andrews, pianist, petrel, countenance, receded, claw,
Nearest to smith: joseph, mcm, geddes, garner, avoirdupois, huck, filename, thera,
Epoch 7/10 Iteration: 75100 Avg. Training loss: 4.2278 0.2713 sec/batch
Epoch 7/10 Iteration: 75200 Avg. Training loss: 4.2401 0.2715 sec/batch
Epoch 7/10 Iteration: 75300 Avg. Training loss: 4.2839 0.2661 sec/batch
Epoch 7/10 Iteration: 75400 Avg. Training loss: 4.3060 0.2667 sec/batch
Epoch 7/10 Iteration: 75500 Avg. Training loss: 4.2785 0.2689 sec/batch
Epoch 7/10 Iteration: 75600 Avg. Training loss: 4.3198 0.2711 sec/batch
Epoch 7/10 Iteration: 75700 Avg. Training loss: 4.2986 0.2693 sec/batch
Epoch 7/10 Iteration: 75800 Avg. Training loss: 4.2404 0.2673 sec/batch
Epoch 7/10 Iteration: 75900 Avg. Training loss: 4.3109 0.2719 sec/batch
Epoch 7/10 Iteration: 76000 Avg. Training loss: 4.2442 0.2692 sec/batch
Nearest to into: back, sub, distancing, categories, onto, destry, further, divided,
Nearest to during: late, antiproton, before, throughout, after, beginning, following, prior,
Nearest to about: regarding, main, rugians, shortening, thacker, approximately, depleted, discography,
Nearest to five: four, one, six, two, seven, zero, three, eight,
Nearest to six: five, one, eight, four, seven, nine, three, two,
Nearest to some: many, several, these, most, various, certain, such, other,
Nearest to however: but, although, though, because, that, while, since, despite,
Nearest to war: ii, outbreak, wwii, campaign, veterans, branson, manassas, nationalists,
Nearest to experience: maybe, subjective, isopropanol, auditory, gz, experiences, bearable, incoherent,
Nearest to magazine: magazines, macmullen, coil, cariou, graven, newspaper, fetus, recoil,
Nearest to joseph: smith, sarah, higginbotham, moglen, phylum, unemployment, ssen, shop,
Nearest to universe: gallen, lvares, constellations, fictional, exposure, monkey, unicef, cbe,
Nearest to defense: guard, comix, pitman, ecclesiastic, ignition, rodentia, parody, caustic,
Nearest to square: knee, dock, gonzaga, upland, shaper, burton, wadis, triangle,
Nearest to egypt: egyptian, jericho, andrews, receded, claw, petrel, pianist, unspecified,
Nearest to smith: joseph, mcm, geddes, garner, thera, huck, avoirdupois, filename,
Epoch 7/10 Iteration: 76100 Avg. Training loss: 4.3168 0.2681 sec/batch
Epoch 7/10 Iteration: 76200 Avg. Training loss: 4.1634 0.2702 sec/batch
Epoch 7/10 Iteration: 76300 Avg. Training loss: 4.3198 0.2666 sec/batch
Epoch 7/10 Iteration: 76400 Avg. Training loss: 4.3198 0.2686 sec/batch
Epoch 7/10 Iteration: 76500 Avg. Training loss: 4.2457 0.2741 sec/batch
Epoch 7/10 Iteration: 76600 Avg. Training loss: 4.1680 0.2760 sec/batch
Epoch 7/10 Iteration: 76700 Avg. Training loss: 4.3202 0.2667 sec/batch
Epoch 7/10 Iteration: 76800 Avg. Training loss: 4.3201 0.2686 sec/batch
Epoch 7/10 Iteration: 76900 Avg. Training loss: 4.3019 0.2674 sec/batch
Epoch 7/10 Iteration: 77000 Avg. Training loss: 4.3182 0.2697 sec/batch
Nearest to into: back, further, sub, categories, distancing, then, onto, divided,
Nearest to during: before, antiproton, late, throughout, after, prior, following, lamanites,
Nearest to about: regarding, main, approximately, thacker, rugians, shortening, depleted, discography,
Nearest to five: four, six, one, two, zero, three, seven, eight,
Nearest to six: five, one, eight, four, three, seven, nine, two,
Nearest to some: many, several, these, most, various, certain, such, other,
Nearest to however: but, although, though, because, that, while, quite, despite,
Nearest to war: ii, outbreak, wwii, campaign, veterans, conflict, branson, manassas,
Nearest to experience: maybe, bearable, isopropanol, gz, incoherent, subjective, auditory, coincident,
Nearest to magazine: magazines, coil, macmullen, graven, cariou, fetus, newspaper, recoil,
Nearest to joseph: sarah, smith, phylum, higginbotham, unemployment, moglen, ssen, shop,
Nearest to universe: gallen, constellations, lvares, fictional, monkey, exposure, ok, unicef,
Nearest to defense: guard, comix, pitman, ignition, ecclesiastic, parody, rodentia, attrition,
Nearest to square: knee, dock, upland, gonzaga, wadis, shaper, triangle, interesting,
Nearest to egypt: egyptian, jericho, andrews, receded, claw, circa, petrel, pianist,
Nearest to smith: joseph, mcm, garner, geddes, thera, sisley, avoirdupois, dildo,
Epoch 7/10 Iteration: 77100 Avg. Training loss: 4.3673 0.2741 sec/batch
Epoch 7/10 Iteration: 77200 Avg. Training loss: 4.2864 0.2683 sec/batch
Epoch 7/10 Iteration: 77300 Avg. Training loss: 4.2339 0.2688 sec/batch
Epoch 7/10 Iteration: 77400 Avg. Training loss: 4.3289 0.2680 sec/batch
Epoch 7/10 Iteration: 77500 Avg. Training loss: 4.3134 0.2658 sec/batch
Epoch 7/10 Iteration: 77600 Avg. Training loss: 4.3078 0.2687 sec/batch
Epoch 7/10 Iteration: 77700 Avg. Training loss: 4.3058 0.2678 sec/batch
Epoch 7/10 Iteration: 77800 Avg. Training loss: 4.3291 0.2686 sec/batch
Epoch 7/10 Iteration: 77900 Avg. Training loss: 4.3158 0.2668 sec/batch
Epoch 7/10 Iteration: 78000 Avg. Training loss: 4.2875 0.2686 sec/batch
Nearest to into: back, distancing, further, categories, divided, sub, onto, then,
Nearest to during: late, antiproton, throughout, before, lamanites, prior, beginning, after,
Nearest to about: regarding, main, approximately, shortening, rugians, discography, thacker, depleted,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, eight, four, three, seven, nine, two,
Nearest to some: many, these, several, most, certain, various, such, other,
Nearest to however: but, although, though, because, while, that, since, despite,
Nearest to war: ii, outbreak, wwii, conflict, wars, campaign, manassas, branson,
Nearest to experience: maybe, subjective, experiences, incoherent, bearable, gz, isopropanol, coincident,
Nearest to magazine: magazines, coil, macmullen, cariou, fetus, recoil, newspaper, graven,
Nearest to joseph: sarah, smith, phylum, higginbotham, unemployment, ssen, moglen, shop,
Nearest to universe: gallen, fictional, constellations, monkey, lvares, big, unicef, terrific,
Nearest to defense: guard, comix, pitman, rodentia, ecclesiastic, ignition, parody, caustic,
Nearest to square: knee, upland, dock, gonzaga, kilmister, interesting, triangle, wadis,
Nearest to egypt: egyptian, jericho, andrews, receded, claw, circa, cuneiform, pianist,
Nearest to smith: joseph, mcm, garner, thera, geddes, avoirdupois, dildo, filename,
Epoch 7/10 Iteration: 78100 Avg. Training loss: 4.2207 0.2764 sec/batch
Epoch 7/10 Iteration: 78200 Avg. Training loss: 4.2597 0.2661 sec/batch
Epoch 7/10 Iteration: 78300 Avg. Training loss: 4.2697 0.2688 sec/batch
Epoch 7/10 Iteration: 78400 Avg. Training loss: 4.2749 0.2677 sec/batch
Epoch 7/10 Iteration: 78500 Avg. Training loss: 4.2818 0.2762 sec/batch
Epoch 7/10 Iteration: 78600 Avg. Training loss: 4.2563 0.2682 sec/batch
Epoch 7/10 Iteration: 78700 Avg. Training loss: 4.3108 0.2697 sec/batch
Epoch 7/10 Iteration: 78800 Avg. Training loss: 4.0003 0.2662 sec/batch
Epoch 7/10 Iteration: 78900 Avg. Training loss: 4.1668 0.2679 sec/batch
Epoch 7/10 Iteration: 79000 Avg. Training loss: 4.1728 0.2678 sec/batch
Nearest to into: back, categories, distancing, then, further, onto, sub, destry,
Nearest to during: before, late, antiproton, throughout, after, following, prior, early,
Nearest to about: regarding, main, approximately, thacker, told, depleted, discography, shortening,
Nearest to five: six, four, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, these, several, most, such, certain, various, other,
Nearest to however: but, although, though, because, that, while, nevertheless, since,
Nearest to war: ii, outbreak, campaign, manassas, wars, conflict, veterans, branson,
Nearest to experience: experiences, maybe, subjective, bearable, incoherent, disillusioned, gz, puerto,
Nearest to magazine: magazines, coil, macmullen, newspaper, cariou, issue, fetus, recoil,
Nearest to joseph: sarah, smith, unemployment, phylum, higginbotham, ssen, moglen, schlick,
Nearest to universe: gallen, fictional, constellations, monkey, lvares, dune, unicef, terrific,
Nearest to defense: guard, comix, ecclesiastic, pitman, rodentia, ignition, parody, attrition,
Nearest to square: knee, upland, dock, gonzaga, kilmister, triangle, shaper, wadis,
Nearest to egypt: egyptian, jericho, receded, andrews, claw, cuneiform, pianist, circa,
Nearest to smith: joseph, garner, mcm, geddes, thera, dildo, sisley, lila,
Epoch 7/10 Iteration: 79100 Avg. Training loss: 4.2832 0.2696 sec/batch
Epoch 7/10 Iteration: 79200 Avg. Training loss: 4.3054 0.2696 sec/batch
Epoch 7/10 Iteration: 79300 Avg. Training loss: 4.3213 0.2710 sec/batch
Epoch 7/10 Iteration: 79400 Avg. Training loss: 4.3523 0.2660 sec/batch
Epoch 7/10 Iteration: 79500 Avg. Training loss: 4.3717 0.2694 sec/batch
Epoch 7/10 Iteration: 79600 Avg. Training loss: 4.3649 0.2697 sec/batch
Epoch 7/10 Iteration: 79700 Avg. Training loss: 4.3248 0.2685 sec/batch
Epoch 7/10 Iteration: 79800 Avg. Training loss: 4.2462 0.2696 sec/batch
Epoch 7/10 Iteration: 79900 Avg. Training loss: 4.2452 0.2738 sec/batch
Epoch 7/10 Iteration: 80000 Avg. Training loss: 4.2952 0.2740 sec/batch
Nearest to into: back, distancing, categories, sub, divided, onto, divisions, further,
Nearest to during: late, before, antiproton, after, throughout, beginning, early, prior,
Nearest to about: regarding, approximately, thacker, main, jail, question, discography, what,
Nearest to five: four, six, one, two, zero, three, seven, eight,
Nearest to six: five, one, four, eight, three, nine, seven, two,
Nearest to some: many, several, these, most, such, certain, various, other,
Nearest to however: but, although, though, because, while, that, since, nevertheless,
Nearest to war: ii, outbreak, wwii, conflict, campaign, manassas, nationalists, veterans,
Nearest to experience: experiences, bearable, subjective, isopropanol, incoherent, coincident, gz, puerto,
Nearest to magazine: magazines, coil, macmullen, recoil, fetus, cariou, skeptics, newspaper,
Nearest to joseph: sarah, unemployment, smith, phylum, higginbotham, erroneous, moglen, ssen,
Nearest to universe: fictional, constellations, gallen, monkey, comics, dune, exposure, lvares,
Nearest to defense: guard, comix, ecclesiastic, rodentia, ignition, parody, pitman, tage,
Nearest to square: knee, dock, upland, gonzaga, wadis, interesting, kilmister, shaper,
Nearest to egypt: egyptian, jericho, andrews, receded, claw, pianist, handle, cuneiform,
Nearest to smith: joseph, mcm, avoirdupois, dildo, garner, thera, whealey, geddes,
Epoch 7/10 Iteration: 80100 Avg. Training loss: 4.2385 0.2756 sec/batch
Epoch 7/10 Iteration: 80200 Avg. Training loss: 4.2931 0.2669 sec/batch
Epoch 7/10 Iteration: 80300 Avg. Training loss: 4.3140 0.2682 sec/batch
Epoch 7/10 Iteration: 80400 Avg. Training loss: 4.3028 0.2752 sec/batch
Epoch 7/10 Iteration: 80500 Avg. Training loss: 4.2937 0.2660 sec/batch
Epoch 7/10 Iteration: 80600 Avg. Training loss: 4.1144 0.2665 sec/batch
Epoch 7/10 Iteration: 80700 Avg. Training loss: 4.1608 0.2659 sec/batch
Epoch 7/10 Iteration: 80800 Avg. Training loss: 4.3152 0.2687 sec/batch
Epoch 7/10 Iteration: 80900 Avg. Training loss: 4.1921 0.2683 sec/batch
Epoch 7/10 Iteration: 81000 Avg. Training loss: 4.2842 0.2672 sec/batch
Nearest to into: back, distancing, categories, onto, further, toward, sub, divisions,
Nearest to during: before, antiproton, after, beginning, late, prior, early, first,
Nearest to about: regarding, approximately, main, thacker, depleted, discography, question, jail,
Nearest to five: six, four, one, two, seven, three, zero, eight,
Nearest to six: five, one, four, eight, three, nine, seven, two,
Nearest to some: many, several, these, various, most, such, certain, other,
Nearest to however: but, although, though, because, while, that, despite, since,
Nearest to war: ii, outbreak, wwii, campaign, manassas, conflict, veterans, wars,
Nearest to experience: experiences, bearable, subjective, gz, isopropanol, incoherent, coincident, maybe,
Nearest to magazine: magazines, coil, macmullen, newspaper, fetus, cariou, skeptics, recoil,
Nearest to joseph: sarah, unemployment, smith, higginbotham, phylum, moglen, ssen, publication,
Nearest to universe: fictional, gallen, constellations, monkey, dune, unicef, ok, comics,
Nearest to defense: guard, comix, ignition, ecclesiastic, rodentia, parody, pitman, dalek,
Nearest to square: knee, dock, wadis, gonzaga, upland, rim, kilmister, westport,
Nearest to egypt: egyptian, jericho, receded, andrews, claw, pianist, circa, cuneiform,
Nearest to smith: mcm, joseph, garner, avoirdupois, dildo, geddes, thera, huck,
Epoch 7/10 Iteration: 81100 Avg. Training loss: 4.2955 0.2717 sec/batch
Epoch 7/10 Iteration: 81200 Avg. Training loss: 4.2967 0.2648 sec/batch
Epoch 7/10 Iteration: 81300 Avg. Training loss: 4.2516 0.2663 sec/batch
Epoch 7/10 Iteration: 81400 Avg. Training loss: 4.3250 0.2658 sec/batch
Epoch 7/10 Iteration: 81500 Avg. Training loss: 4.2848 0.2655 sec/batch
Epoch 7/10 Iteration: 81600 Avg. Training loss: 4.3595 0.2718 sec/batch
Epoch 8/10 Iteration: 81700 Avg. Training loss: 4.3237 0.2149 sec/batch
Epoch 8/10 Iteration: 81800 Avg. Training loss: 4.2626 0.2703 sec/batch
Epoch 8/10 Iteration: 81900 Avg. Training loss: 4.2987 0.2821 sec/batch
Epoch 8/10 Iteration: 82000 Avg. Training loss: 4.2687 0.2690 sec/batch
Nearest to into: back, distancing, categories, divisions, toward, sub, onto, divided,
Nearest to during: before, beginning, early, late, antiproton, after, throughout, first,
Nearest to about: regarding, approximately, main, thacker, discography, what, rugians, zero,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, eight, four, three, seven, nine, two,
Nearest to some: many, several, these, various, most, such, certain, other,
Nearest to however: but, although, though, because, while, that, since, despite,
Nearest to war: ii, outbreak, wwii, campaign, conflict, veterans, manassas, wars,
Nearest to experience: experiences, bearable, gz, subjective, moles, awaken, isopropanol, maybe,
Nearest to magazine: magazines, coil, newspaper, macmullen, issue, skeptics, fetus, recoil,
Nearest to joseph: sarah, unemployment, smith, higginbotham, moglen, phylum, erroneous, ssen,
Nearest to universe: fictional, gallen, constellations, monkey, unicef, cbe, ok, comics,
Nearest to defense: guard, comix, ignition, ecclesiastic, attrition, parody, pitman, caustic,
Nearest to square: knee, dock, wadis, upland, gonzaga, garden, westport, shaper,
Nearest to egypt: egyptian, jericho, receded, cuneiform, andrews, circa, claw, petrel,
Nearest to smith: mcm, joseph, garner, geddes, avoirdupois, filename, whealey, huck,
Epoch 8/10 Iteration: 82100 Avg. Training loss: 4.2710 0.2717 sec/batch
Epoch 8/10 Iteration: 82200 Avg. Training loss: 4.2932 0.2717 sec/batch
Epoch 8/10 Iteration: 82300 Avg. Training loss: 4.2222 0.2686 sec/batch
Epoch 8/10 Iteration: 82400 Avg. Training loss: 4.2164 0.2667 sec/batch
Epoch 8/10 Iteration: 82500 Avg. Training loss: 4.3168 0.2676 sec/batch
Epoch 8/10 Iteration: 82600 Avg. Training loss: 4.3021 0.2683 sec/batch
Epoch 8/10 Iteration: 82700 Avg. Training loss: 4.2851 0.2674 sec/batch
Epoch 8/10 Iteration: 82800 Avg. Training loss: 4.2816 0.2678 sec/batch
Epoch 8/10 Iteration: 82900 Avg. Training loss: 4.3292 0.2730 sec/batch
Epoch 8/10 Iteration: 83000 Avg. Training loss: 4.2945 0.2661 sec/batch
Nearest to into: back, distancing, onto, separate, further, through, divided, categories,
Nearest to during: late, before, throughout, antiproton, after, early, first, beginning,
Nearest to about: regarding, approximately, main, rugians, discography, thacker, depleted, covering,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, most, various, certain, such, other,
Nearest to however: but, although, though, because, that, while, and, despite,
Nearest to war: ii, wwii, outbreak, manassas, veterans, campaign, wars, conflict,
Nearest to experience: experiences, bearable, gz, subjective, moles, isopropanol, gabrieli, sensory,
Nearest to magazine: magazines, coil, newspaper, macmullen, fetus, issue, skeptics, recoil,
Nearest to joseph: sarah, smith, unemployment, phylum, moglen, ssen, erroneous, higginbotham,
Nearest to universe: fictional, gallen, constellations, monkey, unicef, big, comics, universes,
Nearest to defense: guard, ignition, comix, attrition, ecclesiastic, parody, rodentia, pitman,
Nearest to square: knee, dock, upland, wadis, gonzaga, kilmister, shaper, specimen,
Nearest to egypt: egyptian, jericho, andrews, receded, circa, claw, cuneiform, pianist,
Nearest to smith: joseph, mcm, garner, geddes, dildo, avoirdupois, thera, whealey,
Epoch 8/10 Iteration: 83100 Avg. Training loss: 4.2768 0.2721 sec/batch
Epoch 8/10 Iteration: 83200 Avg. Training loss: 4.3054 0.2742 sec/batch
Epoch 8/10 Iteration: 83300 Avg. Training loss: 4.1506 0.2738 sec/batch
Epoch 8/10 Iteration: 83400 Avg. Training loss: 4.2989 0.2712 sec/batch
Epoch 8/10 Iteration: 83500 Avg. Training loss: 4.2915 0.2679 sec/batch
Epoch 8/10 Iteration: 83600 Avg. Training loss: 4.3341 0.2654 sec/batch
Epoch 8/10 Iteration: 83700 Avg. Training loss: 4.3076 0.2698 sec/batch
Epoch 8/10 Iteration: 83800 Avg. Training loss: 4.3022 0.2726 sec/batch
Epoch 8/10 Iteration: 83900 Avg. Training loss: 4.2872 0.2655 sec/batch
Epoch 8/10 Iteration: 84000 Avg. Training loss: 4.3337 0.2685 sec/batch
Nearest to into: back, onto, through, further, then, divided, eastbourne, divisions,
Nearest to during: late, before, throughout, after, early, antiproton, prior, beginning,
Nearest to about: regarding, approximately, rugians, discography, main, five, depleted, thacker,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, three, eight, seven, nine, two,
Nearest to some: many, these, several, most, various, certain, other, such,
Nearest to however: but, although, though, because, that, while, since, despite,
Nearest to war: ii, wwii, outbreak, manassas, campaign, wars, nationalists, branson,
Nearest to experience: experiences, subjective, bearable, gz, auditory, isopropanol, gabrieli, moles,
Nearest to magazine: coil, magazines, macmullen, fetus, newspaper, issue, skeptics, recoil,
Nearest to joseph: smith, sarah, unemployment, phylum, moglen, erroneous, ssen, abaddon,
Nearest to universe: fictional, gallen, constellations, big, comics, monkey, ok, unicef,
Nearest to defense: guard, comix, ignition, attrition, rodentia, ecclesiastic, pitman, caustic,
Nearest to square: knee, dock, wadis, upland, shaper, gonzaga, triangle, specimen,
Nearest to egypt: jericho, egyptian, andrews, receded, claw, circa, cuneiform, pianist,
Nearest to smith: joseph, mcm, geddes, garner, dildo, thera, avoirdupois, filename,
Epoch 8/10 Iteration: 84100 Avg. Training loss: 4.3133 0.2704 sec/batch
Epoch 8/10 Iteration: 84200 Avg. Training loss: 4.3296 0.2673 sec/batch
Epoch 8/10 Iteration: 84300 Avg. Training loss: 4.2492 0.2678 sec/batch
Epoch 8/10 Iteration: 84400 Avg. Training loss: 4.2273 0.2669 sec/batch
Epoch 8/10 Iteration: 84500 Avg. Training loss: 4.0754 0.2678 sec/batch
Epoch 8/10 Iteration: 84600 Avg. Training loss: 4.2451 0.2673 sec/batch
Epoch 8/10 Iteration: 84700 Avg. Training loss: 4.2609 0.2692 sec/batch
Epoch 8/10 Iteration: 84800 Avg. Training loss: 4.2915 0.2706 sec/batch
Epoch 8/10 Iteration: 84900 Avg. Training loss: 4.2445 0.2697 sec/batch
Epoch 8/10 Iteration: 85000 Avg. Training loss: 4.2591 0.2680 sec/batch
Nearest to into: back, distancing, further, categories, divided, sub, onto, divisions,
Nearest to during: late, throughout, before, beginning, early, period, after, antiproton,
Nearest to about: regarding, approximately, main, thacker, depleted, discography, rugians, zero,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, nine, seven, two,
Nearest to some: many, several, these, certain, most, various, other, such,
Nearest to however: but, although, though, because, that, while, since, indeed,
Nearest to war: ii, wwii, outbreak, wars, campaign, manassas, nationalists, veterans,
Nearest to experience: subjective, experiences, gz, bearable, gabrieli, auditory, isopropanol, coincident,
Nearest to magazine: coil, magazines, macmullen, fetus, issue, skeptics, newspaper, handguns,
Nearest to joseph: smith, sarah, phylum, moglen, unemployment, publication, ssen, erroneous,
Nearest to universe: gallen, fictional, constellations, monkey, comics, big, ok, nolan,
Nearest to defense: guard, comix, ignition, ecclesiastic, attrition, pitman, rodentia, parody,
Nearest to square: knee, dock, wadis, westport, upland, shaper, gonzaga, specimen,
Nearest to egypt: egyptian, jericho, andrews, circa, receded, claw, cuneiform, pianist,
Nearest to smith: joseph, mcm, geddes, garner, avoirdupois, thera, dildo, sisley,
Epoch 8/10 Iteration: 85100 Avg. Training loss: 4.1644 0.2688 sec/batch
Epoch 8/10 Iteration: 85200 Avg. Training loss: 4.2761 0.2750 sec/batch
Epoch 8/10 Iteration: 85300 Avg. Training loss: 4.3161 0.2694 sec/batch
Epoch 8/10 Iteration: 85400 Avg. Training loss: 4.3028 0.2675 sec/batch
Epoch 8/10 Iteration: 85500 Avg. Training loss: 4.2930 0.2683 sec/batch
Epoch 8/10 Iteration: 85600 Avg. Training loss: 4.2916 0.2669 sec/batch
Epoch 8/10 Iteration: 85700 Avg. Training loss: 4.2628 0.2670 sec/batch
Epoch 8/10 Iteration: 85800 Avg. Training loss: 4.3359 0.2673 sec/batch
Epoch 8/10 Iteration: 85900 Avg. Training loss: 4.2290 0.2715 sec/batch
Epoch 8/10 Iteration: 86000 Avg. Training loss: 4.2922 0.2680 sec/batch
Nearest to into: back, further, categories, distancing, sub, through, divisions, divided,
Nearest to during: late, before, early, beginning, throughout, after, first, prior,
Nearest to about: regarding, approximately, main, shortening, thacker, depleted, rugians, discography,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, nine, seven, two,
Nearest to some: many, several, these, various, most, certain, such, other,
Nearest to however: although, but, though, because, while, that, since, despite,
Nearest to war: ii, wwii, outbreak, wars, manassas, campaign, veterans, world,
Nearest to experience: subjective, experiences, gz, bearable, maybe, awaken, auditory, coincident,
Nearest to magazine: coil, magazines, skeptics, issue, macmullen, fetus, recoil, handguns,
Nearest to joseph: smith, sarah, phylum, moglen, publication, unemployment, erroneous, ssen,
Nearest to universe: fictional, gallen, constellations, monkey, ok, big, comics, exposure,
Nearest to defense: guard, comix, ignition, pitman, attrition, ecclesiastic, caustic, tage,
Nearest to square: knee, dock, wadis, westport, gonzaga, upland, shaper, kilmister,
Nearest to egypt: egyptian, andrews, jericho, circa, receded, claw, pianist, cuneiform,
Nearest to smith: joseph, mcm, garner, avoirdupois, geddes, filename, sisley, lila,
Epoch 8/10 Iteration: 86100 Avg. Training loss: 4.2402 0.2736 sec/batch
Epoch 8/10 Iteration: 86200 Avg. Training loss: 4.3009 0.2677 sec/batch
Epoch 8/10 Iteration: 86300 Avg. Training loss: 4.2994 0.2659 sec/batch
Epoch 8/10 Iteration: 86400 Avg. Training loss: 4.3585 0.2690 sec/batch
Epoch 8/10 Iteration: 86500 Avg. Training loss: 4.3305 0.2680 sec/batch
Epoch 8/10 Iteration: 86600 Avg. Training loss: 4.3008 0.2686 sec/batch
Epoch 8/10 Iteration: 86700 Avg. Training loss: 4.1811 0.2721 sec/batch
Epoch 8/10 Iteration: 86800 Avg. Training loss: 4.3269 0.2717 sec/batch
Epoch 8/10 Iteration: 86900 Avg. Training loss: 4.2361 0.2658 sec/batch
Epoch 8/10 Iteration: 87000 Avg. Training loss: 4.2953 0.2692 sec/batch
Nearest to into: back, distancing, further, sub, categories, divided, through, divisions,
Nearest to during: late, before, early, antiproton, throughout, after, prior, beginning,
Nearest to about: regarding, approximately, main, thacker, rugians, zero, depleted, discography,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, most, various, certain, such, few,
Nearest to however: but, although, though, because, that, while, since, indeed,
Nearest to war: ii, wwii, outbreak, campaign, manassas, wars, nationalists, veterans,
Nearest to experience: experiences, subjective, bearable, gz, gabrieli, awaken, coincident, auditory,
Nearest to magazine: coil, macmullen, magazines, skeptics, newspaper, recoil, issue, cariou,
Nearest to joseph: smith, sarah, phylum, moglen, unemployment, publication, erroneous, abaddon,
Nearest to universe: fictional, gallen, constellations, monkey, ok, nolan, big, exposure,
Nearest to defense: guard, comix, ecclesiastic, ignition, pitman, caustic, attrition, parody,
Nearest to square: knee, dock, wadis, westport, upland, kilmister, shaper, gonzaga,
Nearest to egypt: egyptian, jericho, andrews, claw, handle, receded, circa, cuneiform,
Nearest to smith: joseph, mcm, garner, avoirdupois, filename, geddes, theory, thera,
Epoch 8/10 Iteration: 87100 Avg. Training loss: 4.3031 0.2719 sec/batch
Epoch 8/10 Iteration: 87200 Avg. Training loss: 4.2624 0.2709 sec/batch
Epoch 8/10 Iteration: 87300 Avg. Training loss: 4.3180 0.2670 sec/batch
Epoch 8/10 Iteration: 87400 Avg. Training loss: 4.2070 0.2690 sec/batch
Epoch 8/10 Iteration: 87500 Avg. Training loss: 4.2838 0.2651 sec/batch
Epoch 8/10 Iteration: 87600 Avg. Training loss: 4.2352 0.2709 sec/batch
Epoch 8/10 Iteration: 87700 Avg. Training loss: 4.2777 0.2680 sec/batch
Epoch 8/10 Iteration: 87800 Avg. Training loss: 4.3109 0.2657 sec/batch
Epoch 8/10 Iteration: 87900 Avg. Training loss: 4.1881 0.2659 sec/batch
Epoch 8/10 Iteration: 88000 Avg. Training loss: 4.3294 0.2644 sec/batch
Nearest to into: back, distancing, sub, categories, divisions, divided, further, separate,
Nearest to during: late, after, early, antiproton, prior, before, throughout, period,
Nearest to about: regarding, main, thacker, rugians, discography, approximately, depleted, shortening,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, eight, four, three, seven, nine, two,
Nearest to some: many, several, these, most, certain, various, such, other,
Nearest to however: but, although, though, that, because, while, since, only,
Nearest to war: ii, wwii, outbreak, campaign, manassas, wars, nationalists, serials,
Nearest to experience: experiences, subjective, gz, bearable, awaken, maybe, gabrieli, coincident,
Nearest to magazine: coil, magazines, macmullen, skeptics, issue, newspaper, recoil, cariou,
Nearest to joseph: smith, sarah, phylum, unemployment, higginbotham, moglen, publication, abaddon,
Nearest to universe: fictional, gallen, constellations, monkey, dune, nolan, ok, exposure,
Nearest to defense: guard, comix, ecclesiastic, pitman, ignition, caustic, parody, tage,
Nearest to square: knee, dock, wadis, upland, shaper, westport, kilmister, grocery,
Nearest to egypt: egyptian, jericho, andrews, claw, receded, handle, petrel, cuneiform,
Nearest to smith: joseph, mcm, garner, avoirdupois, geddes, filename, thera, sisley,
Epoch 8/10 Iteration: 88100 Avg. Training loss: 4.2684 0.2684 sec/batch
Epoch 8/10 Iteration: 88200 Avg. Training loss: 4.2676 0.2682 sec/batch
Epoch 8/10 Iteration: 88300 Avg. Training loss: 4.1677 0.2691 sec/batch
Epoch 8/10 Iteration: 88400 Avg. Training loss: 4.2962 0.2663 sec/batch
Epoch 8/10 Iteration: 88500 Avg. Training loss: 4.3154 0.2682 sec/batch
Epoch 8/10 Iteration: 88600 Avg. Training loss: 4.2939 0.2712 sec/batch
Epoch 8/10 Iteration: 88700 Avg. Training loss: 4.3321 0.2693 sec/batch
Epoch 8/10 Iteration: 88800 Avg. Training loss: 4.3265 0.2685 sec/batch
Epoch 8/10 Iteration: 88900 Avg. Training loss: 4.2090 0.2680 sec/batch
Epoch 8/10 Iteration: 89000 Avg. Training loss: 4.2914 0.2663 sec/batch
Nearest to into: back, divided, further, categories, eventually, sub, distancing, divisions,
Nearest to during: late, after, before, throughout, prior, antiproton, in, early,
Nearest to about: regarding, main, approximately, zero, thacker, rugians, discography, gssps,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, most, certain, various, such, other,
Nearest to however: but, although, though, that, because, while, since, despite,
Nearest to war: ii, wwii, outbreak, wars, campaign, manassas, occupation, conflict,
Nearest to experience: experiences, subjective, gz, bearable, maybe, coincident, incoherent, awaken,
Nearest to magazine: magazines, coil, skeptics, macmullen, newspaper, recoil, published, issue,
Nearest to joseph: sarah, smith, unemployment, phylum, higginbotham, abaddon, erroneous, moglen,
Nearest to universe: fictional, gallen, constellations, monkey, dune, nolan, handful, ok,
Nearest to defense: guard, comix, ecclesiastic, pitman, ignition, caustic, parody, dalek,
Nearest to square: knee, dock, wadis, upland, westport, shaper, kilmister, gonzaga,
Nearest to egypt: egyptian, jericho, andrews, claw, receded, circa, cuneiform, handle,
Nearest to smith: joseph, mcm, avoirdupois, garner, filename, geddes, dungeons, thera,
Epoch 8/10 Iteration: 89100 Avg. Training loss: 4.3058 0.2749 sec/batch
Epoch 8/10 Iteration: 89200 Avg. Training loss: 4.3236 0.2708 sec/batch
Epoch 8/10 Iteration: 89300 Avg. Training loss: 4.2637 0.2739 sec/batch
Epoch 8/10 Iteration: 89400 Avg. Training loss: 4.3106 0.2709 sec/batch
Epoch 8/10 Iteration: 89500 Avg. Training loss: 4.2765 0.2631 sec/batch
Epoch 8/10 Iteration: 89600 Avg. Training loss: 4.3267 0.2712 sec/batch
Epoch 8/10 Iteration: 89700 Avg. Training loss: 4.2844 0.2678 sec/batch
Epoch 8/10 Iteration: 89800 Avg. Training loss: 4.1884 0.2678 sec/batch
Epoch 8/10 Iteration: 89900 Avg. Training loss: 4.2626 0.2682 sec/batch
Epoch 8/10 Iteration: 90000 Avg. Training loss: 4.2539 0.2696 sec/batch
Nearest to into: back, divided, categories, sub, further, distancing, divisions, then,
Nearest to during: late, prior, throughout, early, after, before, beginning, antiproton,
Nearest to about: regarding, approximately, main, zero, thacker, shortening, depleted, around,
Nearest to five: four, one, six, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, most, such, certain, various, other,
Nearest to however: although, but, though, because, that, while, since, only,
Nearest to war: ii, wars, outbreak, wwii, campaign, occupation, manassas, conflict,
Nearest to experience: experiences, subjective, gz, bearable, awaken, isopropanol, coincident, gabrieli,
Nearest to magazine: magazines, coil, newspaper, skeptics, macmullen, recoil, published, fetus,
Nearest to joseph: sarah, smith, phylum, unemployment, erroneous, abaddon, higginbotham, schlick,
Nearest to universe: fictional, gallen, constellations, monkey, dune, nolan, ok, ehud,
Nearest to defense: comix, guard, ecclesiastic, ignition, pitman, caustic, dalek, tage,
Nearest to square: knee, dock, wadis, shaper, upland, kilmister, westport, stochastic,
Nearest to egypt: egyptian, jericho, andrews, claw, cuneiform, circa, receded, handle,
Nearest to smith: joseph, mcm, avoirdupois, filename, garner, geddes, dungeons, thera,
Epoch 8/10 Iteration: 90100 Avg. Training loss: 4.2837 0.2758 sec/batch
Epoch 8/10 Iteration: 90200 Avg. Training loss: 4.2498 0.2744 sec/batch
Epoch 8/10 Iteration: 90300 Avg. Training loss: 4.3351 0.2700 sec/batch
Epoch 8/10 Iteration: 90400 Avg. Training loss: 4.1894 0.2707 sec/batch
Epoch 8/10 Iteration: 90500 Avg. Training loss: 3.9863 0.2692 sec/batch
Epoch 8/10 Iteration: 90600 Avg. Training loss: 4.1559 0.2650 sec/batch
Epoch 8/10 Iteration: 90700 Avg. Training loss: 4.2292 0.2693 sec/batch
Epoch 8/10 Iteration: 90800 Avg. Training loss: 4.2790 0.2691 sec/batch
Epoch 8/10 Iteration: 90900 Avg. Training loss: 4.3102 0.2684 sec/batch
Epoch 8/10 Iteration: 91000 Avg. Training loss: 4.3012 0.2673 sec/batch
Nearest to into: back, categories, sub, distancing, divided, divisions, then, further,
Nearest to during: late, after, before, prior, early, following, antiproton, throughout,
Nearest to about: regarding, approximately, main, depleted, thacker, discography, jail, told,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, most, such, various, certain, few,
Nearest to however: although, but, though, because, that, while, since, despite,
Nearest to war: ii, campaign, outbreak, wwii, manassas, veterans, conflict, occupation,
Nearest to experience: experiences, subjective, bearable, gz, puerto, incoherent, awaken, coincident,
Nearest to magazine: coil, macmullen, skeptics, magazines, issue, newspaper, recoil, cariou,
Nearest to joseph: smith, sarah, phylum, unemployment, erroneous, abaddon, moglen, accra,
Nearest to universe: gallen, constellations, fictional, dune, monkey, ehud, nolan, ok,
Nearest to defense: comix, guard, ecclesiastic, ignition, parody, tage, pitman, dalek,
Nearest to square: knee, dock, wadis, upland, gonzaga, shaper, westport, kilmister,
Nearest to egypt: egyptian, jericho, claw, andrews, receded, handle, circa, cuneiform,
Nearest to smith: joseph, geddes, filename, mcm, garner, avoirdupois, whealey, dildo,
Epoch 8/10 Iteration: 91100 Avg. Training loss: 4.3552 0.2691 sec/batch
Epoch 8/10 Iteration: 91200 Avg. Training loss: 4.3674 0.2664 sec/batch
Epoch 8/10 Iteration: 91300 Avg. Training loss: 4.3530 0.2679 sec/batch
Epoch 8/10 Iteration: 91400 Avg. Training loss: 4.2961 0.2677 sec/batch
Epoch 8/10 Iteration: 91500 Avg. Training loss: 4.1288 0.2661 sec/batch
Epoch 8/10 Iteration: 91600 Avg. Training loss: 4.3116 0.2687 sec/batch
Epoch 8/10 Iteration: 91700 Avg. Training loss: 4.2765 0.2658 sec/batch
Epoch 8/10 Iteration: 91800 Avg. Training loss: 4.2750 0.2643 sec/batch
Epoch 8/10 Iteration: 91900 Avg. Training loss: 4.2619 0.2714 sec/batch
Epoch 8/10 Iteration: 92000 Avg. Training loss: 4.3406 0.2729 sec/batch
Nearest to into: back, divided, distancing, sub, categories, onto, divisions, further,
Nearest to during: after, late, prior, before, antiproton, early, beginning, throughout,
Nearest to about: regarding, approximately, main, jail, thacker, discography, depleted, soldiers,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, several, these, most, various, such, certain, other,
Nearest to however: although, but, though, because, that, while, since, despite,
Nearest to war: ii, wwii, outbreak, campaign, manassas, occupation, wars, nationalists,
Nearest to experience: experiences, subjective, gz, bearable, isopropanol, coincident, maybe, awaken,
Nearest to magazine: coil, skeptics, magazines, recoil, macmullen, issue, fetus, newspaper,
Nearest to joseph: smith, sarah, unemployment, phylum, erroneous, moglen, abaddon, shop,
Nearest to universe: gallen, constellations, fictional, comics, monkey, ehud, nolan, dune,
Nearest to defense: comix, guard, ignition, ecclesiastic, tage, parody, dalek, caustic,
Nearest to square: knee, dock, wadis, westport, shaper, grocery, upland, gonzaga,
Nearest to egypt: egyptian, jericho, andrews, claw, receded, handle, cuneiform, circa,
Nearest to smith: joseph, mcm, filename, geddes, avoirdupois, dildo, garner, whealey,
Epoch 8/10 Iteration: 92100 Avg. Training loss: 4.2559 0.2703 sec/batch
Epoch 8/10 Iteration: 92200 Avg. Training loss: 4.2394 0.2652 sec/batch
Epoch 8/10 Iteration: 92300 Avg. Training loss: 4.0187 0.2679 sec/batch
Epoch 8/10 Iteration: 92400 Avg. Training loss: 4.3157 0.2682 sec/batch
Epoch 8/10 Iteration: 92500 Avg. Training loss: 4.2861 0.2682 sec/batch
Epoch 8/10 Iteration: 92600 Avg. Training loss: 4.1710 0.2695 sec/batch
Epoch 8/10 Iteration: 92700 Avg. Training loss: 4.2821 0.2708 sec/batch
Epoch 8/10 Iteration: 92800 Avg. Training loss: 4.2792 0.2679 sec/batch
Epoch 8/10 Iteration: 92900 Avg. Training loss: 4.2633 0.2711 sec/batch
Epoch 8/10 Iteration: 93000 Avg. Training loss: 4.2319 0.2729 sec/batch
Nearest to into: back, categories, divided, onto, through, divisions, distancing, sub,
Nearest to during: late, before, after, beginning, early, prior, antiproton, first,
Nearest to about: regarding, approximately, main, discography, jail, thacker, liberian, what,
Nearest to five: four, one, six, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, two, nine,
Nearest to some: many, several, these, various, such, most, certain, other,
Nearest to however: although, but, though, because, while, that, since, and,
Nearest to war: ii, wwii, outbreak, campaign, wars, manassas, veterans, occupation,
Nearest to experience: experiences, subjective, bearable, gz, awaken, coincident, maybe, isopropanol,
Nearest to magazine: coil, skeptics, recoil, magazines, issue, macmullen, fetus, graven,
Nearest to joseph: smith, unemployment, sarah, phylum, moglen, erroneous, ssen, shop,
Nearest to universe: constellations, gallen, fictional, comics, ehud, nolan, dune, ok,
Nearest to defense: comix, guard, ignition, dalek, ecclesiastic, caustic, tage, parody,
Nearest to square: knee, dock, wadis, westport, garden, gonzaga, shaper, grocery,
Nearest to egypt: egyptian, andrews, jericho, claw, receded, cuneiform, circa, handle,
Nearest to smith: joseph, mcm, geddes, avoirdupois, filename, garner, dildo, sisley,
Epoch 8/10 Iteration: 93100 Avg. Training loss: 4.3180 0.2685 sec/batch
Epoch 8/10 Iteration: 93200 Avg. Training loss: 4.3311 0.2678 sec/batch
Epoch 9/10 Iteration: 93300 Avg. Training loss: 4.3215 0.0541 sec/batch
Epoch 9/10 Iteration: 93400 Avg. Training loss: 4.3123 0.2690 sec/batch
Epoch 9/10 Iteration: 93500 Avg. Training loss: 4.2240 0.2738 sec/batch
Epoch 9/10 Iteration: 93600 Avg. Training loss: 4.3278 0.2759 sec/batch
Epoch 9/10 Iteration: 93700 Avg. Training loss: 4.2244 0.2669 sec/batch
Epoch 9/10 Iteration: 93800 Avg. Training loss: 4.2550 0.2729 sec/batch
Epoch 9/10 Iteration: 93900 Avg. Training loss: 4.2902 0.2721 sec/batch
Epoch 9/10 Iteration: 94000 Avg. Training loss: 4.2183 0.2677 sec/batch
Nearest to into: back, then, divisions, divided, further, distancing, categories, toward,
Nearest to during: late, before, after, early, beginning, in, period, throughout,
Nearest to about: regarding, approximately, zero, main, five, around, rugians, discography,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, various, most, certain, such, other,
Nearest to however: although, but, though, because, that, while, since, and,
Nearest to war: ii, outbreak, wwii, manassas, campaign, wars, veterans, branson,
Nearest to experience: experiences, subjective, bearable, awaken, gz, coincident, bloodless, auditory,
Nearest to magazine: coil, skeptics, magazines, recoil, macmullen, fetus, newspaper, issue,
Nearest to joseph: smith, unemployment, sarah, moglen, phylum, shop, erroneous, higginbotham,
Nearest to universe: constellations, gallen, fictional, nolan, dune, comics, ehud, ok,
Nearest to defense: guard, comix, ignition, ecclesiastic, parody, caustic, tage, dalek,
Nearest to square: dock, knee, wadis, westport, gonzaga, garden, upland, shaper,
Nearest to egypt: egyptian, cuneiform, jericho, claw, circa, receded, andrews, rameses,
Nearest to smith: joseph, mcm, geddes, filename, avoirdupois, garner, dildo, whealey,
Epoch 9/10 Iteration: 94100 Avg. Training loss: 4.2357 0.2708 sec/batch
Epoch 9/10 Iteration: 94200 Avg. Training loss: 4.3110 0.2661 sec/batch
Epoch 9/10 Iteration: 94300 Avg. Training loss: 4.3020 0.2644 sec/batch
Epoch 9/10 Iteration: 94400 Avg. Training loss: 4.2482 0.2698 sec/batch
Epoch 9/10 Iteration: 94500 Avg. Training loss: 4.2737 0.2670 sec/batch
Epoch 9/10 Iteration: 94600 Avg. Training loss: 4.2984 0.2674 sec/batch
Epoch 9/10 Iteration: 94700 Avg. Training loss: 4.3163 0.2636 sec/batch
Epoch 9/10 Iteration: 94800 Avg. Training loss: 4.2964 0.2668 sec/batch
Epoch 9/10 Iteration: 94900 Avg. Training loss: 4.2923 0.2661 sec/batch
Epoch 9/10 Iteration: 95000 Avg. Training loss: 4.1281 0.2653 sec/batch
Nearest to into: divided, further, divisions, back, then, through, separate, in,
Nearest to during: late, early, after, throughout, prior, before, in, period,
Nearest to about: regarding, zero, approximately, five, main, four, two, around,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, various, certain, most, such, other,
Nearest to however: although, but, though, because, that, while, since, and,
Nearest to war: ii, wwii, manassas, wars, outbreak, occupation, veterans, campaign,
Nearest to experience: experiences, subjective, bearable, gz, awaken, auditory, coincident, bloodless,
Nearest to magazine: coil, skeptics, magazines, macmullen, fetus, issue, recoil, cariou,
Nearest to joseph: smith, unemployment, sarah, moglen, phylum, erroneous, abaddon, kite,
Nearest to universe: constellations, gallen, fictional, nolan, ehud, dune, exposure, unicef,
Nearest to defense: guard, comix, ignition, ecclesiastic, dalek, parody, icbms, iea,
Nearest to square: wadis, knee, dock, gonzaga, upland, westport, garden, shaper,
Nearest to egypt: egyptian, jericho, circa, claw, cuneiform, receded, handle, andrews,
Nearest to smith: joseph, mcm, geddes, avoirdupois, garner, filename, dildo, sisley,
Epoch 9/10 Iteration: 95100 Avg. Training loss: 4.3021 0.2699 sec/batch
Epoch 9/10 Iteration: 95200 Avg. Training loss: 4.3115 0.2675 sec/batch
Epoch 9/10 Iteration: 95300 Avg. Training loss: 4.3078 0.2731 sec/batch
Epoch 9/10 Iteration: 95400 Avg. Training loss: 4.2985 0.2745 sec/batch
Epoch 9/10 Iteration: 95500 Avg. Training loss: 4.2990 0.2704 sec/batch
Epoch 9/10 Iteration: 95600 Avg. Training loss: 4.2961 0.2681 sec/batch
Epoch 9/10 Iteration: 95700 Avg. Training loss: 4.3326 0.2687 sec/batch
Epoch 9/10 Iteration: 95800 Avg. Training loss: 4.3260 0.2677 sec/batch
Epoch 9/10 Iteration: 95900 Avg. Training loss: 4.2775 0.2667 sec/batch
Epoch 9/10 Iteration: 96000 Avg. Training loss: 4.2519 0.2679 sec/batch
Nearest to into: further, back, divided, divisions, categories, distancing, then, separate,
Nearest to during: late, throughout, after, early, before, prior, period, first,
Nearest to about: regarding, main, approximately, rugians, gssps, zero, discography, thacker,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, certain, various, most, such, few,
Nearest to however: although, but, though, because, that, while, since, despite,
Nearest to war: ii, wwii, wars, manassas, outbreak, occupation, campaign, world,
Nearest to experience: experiences, subjective, bearable, gz, awaken, auditory, isopropanol, coincident,
Nearest to magazine: coil, skeptics, macmullen, magazines, fetus, recoil, issue, bodybuilders,
Nearest to joseph: smith, unemployment, sarah, phylum, moglen, erroneous, shop, abaddon,
Nearest to universe: constellations, gallen, fictional, nolan, dune, ehud, universes, unicef,
Nearest to defense: guard, ignition, comix, ecclesiastic, pitman, dalek, parody, caustic,
Nearest to square: knee, dock, wadis, gonzaga, tiananmen, westport, kilmister, shaper,
Nearest to egypt: jericho, circa, egyptian, claw, cuneiform, handle, receded, rameses,
Nearest to smith: joseph, mcm, geddes, garner, avoirdupois, filename, beeswax, pup,
Epoch 9/10 Iteration: 96100 Avg. Training loss: 4.1685 0.2707 sec/batch
Epoch 9/10 Iteration: 96200 Avg. Training loss: 4.1554 0.2688 sec/batch
Epoch 9/10 Iteration: 96300 Avg. Training loss: 4.2596 0.2710 sec/batch
Epoch 9/10 Iteration: 96400 Avg. Training loss: 4.2555 0.2703 sec/batch
Epoch 9/10 Iteration: 96500 Avg. Training loss: 4.2512 0.2714 sec/batch
Epoch 9/10 Iteration: 96600 Avg. Training loss: 4.2711 0.2712 sec/batch
Epoch 9/10 Iteration: 96700 Avg. Training loss: 4.2621 0.2674 sec/batch
Epoch 9/10 Iteration: 96800 Avg. Training loss: 4.1285 0.2670 sec/batch
Epoch 9/10 Iteration: 96900 Avg. Training loss: 4.2975 0.2704 sec/batch
Epoch 9/10 Iteration: 97000 Avg. Training loss: 4.2982 0.2769 sec/batch
Nearest to into: back, then, further, divided, distancing, divisions, toward, through,
Nearest to during: late, early, after, before, throughout, prior, in, first,
Nearest to about: regarding, main, approximately, gssps, rugians, zero, thacker, shortening,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, three, eight, seven, nine, two,
Nearest to some: many, several, these, certain, various, such, most, few,
Nearest to however: although, but, though, because, while, that, since, despite,
Nearest to war: ii, wwii, manassas, outbreak, wars, campaign, world, occupation,
Nearest to experience: experiences, subjective, bearable, gz, awaken, isopropanol, coincident, sensory,
Nearest to magazine: coil, magazines, skeptics, macmullen, issue, fetus, bodybuilders, recoil,
Nearest to joseph: smith, phylum, moglen, sarah, shop, erroneous, unemployment, publication,
Nearest to universe: constellations, gallen, fictional, nolan, dune, universes, ehud, unicef,
Nearest to defense: guard, ecclesiastic, ignition, comix, dalek, pitman, tage, attrition,
Nearest to square: knee, dock, wadis, grocery, stochastic, gonzaga, kilmister, westport,
Nearest to egypt: circa, jericho, egyptian, claw, cuneiform, handle, receded, andrews,
Nearest to smith: joseph, mcm, geddes, filename, sisley, beeswax, avoirdupois, garner,
Epoch 9/10 Iteration: 97100 Avg. Training loss: 4.2774 0.2696 sec/batch
Epoch 9/10 Iteration: 97200 Avg. Training loss: 4.2977 0.2738 sec/batch
Epoch 9/10 Iteration: 97300 Avg. Training loss: 4.2600 0.2707 sec/batch
Epoch 9/10 Iteration: 97400 Avg. Training loss: 4.2964 0.2701 sec/batch
Epoch 9/10 Iteration: 97500 Avg. Training loss: 4.3177 0.2707 sec/batch
Epoch 9/10 Iteration: 97600 Avg. Training loss: 4.2278 0.2693 sec/batch
Epoch 9/10 Iteration: 97700 Avg. Training loss: 4.2865 0.2721 sec/batch
Epoch 9/10 Iteration: 97800 Avg. Training loss: 4.2453 0.2667 sec/batch
Epoch 9/10 Iteration: 97900 Avg. Training loss: 4.2919 0.2675 sec/batch
Epoch 9/10 Iteration: 98000 Avg. Training loss: 4.3104 0.2681 sec/batch
Nearest to into: then, back, categories, divided, distancing, sub, further, destry,
Nearest to during: late, early, before, after, prior, throughout, beginning, in,
Nearest to about: regarding, main, approximately, shortening, gssps, thacker, rugians, zero,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, certain, various, most, such, few,
Nearest to however: although, but, though, because, that, while, since, despite,
Nearest to war: ii, wwii, outbreak, manassas, wars, conflict, cas, occupation,
Nearest to experience: experiences, subjective, gz, bearable, awaken, coincident, auditory, isopropanol,
Nearest to magazine: coil, issue, magazines, macmullen, skeptics, fetus, bodybuilders, recoil,
Nearest to joseph: smith, moglen, phylum, unemployment, shop, erroneous, sarah, publication,
Nearest to universe: gallen, constellations, dune, nolan, ehud, monkey, fictional, exposure,
Nearest to defense: guard, comix, ecclesiastic, ignition, tage, pitman, caustic, dalek,
Nearest to square: knee, dock, wadis, westport, gonzaga, stochastic, recordable, kilmister,
Nearest to egypt: egyptian, circa, claw, jericho, handle, andrews, cuneiform, receded,
Nearest to smith: joseph, mcm, geddes, filename, avoirdupois, garner, pup, sisley,
Epoch 9/10 Iteration: 98100 Avg. Training loss: 4.3029 0.2698 sec/batch
Epoch 9/10 Iteration: 98200 Avg. Training loss: 4.3219 0.2676 sec/batch
Epoch 9/10 Iteration: 98300 Avg. Training loss: 4.2551 0.2713 sec/batch
Epoch 9/10 Iteration: 98400 Avg. Training loss: 4.1934 0.2701 sec/batch
Epoch 9/10 Iteration: 98500 Avg. Training loss: 4.2794 0.2671 sec/batch
Epoch 9/10 Iteration: 98600 Avg. Training loss: 4.2381 0.2777 sec/batch
Epoch 9/10 Iteration: 98700 Avg. Training loss: 4.3172 0.2726 sec/batch
Epoch 9/10 Iteration: 98800 Avg. Training loss: 4.2851 0.2695 sec/batch
Epoch 9/10 Iteration: 98900 Avg. Training loss: 4.2839 0.2693 sec/batch
Epoch 9/10 Iteration: 99000 Avg. Training loss: 4.2690 0.2684 sec/batch
Nearest to into: back, divided, distancing, then, further, categories, sub, toward,
Nearest to during: late, early, before, after, prior, throughout, period, in,
Nearest to about: regarding, main, approximately, rugians, what, shortening, depleted, thacker,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, several, these, certain, most, various, such, few,
Nearest to however: although, but, though, because, that, while, since, only,
Nearest to war: ii, wwii, outbreak, manassas, wars, world, occupation, serials,
Nearest to experience: experiences, subjective, gz, bearable, awaken, coincident, sensory, auditory,
Nearest to magazine: magazines, coil, issue, macmullen, skeptics, newspaper, monthly, recoil,
Nearest to joseph: smith, unemployment, moglen, erroneous, phylum, sarah, shop, publication,
Nearest to universe: gallen, constellations, dune, nolan, monkey, ehud, exposure, fictional,
Nearest to defense: guard, ecclesiastic, comix, ignition, dalek, defence, tage, pitman,
Nearest to square: dock, knee, wadis, gonzaga, westport, shaper, stochastic, kilmister,
Nearest to egypt: egyptian, jericho, claw, circa, handle, cuneiform, receded, andrews,
Nearest to smith: joseph, mcm, geddes, sisley, avoirdupois, filename, dildo, garner,
Epoch 9/10 Iteration: 99100 Avg. Training loss: 4.2315 0.2724 sec/batch
Epoch 9/10 Iteration: 99200 Avg. Training loss: 4.2752 0.2696 sec/batch
Epoch 9/10 Iteration: 99300 Avg. Training loss: 4.2337 0.2654 sec/batch
Epoch 9/10 Iteration: 99400 Avg. Training loss: 4.2906 0.2684 sec/batch
Epoch 9/10 Iteration: 99500 Avg. Training loss: 4.1524 0.2698 sec/batch
Epoch 9/10 Iteration: 99600 Avg. Training loss: 4.3171 0.2671 sec/batch
Epoch 9/10 Iteration: 99700 Avg. Training loss: 4.3114 0.2704 sec/batch
Epoch 9/10 Iteration: 99800 Avg. Training loss: 4.2432 0.2664 sec/batch
Epoch 9/10 Iteration: 99900 Avg. Training loss: 4.2357 0.2683 sec/batch
Epoch 9/10 Iteration: 100000 Avg. Training loss: 4.2364 0.2695 sec/batch
Nearest to into: divided, back, further, categories, then, distancing, divisions, sub,
Nearest to during: late, throughout, prior, after, early, before, period, antiproton,
Nearest to about: regarding, main, approximately, rugians, thacker, depleted, told, shortening,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, most, certain, such, various, few,
Nearest to however: although, but, though, because, while, that, since, and,
Nearest to war: ii, wwii, outbreak, manassas, wars, occupation, conflict, serials,
Nearest to experience: experiences, subjective, gz, bearable, coincident, maybe, bloodless, sensory,
Nearest to magazine: magazines, coil, issue, macmullen, skeptics, newspaper, monthly, recoil,
Nearest to joseph: smith, unemployment, shop, phylum, kite, erroneous, accra, moglen,
Nearest to universe: gallen, constellations, dune, nolan, exposure, prot, monkey, ehud,
Nearest to defense: guard, ecclesiastic, comix, dalek, defence, ignition, tage, pitman,
Nearest to square: dock, knee, wadis, westport, stochastic, gonzaga, shaper, kilmister,
Nearest to egypt: egyptian, claw, jericho, circa, handle, cuneiform, receded, andrews,
Nearest to smith: joseph, mcm, geddes, sisley, avoirdupois, filename, garner, dildo,
Epoch 9/10 Iteration: 100100 Avg. Training loss: 4.2925 0.2674 sec/batch
Epoch 9/10 Iteration: 100200 Avg. Training loss: 4.2745 0.2660 sec/batch
Epoch 9/10 Iteration: 100300 Avg. Training loss: 4.3124 0.2739 sec/batch
Epoch 9/10 Iteration: 100400 Avg. Training loss: 4.3305 0.2733 sec/batch
Epoch 9/10 Iteration: 100500 Avg. Training loss: 4.2886 0.2739 sec/batch
Epoch 9/10 Iteration: 100600 Avg. Training loss: 4.2182 0.2674 sec/batch
Epoch 9/10 Iteration: 100700 Avg. Training loss: 4.3029 0.2707 sec/batch
Epoch 9/10 Iteration: 100800 Avg. Training loss: 4.3183 0.2676 sec/batch
Epoch 9/10 Iteration: 100900 Avg. Training loss: 4.2769 0.2706 sec/batch
Epoch 9/10 Iteration: 101000 Avg. Training loss: 4.2828 0.2723 sec/batch
Nearest to into: back, then, further, divided, toward, categories, distancing, throats,
Nearest to during: late, prior, throughout, after, before, in, early, antiproton,
Nearest to about: regarding, main, approximately, around, told, gssps, nationalities, rugians,
Nearest to five: four, six, one, three, two, seven, zero, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, several, these, most, such, various, certain, other,
Nearest to however: but, although, though, because, that, while, since, it,
Nearest to war: ii, wwii, wars, outbreak, conflict, manassas, occupation, period,
Nearest to experience: experiences, subjective, gz, bearable, awaken, coincident, difficulty, diploma,
Nearest to magazine: magazines, coil, issue, macmullen, newspaper, monthly, skeptics, recoil,
Nearest to joseph: smith, unemployment, phylum, shop, sarah, ssen, erroneous, kite,
Nearest to universe: gallen, dune, constellations, nolan, exposure, prot, ehud, monkey,
Nearest to defense: ecclesiastic, guard, comix, ignition, dalek, tage, pitman, defence,
Nearest to square: dock, knee, wadis, westport, gonzaga, upland, shaper, recordable,
Nearest to egypt: egyptian, jericho, circa, cuneiform, handle, claw, receded, rameses,
Nearest to smith: joseph, mcm, geddes, garner, avoirdupois, filename, dildo, sisley,
Epoch 9/10 Iteration: 101100 Avg. Training loss: 4.3051 0.2705 sec/batch
Epoch 9/10 Iteration: 101200 Avg. Training loss: 4.2915 0.2674 sec/batch
Epoch 9/10 Iteration: 101300 Avg. Training loss: 4.3096 0.2693 sec/batch
Epoch 9/10 Iteration: 101400 Avg. Training loss: 4.2447 0.2711 sec/batch
Epoch 9/10 Iteration: 101500 Avg. Training loss: 4.1855 0.2651 sec/batch
Epoch 9/10 Iteration: 101600 Avg. Training loss: 4.2268 0.2676 sec/batch
Epoch 9/10 Iteration: 101700 Avg. Training loss: 4.2573 0.2683 sec/batch
Epoch 9/10 Iteration: 101800 Avg. Training loss: 4.2527 0.2662 sec/batch
Epoch 9/10 Iteration: 101900 Avg. Training loss: 4.2284 0.2718 sec/batch
Epoch 9/10 Iteration: 102000 Avg. Training loss: 4.3342 0.2674 sec/batch
Nearest to into: back, divided, then, categories, further, distancing, throats, toward,
Nearest to during: late, early, throughout, before, prior, after, in, beginning,
Nearest to about: regarding, main, approximately, thacker, depleted, nationalities, around, told,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, four, eight, one, three, seven, nine, two,
Nearest to some: many, these, several, most, certain, such, various, other,
Nearest to however: although, but, though, because, while, that, since, and,
Nearest to war: ii, outbreak, wars, wwii, manassas, conflict, occupation, campaign,
Nearest to experience: experiences, subjective, gz, subdivide, bearable, difficulty, isopropanol, awaken,
Nearest to magazine: magazines, coil, issue, macmullen, skeptics, monthly, newspaper, recoil,
Nearest to joseph: smith, unemployment, sarah, phylum, erroneous, shop, ssen, kite,
Nearest to universe: gallen, dune, constellations, nolan, ehud, fictional, exposure, prot,
Nearest to defense: guard, ignition, dalek, comix, ecclesiastic, tage, iea, defence,
Nearest to square: dock, knee, wadis, gonzaga, westport, shaper, recordable, tiananmen,
Nearest to egypt: egyptian, circa, jericho, handle, receded, cuneiform, claw, rameses,
Nearest to smith: joseph, geddes, mcm, dildo, sisley, avoirdupois, garner, filename,
Epoch 9/10 Iteration: 102100 Avg. Training loss: 4.0438 0.2706 sec/batch
Epoch 9/10 Iteration: 102200 Avg. Training loss: 4.1330 0.2713 sec/batch
Epoch 9/10 Iteration: 102300 Avg. Training loss: 4.1161 0.2692 sec/batch
Epoch 9/10 Iteration: 102400 Avg. Training loss: 4.2727 0.2712 sec/batch
Epoch 9/10 Iteration: 102500 Avg. Training loss: 4.3031 0.2675 sec/batch
Epoch 9/10 Iteration: 102600 Avg. Training loss: 4.3503 0.2687 sec/batch
Epoch 9/10 Iteration: 102700 Avg. Training loss: 4.2712 0.2741 sec/batch
Epoch 9/10 Iteration: 102800 Avg. Training loss: 4.3502 0.2739 sec/batch
Epoch 9/10 Iteration: 102900 Avg. Training loss: 4.3622 0.2699 sec/batch
Epoch 9/10 Iteration: 103000 Avg. Training loss: 4.3138 0.2705 sec/batch
Nearest to into: back, divided, categories, distancing, then, further, destry, onto,
Nearest to during: after, late, before, early, prior, throughout, antiproton, beginning,
Nearest to about: regarding, main, approximately, thacker, around, told, depleted, includes,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, several, these, most, such, certain, various, other,
Nearest to however: although, but, though, because, while, that, since, and,
Nearest to war: ii, wwii, outbreak, wars, manassas, conflict, occupation, campaign,
Nearest to experience: experiences, subjective, gz, bearable, subdivide, coincident, puerto, difficulty,
Nearest to magazine: coil, magazines, issue, skeptics, newspaper, recoil, monthly, macmullen,
Nearest to joseph: smith, unemployment, erroneous, phylum, sarah, moglen, ssen, kite,
Nearest to universe: gallen, dune, constellations, exposure, comics, fictional, nolan, ehud,
Nearest to defense: guard, comix, ecclesiastic, tage, ignition, dalek, defence, iea,
Nearest to square: dock, knee, wadis, westport, gonzaga, shaper, upland, recordable,
Nearest to egypt: egyptian, jericho, circa, receded, handle, claw, cuneiform, andrews,
Nearest to smith: joseph, avoirdupois, mcm, dildo, geddes, filename, garner, steepest,
Epoch 9/10 Iteration: 103100 Avg. Training loss: 4.2523 0.2713 sec/batch
Epoch 9/10 Iteration: 103200 Avg. Training loss: 4.1653 0.2697 sec/batch
Epoch 9/10 Iteration: 103300 Avg. Training loss: 4.2728 0.2663 sec/batch
Epoch 9/10 Iteration: 103400 Avg. Training loss: 4.2321 0.2674 sec/batch
Epoch 9/10 Iteration: 103500 Avg. Training loss: 4.3234 0.2667 sec/batch
Epoch 9/10 Iteration: 103600 Avg. Training loss: 4.2372 0.2694 sec/batch
Epoch 9/10 Iteration: 103700 Avg. Training loss: 4.3240 0.2735 sec/batch
Epoch 9/10 Iteration: 103800 Avg. Training loss: 4.2546 0.2698 sec/batch
Epoch 9/10 Iteration: 103900 Avg. Training loss: 4.1378 0.2691 sec/batch
Epoch 9/10 Iteration: 104000 Avg. Training loss: 4.0835 0.2669 sec/batch
Nearest to into: divided, back, further, categories, divisions, distancing, then, toward,
Nearest to during: late, after, early, prior, before, throughout, beginning, period,
Nearest to about: regarding, main, approximately, around, five, includes, zero, estimated,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, most, such, various, certain, other,
Nearest to however: although, but, though, while, because, that, since, and,
Nearest to war: ii, outbreak, wwii, wars, conflict, manassas, occupation, campaign,
Nearest to experience: experiences, subjective, gz, bearable, subdivide, puerto, difficulty, diploma,
Nearest to magazine: coil, magazines, skeptics, issue, newspaper, monthly, recoil, macmullen,
Nearest to joseph: smith, unemployment, phylum, sarah, ssen, kite, moglen, erroneous,
Nearest to universe: gallen, dune, fictional, constellations, exposure, comics, nolan, ehud,
Nearest to defense: guard, comix, defence, ignition, ecclesiastic, dalek, tage, energies,
Nearest to square: dock, knee, wadis, westport, gonzaga, grocery, tiananmen, everton,
Nearest to egypt: egyptian, jericho, receded, handle, circa, cuneiform, rameses, claw,
Nearest to smith: joseph, mcm, avoirdupois, dildo, geddes, garner, filename, sisley,
Epoch 9/10 Iteration: 104100 Avg. Training loss: 4.3080 0.2716 sec/batch
Epoch 9/10 Iteration: 104200 Avg. Training loss: 4.2236 0.2729 sec/batch
Epoch 9/10 Iteration: 104300 Avg. Training loss: 4.2394 0.2674 sec/batch
Epoch 9/10 Iteration: 104400 Avg. Training loss: 4.3006 0.2691 sec/batch
Epoch 9/10 Iteration: 104500 Avg. Training loss: 4.2719 0.2674 sec/batch
Epoch 9/10 Iteration: 104600 Avg. Training loss: 4.2168 0.2700 sec/batch
Epoch 9/10 Iteration: 104700 Avg. Training loss: 4.2918 0.2690 sec/batch
Epoch 9/10 Iteration: 104800 Avg. Training loss: 4.2853 0.2653 sec/batch
Epoch 9/10 Iteration: 104900 Avg. Training loss: 4.3540 0.2664 sec/batch
Epoch 10/10 Iteration: 105000 Avg. Training loss: 4.3267 0.1638 sec/batch
Nearest to into: back, divided, distancing, categories, divisions, further, then, eastbourne,
Nearest to during: late, early, before, after, prior, throughout, beginning, antiproton,
Nearest to about: regarding, main, approximately, thacker, rugians, around, what, homosexuals,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, four, eight, one, three, seven, nine, two,
Nearest to some: many, several, these, such, various, certain, most, other,
Nearest to however: although, but, though, because, that, while, since, despite,
Nearest to war: ii, wwii, outbreak, wars, conflict, manassas, veterans, occupation,
Nearest to experience: experiences, subjective, gz, bearable, subdivide, difficulty, coincident, isopropanol,
Nearest to magazine: coil, magazines, issue, skeptics, newspaper, recoil, macmullen, monthly,
Nearest to joseph: unemployment, smith, sarah, ssen, phylum, moglen, kite, erroneous,
Nearest to universe: gallen, fictional, constellations, exposure, comics, dune, nolan, manor,
Nearest to defense: guard, comix, ecclesiastic, dalek, defence, ignition, tage, energies,
Nearest to square: dock, knee, westport, wadis, gonzaga, tiananmen, stochastic, grocery,
Nearest to egypt: egyptian, jericho, circa, handle, receded, rameses, cuneiform, claw,
Nearest to smith: joseph, mcm, avoirdupois, dildo, geddes, garner, flagship, electing,
Epoch 10/10 Iteration: 105100 Avg. Training loss: 4.2392 0.2705 sec/batch
Epoch 10/10 Iteration: 105200 Avg. Training loss: 4.2686 0.2688 sec/batch
Epoch 10/10 Iteration: 105300 Avg. Training loss: 4.2902 0.2726 sec/batch
Epoch 10/10 Iteration: 105400 Avg. Training loss: 4.2357 0.2722 sec/batch
Epoch 10/10 Iteration: 105500 Avg. Training loss: 4.2670 0.2700 sec/batch
Epoch 10/10 Iteration: 105600 Avg. Training loss: 4.2270 0.2702 sec/batch
Epoch 10/10 Iteration: 105700 Avg. Training loss: 4.2497 0.2676 sec/batch
Epoch 10/10 Iteration: 105800 Avg. Training loss: 4.2733 0.2696 sec/batch
Epoch 10/10 Iteration: 105900 Avg. Training loss: 4.2974 0.2679 sec/batch
Epoch 10/10 Iteration: 106000 Avg. Training loss: 4.2656 0.2682 sec/batch
Nearest to into: then, back, divided, categories, further, onto, distancing, eastbourne,
Nearest to during: early, late, before, throughout, after, antiproton, prior, first,
Nearest to about: regarding, approximately, main, around, five, told, roughly, gssps,
Nearest to five: four, six, one, three, two, seven, zero, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, several, these, certain, such, various, most, other,
Nearest to however: although, but, though, because, that, while, since, and,
Nearest to war: ii, wwii, outbreak, wars, battle, manassas, veterans, campaign,
Nearest to experience: experiences, subjective, gz, bearable, subdivide, isopropanol, diploma, awaken,
Nearest to magazine: coil, magazines, issue, newspaper, skeptics, monthly, recoil, macmullen,
Nearest to joseph: smith, unemployment, kite, sarah, phylum, ssen, moglen, scarecrow,
Nearest to universe: gallen, dune, comics, fictional, constellations, nolan, exposure, monkey,
Nearest to defense: guard, ignition, dalek, comix, defence, tage, ecclesiastic, handoff,
Nearest to square: dock, knee, wadis, westport, tiananmen, grocery, gonzaga, stochastic,
Nearest to egypt: egyptian, circa, receded, cuneiform, jericho, rameses, handle, countenance,
Nearest to smith: joseph, mcm, avoirdupois, dildo, geddes, sisley, garner, flagship,
Epoch 10/10 Iteration: 106100 Avg. Training loss: 4.2384 0.2702 sec/batch
Epoch 10/10 Iteration: 106200 Avg. Training loss: 4.3357 0.2679 sec/batch
Epoch 10/10 Iteration: 106300 Avg. Training loss: 4.2648 0.2682 sec/batch
Epoch 10/10 Iteration: 106400 Avg. Training loss: 4.2838 0.2686 sec/batch
Epoch 10/10 Iteration: 106500 Avg. Training loss: 4.2774 0.2663 sec/batch
Epoch 10/10 Iteration: 106600 Avg. Training loss: 4.2170 0.2670 sec/batch
Epoch 10/10 Iteration: 106700 Avg. Training loss: 4.1869 0.2685 sec/batch
Epoch 10/10 Iteration: 106800 Avg. Training loss: 4.2818 0.2696 sec/batch
Epoch 10/10 Iteration: 106900 Avg. Training loss: 4.3066 0.2758 sec/batch
Epoch 10/10 Iteration: 107000 Avg. Training loss: 4.2884 0.2680 sec/batch
Nearest to into: then, divided, back, further, categories, onto, distancing, destry,
Nearest to during: throughout, late, prior, early, before, after, antiproton, first,
Nearest to about: regarding, approximately, main, roughly, around, gssps, thacker, rugians,
Nearest to five: four, six, two, one, three, zero, seven, eight,
Nearest to six: five, four, one, eight, three, seven, nine, two,
Nearest to some: many, these, several, certain, various, such, most, other,
Nearest to however: although, but, though, because, that, while, and, since,
Nearest to war: ii, wwii, battle, wars, manassas, outbreak, campaign, conflict,
Nearest to experience: experiences, subjective, gz, bearable, subdivide, isopropanol, awaken, puerto,
Nearest to magazine: coil, magazines, issue, skeptics, monthly, newspaper, macmullen, recoil,
Nearest to joseph: smith, unemployment, kite, sarah, moglen, erroneous, phylum, ssen,
Nearest to universe: gallen, exposure, dune, nolan, ehud, fictional, monkey, comics,
Nearest to defense: guard, ignition, dalek, defence, tage, ecclesiastic, comix, handoff,
Nearest to square: dock, knee, wadis, tiananmen, westport, garden, grocery, gonzaga,
Nearest to egypt: egyptian, circa, receded, jericho, cuneiform, handle, rameses, countenance,
Nearest to smith: joseph, mcm, avoirdupois, garner, geddes, dildo, sisley, worthwhile,
Epoch 10/10 Iteration: 107100 Avg. Training loss: 4.2963 0.2793 sec/batch
Epoch 10/10 Iteration: 107200 Avg. Training loss: 4.2911 0.2735 sec/batch
Epoch 10/10 Iteration: 107300 Avg. Training loss: 4.2897 0.2666 sec/batch
Epoch 10/10 Iteration: 107400 Avg. Training loss: 4.3287 0.2671 sec/batch
Epoch 10/10 Iteration: 107500 Avg. Training loss: 4.3038 0.2653 sec/batch
Epoch 10/10 Iteration: 107600 Avg. Training loss: 4.2470 0.2683 sec/batch
Epoch 10/10 Iteration: 107700 Avg. Training loss: 4.2436 0.2719 sec/batch
Epoch 10/10 Iteration: 107800 Avg. Training loss: 4.0593 0.2696 sec/batch
Epoch 10/10 Iteration: 107900 Avg. Training loss: 4.2098 0.2688 sec/batch
Epoch 10/10 Iteration: 108000 Avg. Training loss: 4.3156 0.2708 sec/batch
Nearest to into: divided, further, then, categories, distancing, in, throats, back,
Nearest to during: late, throughout, early, before, prior, after, period, beginning,
Nearest to about: regarding, approximately, main, gssps, depleted, thacker, around, five,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, three, seven, nine, two,
Nearest to some: many, these, several, certain, various, such, most, few,
Nearest to however: although, but, though, because, that, while, since, fact,
Nearest to war: ii, wwii, wars, outbreak, manassas, campaign, occupation, conflict,
Nearest to experience: experiences, subjective, gz, bearable, isopropanol, subdivide, awaken, coincident,
Nearest to magazine: coil, magazines, monthly, issue, skeptics, macmullen, newspaper, wired,
Nearest to joseph: smith, unemployment, phylum, sarah, moglen, kite, erroneous, shop,
Nearest to universe: gallen, exposure, dune, nolan, comics, ehud, ok, manor,
Nearest to defense: guard, ecclesiastic, defence, ignition, tage, dalek, comix, icbms,
Nearest to square: dock, knee, wadis, westport, tiananmen, shaper, grocery, gonzaga,
Nearest to egypt: egyptian, circa, handle, receded, cuneiform, rameses, jericho, countenance,
Nearest to smith: joseph, mcm, geddes, avoirdupois, dildo, sisley, garner, whealey,
Epoch 10/10 Iteration: 108100 Avg. Training loss: 4.2269 0.2714 sec/batch
Epoch 10/10 Iteration: 108200 Avg. Training loss: 4.2451 0.2693 sec/batch
Epoch 10/10 Iteration: 108300 Avg. Training loss: 4.2381 0.2675 sec/batch
Epoch 10/10 Iteration: 108400 Avg. Training loss: 4.1315 0.2672 sec/batch
Epoch 10/10 Iteration: 108500 Avg. Training loss: 4.2485 0.2672 sec/batch
Epoch 10/10 Iteration: 108600 Avg. Training loss: 4.2858 0.2787 sec/batch
Epoch 10/10 Iteration: 108700 Avg. Training loss: 4.2960 0.2728 sec/batch
Epoch 10/10 Iteration: 108800 Avg. Training loss: 4.2807 0.2732 sec/batch
Epoch 10/10 Iteration: 108900 Avg. Training loss: 4.2814 0.2859 sec/batch
Epoch 10/10 Iteration: 109000 Avg. Training loss: 4.2494 0.2820 sec/batch
Nearest to into: then, back, further, toward, divided, distancing, through, throats,
Nearest to during: late, after, before, early, prior, throughout, in, period,
Nearest to about: regarding, approximately, main, gssps, thacker, depleted, around, rugians,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, several, these, certain, various, such, most, other,
Nearest to however: although, but, though, because, that, while, since, and,
Nearest to war: ii, wwii, wars, outbreak, manassas, campaign, conflict, battle,
Nearest to experience: experiences, subjective, gz, awaken, bearable, isopropanol, subdivide, difficulty,
Nearest to magazine: coil, magazines, issue, skeptics, monthly, macmullen, newspaper, recoil,
Nearest to joseph: smith, phylum, sarah, unemployment, moglen, kite, ssen, scarecrow,
Nearest to universe: gallen, exposure, dune, ehud, nolan, everything, ok, comics,
Nearest to defense: guard, ignition, ecclesiastic, comix, tage, dalek, defence, civilians,
Nearest to square: dock, knee, wadis, tiananmen, recordable, westport, shaper, gonzaga,
Nearest to egypt: egyptian, circa, receded, handle, andrews, jericho, claw, cuneiform,
Nearest to smith: joseph, mcm, avoirdupois, geddes, sisley, dildo, worthwhile, flagship,
Epoch 10/10 Iteration: 109100 Avg. Training loss: 4.3066 0.2836 sec/batch
Epoch 10/10 Iteration: 109200 Avg. Training loss: 4.2856 0.2822 sec/batch
Epoch 10/10 Iteration: 109300 Avg. Training loss: 4.2405 0.2856 sec/batch
Epoch 10/10 Iteration: 109400 Avg. Training loss: 4.2868 0.3042 sec/batch
Epoch 10/10 Iteration: 109500 Avg. Training loss: 4.2395 0.2946 sec/batch
Epoch 10/10 Iteration: 109600 Avg. Training loss: 4.3062 0.2921 sec/batch
Epoch 10/10 Iteration: 109700 Avg. Training loss: 4.2981 0.2853 sec/batch
Epoch 10/10 Iteration: 109800 Avg. Training loss: 4.3307 0.2883 sec/batch
Epoch 10/10 Iteration: 109900 Avg. Training loss: 4.2953 0.2846 sec/batch
Epoch 10/10 Iteration: 110000 Avg. Training loss: 4.1947 0.2823 sec/batch
Nearest to into: back, then, divided, further, in, toward, distancing, categories,
Nearest to during: late, after, before, early, throughout, prior, period, antiproton,
Nearest to about: regarding, approximately, main, around, gssps, zero, five, depleted,
Nearest to five: four, six, one, two, three, zero, seven, eight,
Nearest to six: five, four, one, eight, seven, three, nine, two,
Nearest to some: many, these, several, certain, various, such, most, other,
Nearest to however: but, although, though, because, that, while, since, and,
Nearest to war: ii, wwii, wars, outbreak, campaign, occupation, manassas, battle,
Nearest to experience: experiences, subjective, gz, bearable, coincident, subdivide, awaken, isopropanol,
Nearest to magazine: magazines, issue, coil, newspaper, skeptics, monthly, macmullen, publication,
Nearest to joseph: smith, unemployment, moglen, phylum, erroneous, sarah, shop, ssen,
Nearest to universe: gallen, dune, exposure, nolan, fictional, ehud, monkey, everything,
Nearest to defense: guard, defence, dalek, comix, ecclesiastic, ignition, civilians, pitman,
Nearest to square: dock, wadis, knee, tiananmen, westport, blessed, gonzaga, recordable,
Nearest to egypt: egyptian, handle, circa, receded, andrews, cuneiform, claw, jericho,
Nearest to smith: joseph, mcm, avoirdupois, dildo, geddes, worthwhile, sisley, flagship,
Epoch 10/10 Iteration: 110100 Avg. Training loss: 4.2780 0.2826 sec/batch
Epoch 10/10 Iteration: 110200 Avg. Training loss: 4.2035 0.2804 sec/batch
Epoch 10/10 Iteration: 110300 Avg. Training loss: 4.2942 0.2826 sec/batch
Epoch 10/10 Iteration: 110400 Avg. Training loss: 4.2988 0.2924 sec/batch
Epoch 10/10 Iteration: 110500 Avg. Training loss: 4.2455 0.2854 sec/batch
Epoch 10/10 Iteration: 110600 Avg. Training loss: 4.3050 0.2841 sec/batch
Epoch 10/10 Iteration: 110700 Avg. Training loss: 4.2802 0.2810 sec/batch
Epoch 10/10 Iteration: 110800 Avg. Training loss: 4.2375 0.2824 sec/batch
Epoch 10/10 Iteration: 110900 Avg. Training loss: 4.2670 0.2812 sec/batch
Epoch 10/10 Iteration: 111000 Avg. Training loss: 4.2289 0.2840 sec/batch
Nearest to into: back, divided, then, distancing, further, onto, categories, throats,
Nearest to during: late, after, before, throughout, period, prior, early, antiproton,
Nearest to about: regarding, main, approximately, what, depleted, gssps, told, around,
Nearest to five: four, six, one, seven, three, two, eight, zero,
Nearest to six: five, one, four, eight, seven, nine, three, two,
Nearest to some: many, these, several, certain, most, various, such, other,
Nearest to however: but, although, though, because, that, while, since, and,
Nearest to war: ii, wwii, outbreak, wars, campaign, manassas, veterans, occupation,
Nearest to experience: experiences, subjective, gz, awaken, bearable, coincident, difficulty, isopropanol,
Nearest to magazine: magazines, issue, coil, newspaper, monthly, skeptics, publication, macmullen,
Nearest to joseph: smith, sarah, unemployment, phylum, shop, moglen, ssen, kite,
Nearest to universe: gallen, dune, exposure, nolan, ehud, monkey, manor, fictional,
Nearest to defense: dalek, ecclesiastic, comix, ignition, guard, defence, tage, pitman,
Nearest to square: dock, wadis, knee, shaper, westport, gonzaga, tiananmen, grocery,
Nearest to egypt: egyptian, handle, andrews, claw, jericho, receded, circa, cuneiform,
Nearest to smith: joseph, mcm, avoirdupois, dildo, lila, dungeons, geddes, sisley,
Epoch 10/10 Iteration: 111100 Avg. Training loss: 4.3183 0.2877 sec/batch
Epoch 10/10 Iteration: 111200 Avg. Training loss: 4.1560 0.2809 sec/batch
Epoch 10/10 Iteration: 111300 Avg. Training loss: 4.2939 0.2830 sec/batch
Epoch 10/10 Iteration: 111400 Avg. Training loss: 4.2958 0.2799 sec/batch
Epoch 10/10 Iteration: 111500 Avg. Training loss: 4.2645 0.2820 sec/batch
Epoch 10/10 Iteration: 111600 Avg. Training loss: 4.1110 0.2892 sec/batch
Epoch 10/10 Iteration: 111700 Avg. Training loss: 4.3194 0.2871 sec/batch
Epoch 10/10 Iteration: 111800 Avg. Training loss: 4.2991 0.2897 sec/batch
Epoch 10/10 Iteration: 111900 Avg. Training loss: 4.2653 0.2841 sec/batch
Epoch 10/10 Iteration: 112000 Avg. Training loss: 4.3020 0.2902 sec/batch
Nearest to into: back, further, then, divided, categories, toward, onto, throats,
Nearest to during: throughout, after, late, before, prior, period, in, antiproton,
Nearest to about: regarding, main, approximately, around, thacker, what, told, discography,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, these, several, certain, most, various, such, other,
Nearest to however: but, although, though, that, because, while, since, it,
Nearest to war: ii, wars, wwii, outbreak, campaign, conflict, occupation, manassas,
Nearest to experience: experiences, subjective, gz, bearable, awaken, coincident, difficulty, subdivide,
Nearest to magazine: issue, magazines, coil, monthly, newspaper, skeptics, macmullen, publication,
Nearest to joseph: smith, unemployment, sarah, phylum, ssen, shop, erroneous, kite,
Nearest to universe: gallen, dune, exposure, nolan, ehud, ok, unicef, manor,
Nearest to defense: defence, dalek, guard, comix, ignition, ecclesiastic, tage, pitman,
Nearest to square: dock, wadis, knee, gonzaga, grocery, shaper, westport, blessed,
Nearest to egypt: egyptian, handle, jericho, circa, claw, receded, cuneiform, andrews,
Nearest to smith: joseph, mcm, avoirdupois, dungeons, dildo, sisley, oswald, flagship,
Epoch 10/10 Iteration: 112100 Avg. Training loss: 4.3500 0.2857 sec/batch
Epoch 10/10 Iteration: 112200 Avg. Training loss: 4.2449 0.2874 sec/batch
Epoch 10/10 Iteration: 112300 Avg. Training loss: 4.2389 0.2823 sec/batch
Epoch 10/10 Iteration: 112400 Avg. Training loss: 4.3159 0.2855 sec/batch
Epoch 10/10 Iteration: 112500 Avg. Training loss: 4.2911 0.2817 sec/batch
Epoch 10/10 Iteration: 112600 Avg. Training loss: 4.2513 0.2861 sec/batch
Epoch 10/10 Iteration: 112700 Avg. Training loss: 4.2901 0.3048 sec/batch
Epoch 10/10 Iteration: 112800 Avg. Training loss: 4.2810 0.2999 sec/batch
Epoch 10/10 Iteration: 112900 Avg. Training loss: 4.3006 0.2951 sec/batch
Epoch 10/10 Iteration: 113000 Avg. Training loss: 4.2613 0.2909 sec/batch
Nearest to into: back, further, then, divided, onto, throats, categories, distancing,
Nearest to during: late, throughout, prior, before, after, period, in, around,
Nearest to about: regarding, main, approximately, around, what, zero, told, iqs,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, these, several, certain, most, such, various, other,
Nearest to however: but, although, though, because, while, that, since, only,
Nearest to war: ii, wars, wwii, outbreak, campaign, conflict, occupation, manassas,
Nearest to experience: experiences, subjective, awaken, gz, difficulty, bearable, coincident, subdivide,
Nearest to magazine: coil, magazines, issue, monthly, newspaper, skeptics, macmullen, recoil,
Nearest to joseph: smith, phylum, sarah, unemployment, ssen, shop, scarecrow, kite,
Nearest to universe: dune, gallen, fictional, nolan, comics, exposure, ehud, ok,
Nearest to defense: dalek, comix, defence, ignition, guard, ecclesiastic, tage, civilians,
Nearest to square: dock, knee, wadis, tiananmen, gonzaga, blessed, westport, shaper,
Nearest to egypt: egyptian, jericho, circa, handle, cuneiform, claw, receded, andrews,
Nearest to smith: joseph, avoirdupois, mcm, dildo, flagship, dungeons, worthwhile, geddes,
Epoch 10/10 Iteration: 113100 Avg. Training loss: 4.1689 0.2883 sec/batch
Epoch 10/10 Iteration: 113200 Avg. Training loss: 4.2596 0.2858 sec/batch
Epoch 10/10 Iteration: 113300 Avg. Training loss: 4.2456 0.2859 sec/batch
Epoch 10/10 Iteration: 113400 Avg. Training loss: 4.2707 0.2862 sec/batch
Epoch 10/10 Iteration: 113500 Avg. Training loss: 4.2391 0.2846 sec/batch
Epoch 10/10 Iteration: 113600 Avg. Training loss: 4.2831 0.2970 sec/batch
Epoch 10/10 Iteration: 113700 Avg. Training loss: 4.2551 0.2885 sec/batch
Epoch 10/10 Iteration: 113800 Avg. Training loss: 3.9714 0.2846 sec/batch
Epoch 10/10 Iteration: 113900 Avg. Training loss: 4.1254 0.2866 sec/batch
Epoch 10/10 Iteration: 114000 Avg. Training loss: 4.2200 0.2883 sec/batch
Nearest to into: back, then, toward, further, distancing, onto, categories, eastbourne,
Nearest to during: throughout, before, late, after, prior, early, in, period,
Nearest to about: regarding, main, told, approximately, what, around, thacker, depleted,
Nearest to five: four, six, one, three, two, seven, zero, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, these, several, most, certain, such, various, other,
Nearest to however: although, but, though, because, that, while, since, despite,
Nearest to war: ii, wars, outbreak, wwii, campaign, battle, manassas, conflict,
Nearest to experience: experiences, subjective, gz, awaken, puerto, bearable, distracting, subdivide,
Nearest to magazine: issue, magazines, monthly, coil, published, newspaper, publication, skeptics,
Nearest to joseph: smith, phylum, sarah, unemployment, moglen, ssen, erroneous, shop,
Nearest to universe: gallen, dune, fictional, ehud, nolan, exposure, comics, ok,
Nearest to defense: dalek, comix, defence, guard, ignition, tage, ecclesiastic, civilians,
Nearest to square: dock, knee, wadis, blessed, tiananmen, gonzaga, shaper, westport,
Nearest to egypt: egyptian, jericho, handle, circa, cuneiform, claw, receded, rameses,
Nearest to smith: joseph, avoirdupois, dildo, mcm, lila, flagship, garner, steepest,
Epoch 10/10 Iteration: 114100 Avg. Training loss: 4.2558 0.2885 sec/batch
Epoch 10/10 Iteration: 114200 Avg. Training loss: 4.3037 0.2840 sec/batch
Epoch 10/10 Iteration: 114300 Avg. Training loss: 4.2805 0.2872 sec/batch
Epoch 10/10 Iteration: 114400 Avg. Training loss: 4.3431 0.2877 sec/batch
Epoch 10/10 Iteration: 114500 Avg. Training loss: 4.3658 0.2865 sec/batch
Epoch 10/10 Iteration: 114600 Avg. Training loss: 4.3383 0.2846 sec/batch
Epoch 10/10 Iteration: 114700 Avg. Training loss: 4.2855 0.2881 sec/batch
Epoch 10/10 Iteration: 114800 Avg. Training loss: 4.1517 0.2874 sec/batch
Epoch 10/10 Iteration: 114900 Avg. Training loss: 4.2656 0.2934 sec/batch
Epoch 10/10 Iteration: 115000 Avg. Training loss: 4.2973 0.2857 sec/batch
Nearest to into: further, onto, back, divided, categories, distancing, toward, through,
Nearest to during: late, throughout, after, before, early, prior, period, antiproton,
Nearest to about: regarding, main, approximately, what, thacker, around, discography, told,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, four, one, eight, seven, three, nine, two,
Nearest to some: many, these, several, most, certain, such, various, other,
Nearest to however: although, but, though, that, because, while, since, only,
Nearest to war: ii, outbreak, wwii, wars, conflict, campaign, manassas, veterans,
Nearest to experience: experiences, subjective, gz, bearable, subdivide, euphoria, coincident, distracting,
Nearest to magazine: coil, magazines, issue, monthly, skeptics, newspaper, publication, recoil,
Nearest to joseph: smith, phylum, unemployment, sarah, moglen, erroneous, hydrolysis, ssen,
Nearest to universe: gallen, dune, comics, nolan, exposure, ehud, fictional, ok,
Nearest to defense: defence, guard, comix, dalek, ignition, tage, civilians, ecclesiastic,
Nearest to square: dock, knee, wadis, gonzaga, blessed, tiananmen, grocery, shaper,
Nearest to egypt: egyptian, handle, jericho, receded, cuneiform, claw, rameses, andrews,
Nearest to smith: joseph, avoirdupois, dildo, mcm, lila, flagship, geddes, steepest,
Epoch 10/10 Iteration: 115100 Avg. Training loss: 4.2270 0.2834 sec/batch
Epoch 10/10 Iteration: 115200 Avg. Training loss: 4.2532 0.2849 sec/batch
Epoch 10/10 Iteration: 115300 Avg. Training loss: 4.3257 0.2866 sec/batch
Epoch 10/10 Iteration: 115400 Avg. Training loss: 4.2499 0.2878 sec/batch
Epoch 10/10 Iteration: 115500 Avg. Training loss: 4.2659 0.2860 sec/batch
Epoch 10/10 Iteration: 115600 Avg. Training loss: 4.0685 0.2847 sec/batch
Epoch 10/10 Iteration: 115700 Avg. Training loss: 4.2337 0.2873 sec/batch
Epoch 10/10 Iteration: 115800 Avg. Training loss: 4.2770 0.2912 sec/batch
Epoch 10/10 Iteration: 115900 Avg. Training loss: 4.1541 0.2901 sec/batch
Epoch 10/10 Iteration: 116000 Avg. Training loss: 4.2565 0.2854 sec/batch
Nearest to into: further, onto, back, divided, toward, distancing, categories, eastbourne,
Nearest to during: before, late, after, prior, throughout, period, early, antiproton,
Nearest to about: regarding, main, approximately, what, thacker, around, liberian, includes,
Nearest to five: four, six, one, two, three, seven, zero, eight,
Nearest to six: five, one, four, eight, seven, three, nine, two,
Nearest to some: many, these, several, certain, various, most, such, other,
Nearest to however: although, but, though, that, because, while, despite, and,
Nearest to war: ii, outbreak, wars, wwii, campaign, manassas, veterans, battle,
Nearest to experience: experiences, subjective, gz, bearable, euphoria, subdivide, awaken, coincident,
Nearest to magazine: coil, magazines, monthly, issue, skeptics, newspaper, publication, macmullen,
Nearest to joseph: smith, sarah, unemployment, phylum, moglen, scarecrow, ssen, kite,
Nearest to universe: gallen, dune, comics, ehud, exposure, ok, nolan, fictional,
Nearest to defense: defence, comix, dalek, guard, ignition, tage, ecclesiastic, civilians,
Nearest to square: dock, knee, wadis, grocery, tiananmen, blessed, stochastic, westport,
Nearest to egypt: egyptian, handle, jericho, cuneiform, receded, claw, rameses, andrews,
Nearest to smith: joseph, avoirdupois, dildo, mcm, flagship, lila, geddes, lancaster,
Epoch 10/10 Iteration: 116100 Avg. Training loss: 4.2612 0.2883 sec/batch
Epoch 10/10 Iteration: 116200 Avg. Training loss: 4.2712 0.2852 sec/batch
Epoch 10/10 Iteration: 116300 Avg. Training loss: 4.2564 0.2853 sec/batch
Epoch 10/10 Iteration: 116400 Avg. Training loss: 4.2684 0.2835 sec/batch
Epoch 10/10 Iteration: 116500 Avg. Training loss: 4.3113 0.2852 sec/batch
Epoch 10/10 Iteration: 116600 Avg. Training loss: 4.3255 0.2858 sec/batch

Restore the trained network if you need to:



In [ ]:

    
with train_graph.as_default():
    saver = tf.train.Saver()

with tf.Session(graph=train_graph) as sess:
    saver.restore(sess, tf.train.latest_checkpoint('checkpoints'))
    embed_mat = sess.run(embedding)

Visualizing the word vectors

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 to learn more about T-SNE and other ways to visualize high-dimensional data.



In [16]:

    
%matplotlib inline
%config InlineBackend.figure_format = 'retina'

import matplotlib.pyplot as plt
from sklearn.manifold import TSNE



In [17]:

    
viz_words = 500
tsne = TSNE()
embed_tsne = tsne.fit_transform(embed_mat[:viz_words, :])



In [18]:

    
fig, ax = plt.subplots(figsize=(14, 14))
for idx in range(viz_words):
    plt.scatter(*embed_tsne[idx, :], color='steelblue')
    plt.annotate(int_to_vocab[idx], (embed_tsne[idx, 0], embed_tsne[idx, 1]), alpha=0.7)



In [ ]: