In [1]:

    

from __future__ import print_function
import collections
import numpy as np
import pandas as pd
import random
import math
import pickle
import sklearn
from sklearn.model_selection import train_test_split
from sklearn.manifold import TSNE

import tensorflow as tf

import gensim
import logging

from matplotlib import pylab
%matplotlib inline

# fix random seed for reproducibility
np.random.seed(7)

from pprint import pprint


    

In [2]:

    

data_file = "dump_2017_words.txt"


    

In [3]:

    

############################################################
# 1. load data
############################################################

# load dataset
print("Loading data...")
data=[]
y=[]

# count words
c = collections.Counter()

count = 0
with open(data_file, "r") as infile:
    for line in infile:
        count += 1
        if count % 50000 == 0:
            print (count)

        l = line.rstrip('\n').split(",")
        # status is first element
        label = l.pop(0)
        # headline is rest
        
        data.append(l)
        y.append(label)
        c.update(l)
        
print("Loaded %d lines." % count)


    

    




Loading data...
50000
100000
150000
200000
Loaded 218419 lines.

In [4]:

    

# dump data as one giant text string

with open("dump_words.txt", "w") as outfile:
    for headline in data:
        for word in headline:
            outfile.write(word)
            outfile.write(" ")


    

In [5]:

    

def read_data(filename):
  """Extract the file as a list of words"""
  with open(filename) as f:
    data = tf.compat.as_str(f.read()).split()
  return data

filename = "dump_words.txt"
words = read_data(filename)
print('Data size %d' % len(words))


    

    




Data size 2196850

In [8]:

    

############################################################
# build dataset
############################################################

vocabulary_size = 10000 # pick this to have ~10 occurrences for least popular

def build_dataset(words):
  count = [['UNK', -1]]
  count.extend(collections.Counter(words).most_common(vocabulary_size - 1))
  dictionary = dict()
  for word, _ in count:
    dictionary[word] = len(dictionary)
  data = list()
  unk_count = 0
  for word in words:
    if word in dictionary:
      index = dictionary[word]
    else:
      index = 0  # dictionary['UNK']
      unk_count = unk_count + 1
    data.append(index)
  count[0][1] = unk_count
  reverse_dictionary = dict(zip(dictionary.values(), dictionary.keys()))
  return data, count, dictionary, reverse_dictionary

data, count, dictionary, reverse_dictionary = build_dataset(words)
print('Most common words (+UNK)', count[:100])
print("\n")
print('Least common words', count[-10:])
print("\n")
print('Sample data', data[:100])


    

    




Most common words (+UNK) [['UNK', 168108], ('domain_otherdomain', 119708), ('subsource_othersubsource', 47862), ('trump', 21141), ('with', 10761), ('domain_youtube.com', 8908), ('us', 8434), ('2017', 7862), ('from', 7768), ('subsource_memeorandum', 7712), ('how', 7301), ('say', 6944), ('new', 6923), ('have', 6614), ('domain_bloomberg.com', 6053), ('domain_nytimes.com', 5645), ('domain_washingtonpost.com', 4919), ('will', 4720), ('about', 4594), ('domain_reuters.com', 4520), ('you', 4509), ('do', 4408), ('after', 4406), ('domain_wsj.com', 4323), ('subsource_SpirosMargaris', 4314), ('that', 4290), ('what', 4273), ('subsource_DRUDGE_REPORT', 4242), ('why', 4216), ('subsource_pdacosta', 4108), ('not', 4033), ('up', 3993), ('over', 3932), ('make', 3918), ('get', 3908), ('your', 3696), ('domain_theguardian.com', 3618), ('domain_ft.com', 3560), ('its', 3538), ('this', 3454), ('subsource_JohnLothian', 3175), ('pdf', 3170), ('subsource_thegarance', 3148), ('year', 3100), ('utc', 3064), ('out', 3058), ('world', 3018), ('more', 2979), ('china', 2927), ('president', 2913), ('subsource_lrozen', 2893), ('his', 2855), ('woman', 2834), ('report', 2821), ('take', 2782), ('can', 2778), ('may', 2767), ('subsource_aleks_kins', 2721), ('could', 2704), ('big', 2695), ('day', 2688), ('who', 2678), ('donald_trump', 2678), ('first', 2601), ('go', 2598), ('source_Huffington_Post', 2596), ('domain_businessinsider.com', 2595), ('plan', 2582), ('market', 2579), ('one', 2576), ('subsource_moorehn', 2571), ('brexit', 2523), ('time', 2485), ('instagram_post', 2406), ('russia', 2398), ('news', 2395), ('subsource_maggieNYT', 2380), ('state', 2376), ('american', 2375), ('subsource_MissTrade', 2329), ('election', 2306), ('show', 2273), ('call', 2200), ('all', 2198), ('but', 2180), ('now', 2178), ('into', 2159), ('just', 2141), ('want', 2133), ('job', 2133), ('source_Abnormal_Returns', 2124), ('back', 2108), ('domain_cnbc.com', 2106), ('people', 2103), ('uk', 2103), ('million', 2083), ('domain_thehill.com', 2045), ('subsource_HotlineJosh', 2012), ('business', 1997), ('work', 1974)]


Least common words [('hazard', 17), ('alexei', 17), ('molly', 17), ('expel', 17), ('champ', 17), ('admiral', 17), ('conversational', 17), ('memorable', 17), ('wharton', 17), ('torn', 17)]


Sample data [1, 108, 0, 864, 0, 8153, 8057, 6095, 285, 0, 502, 5598, 1, 108, 0, 168, 29, 3, 2976, 272, 378, 39, 72, 2519, 45, 120, 1865, 292, 0, 164, 2822, 23, 2259, 6223, 1, 2, 2675, 4205, 1602, 0, 0, 359, 133, 9, 3337, 220, 216, 6500, 7382, 75, 954, 53, 1, 641, 2135, 56, 443, 3790, 2852, 2112, 12, 43, 60, 1, 991, 535, 1641, 2850, 693, 7, 0, 44, 1, 656, 541, 178, 189, 84, 74, 13, 38, 1379, 7732, 1, 108, 0, 139, 0, 0, 0, 1, 1970, 529, 601, 35, 5905, 218, 21, 33, 4836]

In [9]:

    

############################################################
# function to generate mini batches
############################################################

# generate pairs of words that are collocated as two parallel lists
data_index = 0

def generate_batch(batch_size, num_skips, skip_window):
  global data_index
  assert batch_size % num_skips == 0
  assert num_skips <= 2 * skip_window
  batch = np.ndarray(shape=(batch_size), dtype=np.int32)
  labels = np.ndarray(shape=(batch_size, 1), dtype=np.int32)
  span = 2 * skip_window + 1 # [ skip_window target skip_window ]
  buffer = collections.deque(maxlen=span)
  for _ in range(span):
    buffer.append(data[data_index])
    data_index = (data_index + 1) % len(data)
  for i in range(batch_size // num_skips):
    target = skip_window  # target label at the center of the buffer
    targets_to_avoid = [ skip_window ]
    for j in range(num_skips):
      while target in targets_to_avoid:
        target = random.randint(0, span - 1)
      targets_to_avoid.append(target)
      batch[i * num_skips + j] = buffer[skip_window]
      labels[i * num_skips + j, 0] = buffer[target]
    buffer.append(data[data_index])
    data_index = (data_index + 1) % len(data)
  return batch, labels

print('data:', [reverse_dictionary[di] for di in data[:8]])

for num_skips, skip_window in [(2, 1), (4, 2)]:
    data_index = 0
    batch, labels = generate_batch(batch_size=8, num_skips=num_skips, skip_window=skip_window)
    print('\nwith num_skips = %d and skip_window = %d:' % (num_skips, skip_window))
    print('    batch:', [reverse_dictionary[bi] for bi in batch])
    print('    labels:', [reverse_dictionary[li] for li in labels.reshape(8)])


    

    




data: ['domain_otherdomain', 'subsource_newsycombinator', 'UNK', 'hack', 'UNK', 'demonstrate', 'linux', 'steam']

with num_skips = 2 and skip_window = 1:
    batch: ['subsource_newsycombinator', 'subsource_newsycombinator', 'UNK', 'UNK', 'hack', 'hack', 'UNK', 'UNK']
    labels: ['domain_otherdomain', 'UNK', 'hack', 'subsource_newsycombinator', 'UNK', 'UNK', 'demonstrate', 'hack']

with num_skips = 4 and skip_window = 2:
    batch: ['UNK', 'UNK', 'UNK', 'UNK', 'hack', 'hack', 'hack', 'hack']
    labels: ['hack', 'UNK', 'subsource_newsycombinator', 'domain_otherdomain', 'demonstrate', 'subsource_newsycombinator', 'UNK', 'UNK']

In [10]:

    

# Initialize embeddings to random values or Google News embedddings where available

embedding_size = 300 # Dimension of the embedding vector.
print ("Initializing random embeddings")
initial_vals = np.float32(np.random.uniform(-1,1,size=(vocabulary_size,embedding_size)))

# Look up Google News embeddings and replace where they exist
print ("Loading Google News embedding model")
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)

# do this once each session, takes a while
try:
    model
except NameError:
    model = gensim.models.Word2Vec.load_word2vec_format('~/mount/ubuntu/pretrained/GoogleNews-vectors-negative300.bin', binary=True)

#copy file above to ramdisk to maybe speedup
#sudo mount -t tmpfs -o size=4096m tmpfs /home/ubuntu/mount/ramdisk

#print(model['computer'])
#print(dictionary['computer'])
#print(reverse_dictionary[2297])
#model.similarity('computer', 'PC')
#print(initial_vals[2297])

print ("Re-initializing to Google News embedding where available")
# for each word in model, use its embeddings as initial val

count = 0
for (v,k) in reverse_dictionary.items():
    if k in model:
        count+=1
        if (count % 1000) == 0:
            print ("Re-initialized %d words" % count)
        initial_vals[v] = model[k]

print ("Re-initialized %d of %d words" % (count,vocabulary_size))
# delete model which takes up a lot of memory (5-6 GB)
del model


    

    




2017-07-18 20:16:59,318 : INFO : loading projection weights from ~/mount/ubuntu/pretrained/GoogleNews-vectors-negative300.bin






    




Initializing random embeddings
Loading Google News embedding model






    




2017-07-18 20:18:39,318 : INFO : loaded (3000000, 300) matrix from ~/mount/ubuntu/pretrained/GoogleNews-vectors-negative300.bin






    




Re-initializing to Google News embedding where available
Re-initialized 1000 words
Re-initialized 2000 words
Re-initialized 3000 words
Re-initialized 4000 words
Re-initialized 5000 words
Re-initialized 6000 words
Re-initialized 7000 words
Re-initialized 7296 of 10000 words

In [11]:

    

############################################################
# word2vec computation graph
############################################################

batch_size = 128
skip_window = 1 # How many words to consider left and right.
num_skips = 2 # How many times to reuse an input to generate a label.
# We pick a random validation set to sample nearest neighbors. here we limit the
# validation samples to the words that have a low numeric ID, which by
# construction are also the most frequent.
valid_size = 20 # Random set of words to evaluate similarity on.
valid_window = 100 # Only pick dev samples in the head of the distribution.
valid_examples = np.array(random.sample(range(valid_window), valid_size))
num_sampled = 64 # Number of negative examples to sample.

graph = tf.Graph()

with graph.as_default():
    #, tf.device('/cpu:0'): 
    # unsure why I have to pin to CPU to make it work
    # defeats the purpose of tensorflow if you can't use GPU
    
    # Input data.
    train_dataset = tf.placeholder(tf.int32, shape=[batch_size])
    train_labels = tf.placeholder(tf.int32, shape=[batch_size, 1])
    valid_dataset = tf.constant(valid_examples, dtype=tf.int32)
    tf_learning_rate = tf.placeholder(tf.float32)
    
    # Variables.
    embeddings = tf.Variable(initial_vals)
    # (initialize to random or Google News value from previous cell)
    # tf.random_uniform([vocabulary_size, embedding_size], -1.0, 1.0))
    softmax_weights = tf.Variable(tf.truncated_normal([vocabulary_size, embedding_size],
                                                      stddev=1.0 / math.sqrt(embedding_size)))
    softmax_biases = tf.Variable(tf.zeros([vocabulary_size]))
    
    # Model.
    # Look up embeddings for inputs.
    embed = tf.nn.embedding_lookup(embeddings, train_dataset)
    # Compute the softmax loss, using a sample of the negative labels each time.
    loss = tf.reduce_mean(tf.nn.sampled_softmax_loss(softmax_weights, softmax_biases, train_labels, embed,
                                                     num_sampled, vocabulary_size))
    
    # Optimizer.
    # Note: The optimizer will optimize the softmax_weights AND the embeddings.
    optimizer = tf.train.AdagradOptimizer(tf_learning_rate).minimize(loss)
    
    # Compute the similarity between minibatch examples and all embeddings.
    # We use the cosine distance:
    norm = tf.sqrt(tf.reduce_sum(tf.square(embeddings), 1, keep_dims=True))
    normalized_embeddings = embeddings / norm
    valid_embeddings = tf.nn.embedding_lookup(normalized_embeddings, valid_dataset)
    similarity = tf.matmul(valid_embeddings, tf.transpose(normalized_embeddings))
    
    # initialize op
    init_op = tf.global_variables_initializer()


    

In [12]:

    

############################################################
# run computation graph
############################################################

num_steps = 1200001

# high initial learning rate, reduce gradually
initial_lr = 1.0
final_lr = 0.1
#final_lr = 0.01
lr_reduce_rate=(final_lr / initial_lr) ** (1.0/num_steps)
print("lr_reduce_rate: %.10f" % lr_reduce_rate)

with tf.Session(graph=graph) as session:
    session.run(init_op)
    average_loss = 0
    lr = initial_lr
    
    for step in range(num_steps):
        batch_data, batch_labels = generate_batch(batch_size, num_skips, skip_window)
        feed_dict = {train_dataset : batch_data, 
                     train_labels : batch_labels, 
                     tf_learning_rate : lr,}
        _, l = session.run([optimizer, loss], 
                           feed_dict=feed_dict,)
        average_loss += l
        if step % 2000 == 0:
            if step > 0:
                average_loss = average_loss / 2000
                # The average loss is an estimate of the loss over the last 2000 batches.
                print('Average loss at step %d: %f' % (step, average_loss))
            average_loss = 0
            
        # note that this is expensive (~20% slowdown if computed every 500 steps)
        if step % 10000 == 0:
            sim = similarity.eval()
            for i in range(valid_size):
                valid_word = reverse_dictionary[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 = reverse_dictionary[nearest[k]]
                    log = '%s %s,' % (log, close_word)
                print(log)
        lr *= lr_reduce_rate
        
    final_embeddings = normalized_embeddings.eval()


    

    




lr_reduce_rate: 0.9999980812
Nearest to job: work, employment, salary, internship, hiring, position, apprenticeship, hire,
Nearest to go: come, get, do, sit, stay, take, put, walk,
Nearest to make: making, made, get, give, create, bring, produce, deliver,
Nearest to who: whose, fellow, young, also, veteran, former, him, personally,
Nearest to after: before, later, last, earlier, when, ago, prior, during,
Nearest to from: after, where, through, outside, away, within, around, before,
Nearest to million: billion, mln, 1bn, 2bn, bln, percent, 7bn, total,
Nearest to donald_trump: rubio, palin, hillary, boehner, lindsey, hillary_clinton, mccarthy, mccain,
Nearest to utc: pmi, mattis, gma, rba, vix, davidson, norway, alexis,
Nearest to may: might, could, should, can, will, would, must, likely,
Nearest to domain_youtube.com: calendar, presenter, freelance, lifestyle, clock, buzzfeed, subsource_Jeffrey_Cane, constellation,
Nearest to first: second, third, fourth, fifth, sixth, last, final, ninth,
Nearest to domain_washingtonpost.com: equipment, lindzon, subsource_steve_hanke, subsource_jotted, subsource_NaomiAKlein, star, unit, logistics,
Nearest to domain_otherdomain: poop, wearable, range, diplomatic, battery, haul, yen, subsource_carlquintanilla,
Nearest to subsource_DRUDGE_REPORT: human, subsource_GrantLewis1, patch, replay, study, sized, source_Huffington_Post, kalanick,
Nearest to domain_wsj.com: format, qui, petersburg, entre, pit, tactic, ici, word,
Nearest to domain_ft.com: subsource_josheidelson, victor, subsource_lheron, twist, celebration, cord, anniversary, peaceful,
Nearest to subsource_pdacosta: surgeon, knife, smoking, weapon, firearm, meddle, subsource_reidepstein, politicalbetting,
Nearest to source_Huffington_Post: spotify, dispatch, structure, subsource_TaylorLorenz, subsource_DRUDGE_REPORT, 1977, subsource_GregorMacdonald, recon,
Nearest to domain_nytimes.com: subsource_davidgaffen, pose, subsource_BretBaier, relax, showtime, subsource_janzilinsky, lax, match,
Average loss at step 2000: 4.082999
Average loss at step 4000: 3.751330
Average loss at step 6000: 3.651899
Average loss at step 8000: 3.606206
Average loss at step 10000: 3.555524
Nearest to job: employment, hire, position, hiring, salary, staff, skilled, internship,
Nearest to go: come, happen, get, coming, wait, sit, turn, roll,
Nearest to make: made, making, give, become, create, deliver, get, build,
Nearest to who: whose, young, still, wont, never, they, try, she,
Nearest to after: over, during, before, amid, while, despite, since, from,
Nearest to from: after, through, outside, with, via, across, over, into,
Nearest to million: billion, mln, percent, bln, 2bn, 1bn, nearly, total,
Nearest to donald_trump: trump, trumps, rubio, obama, democrat, barack_obama, they, hillary_clinton,
Nearest to utc: 2017, pdf, feb, pmi, nach, jpg, claude, thu,
Nearest to may: might, will, must, could, should, would, likely, wont,
Nearest to domain_youtube.com: domain_otherdomain, domain_washingtonpost.com, inside, tencent, subsource_DeliveringAlpha, domain_ft.com, domain_miamiherald.com, known,
Nearest to first: second, third, fourth, fifth, inaugural, last, sixth, 1st,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_youtube.com, domain_thehill.com, amid, subsource_NaomiAKlein, despite,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_reuters.com, domain_washingtonpost.com, domain_nytimes.com, domain_theguardian.com, domain_wsj.com, how,
Nearest to subsource_DRUDGE_REPORT: patch, replay, subsource_GrantLewis1, subsource_HeerJeet, riot, subsource_ElliotTurn, subsource_burke_jason, compound,
Nearest to domain_wsj.com: domain_otherdomain, subsource_aleks_kins, euronext, early, domain_nytimes.com, format, real, domain_reuters.com,
Nearest to domain_ft.com: domain_bloomberg.com, subsource_josheidelson, great, domain_reuters.com, domain_independent.co.uk, why_do, sadiq, domain_youtube.com,
Nearest to subsource_pdacosta: subsource_JohnLothian, subsource_lrozen, opinion, subsource_memeorandum, subsource_reidepstein, subsource_KimZetter, former, subsource_jbarro,
Nearest to source_Huffington_Post: source_NYT_Dealbook, govern, source_Business_Insider, topple, dispatch, spotify, structure, command,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_forbes.com, domain_thehill.com, domain_businessinsider.com, domain_wsj.com, domain_theonion.com, domain_bloomberg.com,
Average loss at step 12000: 3.529433
Average loss at step 14000: 3.479088
Average loss at step 16000: 3.460511
Average loss at step 18000: 3.419363
Average loss at step 20000: 3.410041
Nearest to job: employment, salary, worker, workforce, wage, employer, hiring, productivity,
Nearest to go: come, move, walk, stay, sit, happen, coming, know,
Nearest to make: making, made, give, become, create, convince, transform, prove,
Nearest to who: whose, young, also, try, dont, still, actually, cant,
Nearest to after: before, amid, during, despite, while, over, because, due,
Nearest to from: after, through, outside, into, onto, during, behind, via,
Nearest to million: billion, 1bn, mln, 2bn, bln, 7bn, nearly, 5bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, barack_obama, obama, she, him, democrat,
Nearest to utc: 2017, pdf, html, jpg, apr, htm, del, nach,
Nearest to may: might, will, could, should, wont, must, would, can,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_washingtonpost.com, domain_wsj.com, domain_bloomberg.com, domain_theguardian.com, domain_ft.com, domain_businessinsider.com,
Nearest to first: second, fourth, third, fifth, 1st, last, sixth, inaugural,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_wsj.com, domain_politico.com, domain_bloomberg.com, domain_thehill.com, domain_youtube.com, domain_www.cnn.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_medium.com, domain_huffingtonpost.com, domain_wsj.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, patch, replay, breaking, subsource_burke_jason, subsource_thegarance, subsource_helentbbc, subsource_AnnCoulter,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_washingtonpost.com, domain_otherdomain, domain_bloomberg.com, domain_ft.com, domain_youtube.com, subsource_aleks_kins,
Nearest to domain_ft.com: domain_bloomberg.com, domain_wsj.com, domain_reuters.com, domain_independent.co.uk, domain_otherdomain, domain_youtube.com, domain_nytimes.com, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_memeorandum, subsource_JohnLothian, domain_reuters.com, why, subsource_GreekAnalyst, subsource_sdonnan, subsource_george_chen, subsource_lrozen,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_Here_Is_The_City, source_Abnormal_Returns, source_othersource, topple, govern, dispatch,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_youtube.com, domain_theguardian.com, domain_reuters.com, domain_businessinsider.com, domain_variety.com,
Average loss at step 22000: 3.392385
Average loss at step 24000: 3.381102
Average loss at step 26000: 3.361254
Average loss at step 28000: 3.358935
Average loss at step 30000: 3.347666
Nearest to job: employment, worker, workforce, salary, hiring, employer, internship, factory,
Nearest to go: come, sit, turn, stuff, happen, coming, hell, always,
Nearest to make: create, made, making, earn, convince, achieve, build, become,
Nearest to who: young, whose, actually, also, dont, allegedly, try, elderly,
Nearest to after: amid, despite, during, near, over, within, because, before,
Nearest to from: after, through, during, into, via, outside, thanks, then,
Nearest to million: billion, 1bn, 2bn, mln, bln, trillion, 7bn, 5bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, macron, him, barack_obama, rubio, she,
Nearest to utc: 2017, pdf, html, shtml, htm, jpg, del, png,
Nearest to may: feb, could, might, wont, apr, jan, will, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_bloomberg.com, domain_medium.com, domain_ft.com, domain_washingtonpost.com, domain_politico.com, domain_wsj.com,
Nearest to first: second, fourth, 1st, fifth, third, inaugural, sixth, 2nd,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_politico.com, domain_nytimes.com, domain_wsj.com, domain_www.cnn.com, domain_theguardian.com, domain_reuters.com, domain_businessinsider.com,
Nearest to domain_otherdomain: domain_theguardian.com, domain_washingtonpost.com, domain_youtube.com, domain_wsj.com, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_maggieNYT, replay, subsource_whpresscorps, breaking, subsource_cfarivar, late,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_otherdomain, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_cnbc.com, domain_politico.com,
Nearest to domain_ft.com: domain_reuters.com, domain_otherdomain, domain_wsj.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_bbc.co.uk, domain_washingtonpost.com,
Nearest to subsource_pdacosta: former, subsource_JohnLothian, meddle, analysis, subsource_AmyResnick, wary, treasury, subsource_GTCost,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Abnormal_Returns, source_reddit/Economics, source_Here_Is_The_City, source_othersource, owen, iain,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_wsj.com, domain_otherdomain, domain_reuters.com, domain_youtube.com, domain_ft.com, domain_theguardian.com, domain_bloomberg.com,
Average loss at step 32000: 3.335136
Average loss at step 34000: 3.289589
Average loss at step 36000: 3.177596
Average loss at step 38000: 3.191615
Average loss at step 40000: 3.199860
Nearest to job: employment, salary, career, internship, workforce, worker, hiring, employer,
Nearest to go: come, stuff, walk, probably, coming, turn, happen, skip,
Nearest to make: making, made, give, earn, remind, save, inspire, convince,
Nearest to who: whose, able, young, also, actually, dont, allegedly, still,
Nearest to after: amid, despite, over, because, during, wake, due, within,
Nearest to from: through, outside, thanks, after, hailing, via, made, based,
Nearest to million: billion, 1bn, mln, 2bn, trillion, bln, 5bn, 7bn,
Nearest to donald_trump: trump, trumps, macron, barack_obama, hillary_clinton, britain, rubio, conway,
Nearest to utc: 2017, pdf, htm, shtml, html, jpg, del, png,
Nearest to may: might, could, wont, will, likely, able, cant, try,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_businessinsider.com, domain_politico.com, domain_techcrunch.com, domain_usatoday.com, domain_nytimes.com, domain_dailymail.co.uk,
Nearest to first: second, fourth, 1st, fifth, third, sixth, 2nd, inaugural,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_thehill.com, domain_www.cnn.com, domain_wsj.com, domain_theatlantic.com, domain_reuters.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_reuters.com, domain_medium.com, domain_bloomberg.com, domain_nytimes.com, domain_wsj.com, domain_thehill.com,
Nearest to subsource_DRUDGE_REPORT: subsource_instapundit, subsource_memeorandum, subsource_jonnajarian, replay, subsource_thegarance, subsource_whpresscorps, subsource_benshapiro, wound,
Nearest to domain_wsj.com: domain_nytimes.com, domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_ft.com, domain_washingtonpost.com, domain_cnbc.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_otherdomain, domain_businessinsider.com, domain_nytimes.com, domain_bbc.co.uk, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_JohnLothian, subsource_thegarance, subsource_jessefelder, subsource_AmyResnick, subsource_george_chen, wary, brazil,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_Abnormal_Returns, source_Here_Is_The_City, source_reddit/Economics, source_othersource, lauren, create,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_wsj.com, domain_otherdomain, domain_bloomberg.com, domain_ft.com, domain_politico.com, domain_thehill.com, domain_reuters.com,
Average loss at step 42000: 3.183637
Average loss at step 44000: 3.182395
Average loss at step 46000: 3.165450
Average loss at step 48000: 3.166611
Average loss at step 50000: 3.154944
Nearest to job: employment, worker, workforce, internship, salary, hiring, productivity, tenure,
Nearest to go: come, sit, stay, enter, stuff, skip, walk, hell,
Nearest to make: making, made, earn, convince, give, create, enjoy, steer,
Nearest to who: whose, able, young, allegedly, also, innocent, prosecute, disabled,
Nearest to after: amid, despite, prompt, during, wake, due, earlier, before,
Nearest to from: onto, outside, after, then, via, through, front, seek,
Nearest to million: billion, 1bn, mln, 2bn, 000, 5bn, bln, 7bn,
Nearest to donald_trump: trump, trumps, barack_obama, conway, hillary_clinton, him, marco_rubio, democrat,
Nearest to utc: 2017, pdf, shtml, html, htm, jpg, png, del,
Nearest to may: might, could, will, wont, would, likely, able, probably,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_businessinsider.com, domain_cnbc.com, domain_theguardian.com, domain_bloomberg.com, domain_politico.com, domain_wsj.com,
Nearest to first: 1st, second, fourth, fifth, 2nd, third, sixth, inaugural,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_politico.com, domain_otherdomain, domain_thehill.com, domain_wsj.com, domain_reuters.com, domain_bloomberg.com, domain_www.cnn.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_theguardian.com, domain_washingtonpost.com, domain_nytimes.com, domain_ft.com, domain_bloomberg.com, domain_medium.com, domain_reuters.com,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_whpresscorps, subsource_memeorandum, subsource_maggieNYT, subsource_instapundit, replay, breaking, novartis,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_washingtonpost.com, domain_otherdomain, domain_bloomberg.com, domain_ft.com, domain_politico.com, domain_cnbc.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bloomberg.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_businessinsider.com, domain_cnbc.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_AdamPosen, subsource_JohnLothian, subsource_TimAeppel, subsource_AmyResnick, subsource_thegarance, why, subsource_Frances_Coppola, subsource_jessefelder,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Abnormal_Returns, source_Here_Is_The_City, source_reddit/Economics, source_othersource, owen, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_wsj.com, domain_otherdomain, domain_reuters.com, domain_theguardian.com, domain_politico.com, domain_latimes.com, domain_bloomberg.com,
Average loss at step 52000: 3.146661
Average loss at step 54000: 3.154162
Average loss at step 56000: 3.139187
Average loss at step 58000: 3.153710
Average loss at step 60000: 3.133538
Nearest to job: worker, internship, employment, salary, factory, workforce, hiring, payroll,
Nearest to go: come, roar, stuff, sit, hell, coming, enter, crawl,
Nearest to make: earn, convince, making, give, steer, made, achieve, pave,
Nearest to who: whose, allegedly, actually, young, also, elderly, prosecute, able,
Nearest to after: during, despite, amid, within, because, prompt, earlier, before,
Nearest to from: thanks, then, attempt, via, during, accord, onto, outside,
Nearest to million: billion, 1bn, 2bn, mln, 000, 5bn, 7bn, bln,
Nearest to donald_trump: trump, trumps, barack_obama, hillary_clinton, marco_rubio, conway, macron, democrat,
Nearest to utc: 2017, pdf, shtml, html, htm, jpg, png, philippe,
Nearest to may: feb, apr, jul, jan, mar, june, jun, might,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_www.cnn.com, domain_cnbc.com, domain_bloomberg.com, domain_dailymail.co.uk, domain_thedailybeast.com,
Nearest to first: 1st, fourth, second, fifth, 2nd, sixth, inaugural, maiden,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_theguardian.com, domain_theatlantic.com, domain_reuters.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_theguardian.com, domain_bloomberg.com, domain_medium.com, domain_nytimes.com, domain_usatoday.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_maggieNYT, replay, subsource_instapundit, subsource_jonnajarian, subsource_benshapiro,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_otherdomain, domain_cnbc.com, domain_marketwatch.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_cnbc.com, domain_businessinsider.com, domain_otherdomain, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_xeni, subsource_lrozen, subsource_Frances_Coppola, subsource_TimAeppel, subsource_thegarance, subsource_GlennKesslerWP, subsource_AmyResnick, subsource_memeorandum,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_othersource, iain, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_thehill.com, domain_theguardian.com, domain_youtube.com, domain_huffingtonpost.com, domain_economist.com,
Average loss at step 62000: 3.141616
Average loss at step 64000: 3.125544
Average loss at step 66000: 3.148371
Average loss at step 68000: 3.068284
Average loss at step 70000: 3.055412
Nearest to job: employment, internship, factory, workforce, worker, productivity, hiring, career,
Nearest to go: come, roar, run, walk, always, happen, hell, belong,
Nearest to make: making, convince, achieve, earn, transform, steer, build, spare,
Nearest to who: able, allegedly, also, dont, young, actually, whose, prosecute,
Nearest to after: despite, amid, during, because, before, wake, earlier, prompt,
Nearest to from: thanks, via, hailing, through, outside, made, into, after,
Nearest to million: billion, mln, 1bn, 000, 2bn, 5bn, thousand, bln,
Nearest to donald_trump: trump, trumps, macron, barack_obama, marco_rubio, hillary_clinton, conway, obamas,
Nearest to utc: 2017, pdf, shtml, htm, jpg, html, png, mp3,
Nearest to may: might, could, wont, will, likely, would, cant, able,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_www.cnn.com, domain_businessinsider.com, domain_theguardian.com, domain_bloomberg.com, domain_dailymail.co.uk,
Nearest to first: 1st, fourth, second, 2nd, fifth, sixth, inaugural, final,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_politico.com, domain_otherdomain, domain_www.cnn.com, domain_reuters.com, domain_thehill.com, domain_yahoo.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_medium.com, domain_theguardian.com, domain_washingtonpost.com, domain_bloomberg.com, domain_wsj.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_instapundit, subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_maggieNYT, subsource_jonnajarian, replay, subsource_AnnCoulter,
Nearest to domain_wsj.com: domain_otherdomain, domain_ft.com, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_politico.com, domain_cnbc.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_wsj.com, domain_otherdomain, domain_reuters.com, domain_washingtonpost.com, domain_cnbc.com, domain_businessinsider.com, domain_bbc.co.uk,
Nearest to subsource_pdacosta: subsource_memeorandum, subsource_lrozen, subsource_Frances_Coppola, subsource_AmyResnick, subsource_thegarance, subsource_TimAeppel, subsource_george_chen, subsource_jessefelder,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Marginal_Revolution, source_othersource, owen,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_politico.com, domain_youtube.com, domain_thehill.com, domain_economist.com, domain_vox.com,
Average loss at step 72000: 3.063107
Average loss at step 74000: 3.056354
Average loss at step 76000: 3.054032
Average loss at step 78000: 3.049367
Average loss at step 80000: 3.035456
Nearest to job: employment, internship, worker, workforce, salary, career, productivity, payroll,
Nearest to go: come, roar, enter, descend, belong, really, hell, crawl,
Nearest to make: earn, made, making, give, achieve, pave, convince, steer,
Nearest to who: allegedly, whose, young, able, innocent, actually, dont, also,
Nearest to after: amid, despite, prompt, within, wake, before, earlier, briefly,
Nearest to from: via, onto, based, sends, after, bound, then, outside,
Nearest to million: billion, 1bn, mln, 000, 2bn, 5bn, thousand, billions,
Nearest to donald_trump: trump, trumps, marco_rubio, barack_obama, conway, christie, hillary_clinton, franken,
Nearest to utc: 2017, pdf, shtml, htm, jpg, html, png, mp3,
Nearest to may: could, might, will, wont, probably, able, likely, would,
Nearest to domain_youtube.com: domain_otherdomain, domain_cnbc.com, domain_medium.com, domain_www.cnn.com, domain_thedailybeast.com, domain_bloomberg.com, domain_nytimes.com, domain_washingtonpost.com,
Nearest to first: 1st, second, fourth, 2nd, fifth, sixth, inaugural, pivotal,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_thehill.com, domain_yahoo.com, domain_www.cnn.com, domain_wsj.com, domain_reuters.com,
Nearest to domain_otherdomain: domain_bloomberg.com, domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_cnbc.com, domain_medium.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_instapundit, subsource_thegarance, subsource_whpresscorps, subsource_AnnCoulter, subsource_maggieNYT, subsource_deray, subsource_memeorandum, subsource_xeni,
Nearest to domain_wsj.com: domain_reuters.com, domain_otherdomain, domain_washingtonpost.com, domain_nytimes.com, domain_bloomberg.com, domain_ft.com, domain_cnbc.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_businessinsider.com, domain_telegraph.co.uk, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_dandrezner, subsource_JohnLothian, subsource_TimAeppel, subsource_sdonnan, subsource_AdamPosen, subsource_memeorandum,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_othersource, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_bloomberg.com, domain_huffingtonpost.com, domain_npr.org, domain_apnews.com, domain_politico.com,
Average loss at step 82000: 3.043668
Average loss at step 84000: 3.022660
Average loss at step 86000: 3.036595
Average loss at step 88000: 3.029610
Average loss at step 90000: 3.029005
Nearest to job: worker, internship, workforce, employment, salary, payroll, hiring, tenure,
Nearest to go: come, roar, crawl, stuff, enter, belong, hell, run,
Nearest to make: making, earn, steer, convince, rake, pave, give, enjoy,
Nearest to who: allegedly, also, able, whose, young, prosecute, elderly, convict,
Nearest to after: amid, despite, because, briefly, unexpectedly, from, before, earlier,
Nearest to from: after, onto, thanks, attempt, via, because, into, during,
Nearest to million: billion, 1bn, mln, 2bn, 5bn, 000, 7bn, bln,
Nearest to donald_trump: trump, barack_obama, trumps, marco_rubio, conway, hillary_clinton, alex_jones, franken,
Nearest to utc: 2017, shtml, pdf, htm, jpg, html, png, mp3,
Nearest to may: apr, feb, jul, jan, jun, mar, june, might,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_dailymail.co.uk, domain_washingtonpost.com, domain_nationalreview.com, domain_www.cnn.com, domain_washingtonexaminer.com,
Nearest to first: 1st, fourth, 2nd, sixth, second, maiden, fifth, pivotal,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_www.cnn.com, domain_wsj.com, domain_businessinsider.com, domain_theatlantic.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_medium.com, domain_theguardian.com, domain_nytimes.com, domain_telegraph.co.uk, domain_reuters.com, domain_wsj.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_benshapiro, subsource_maggieNYT, subsource_whpresscorps, subsource_nycjim, subsource_DividendMaster, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_reuters.com, domain_washingtonpost.com, domain_ft.com, domain_otherdomain, domain_nytimes.com, domain_bloomberg.com, domain_politico.com, domain_cnbc.com,
Nearest to domain_ft.com: domain_reuters.com, domain_wsj.com, domain_bbc.co.uk, domain_bloomberg.com, domain_otherdomain, domain_cnbc.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_xeni, subsource_TimAeppel, subsource_maggieNYT, subsource_lrozen, subsource_edwardnh, subsource_GlennKesslerWP,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_Abnormal_Returns, source_Here_Is_The_City, source_reddit/Economics, source_othersource, source_Incidental_Economist, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_apnews.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_foxnews.com, domain_nymag.com,
Average loss at step 92000: 3.031374
Average loss at step 94000: 3.025721
Average loss at step 96000: 3.018120
Average loss at step 98000: 3.035126
Average loss at step 100000: 3.024996
Nearest to job: employment, internship, career, workforce, salary, worker, apprenticeship, payroll,
Nearest to go: roar, come, cry, stuff, trickle, sneak, crawl, turn,
Nearest to make: steer, making, create, convince, spare, give, transform, achieve,
Nearest to who: allegedly, able, also, prosecute, accuse, portman, accidentally, unable,
Nearest to after: amid, despite, prompt, because, during, wake, within, briefly,
Nearest to from: via, onto, thanks, after, into, hailing, during, outside,
Nearest to million: billion, 000, mln, 1bn, 2bn, 5bn, bln, billions,
Nearest to donald_trump: trump, marco_rubio, trumps, conway, barack_obama, kerry, hillary_clinton, christie,
Nearest to utc: 2017, shtml, pdf, htm, jpg, html, png, mp3,
Nearest to may: feb, might, could, hammond, jul, june, apr, april,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_businessinsider.com, domain_dailymail.co.uk, domain_theguardian.com, domain_bloomberg.com, domain_nydailynews.com,
Nearest to first: 1st, fourth, 2nd, maiden, second, fifth, sixth, 7th,
Nearest to domain_washingtonpost.com: domain_politico.com, domain_nytimes.com, domain_www.cnn.com, domain_theatlantic.com, domain_thehill.com, domain_wsj.com, domain_theguardian.com, domain_otherdomain,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_medium.com, domain_theguardian.com, domain_bloomberg.com, domain_forbes.com, domain_reuters.com, domain_washingtonpost.com,
Nearest to subsource_DRUDGE_REPORT: subsource_instapundit, subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_jonnajarian, novartis, subsource_aussietorres, subsource_vcutrader,
Nearest to domain_wsj.com: domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_nytimes.com, domain_washingtonpost.com, domain_otherdomain, domain_apnews.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_bbc.co.uk, domain_businessinsider.com, domain_economist.com, domain_theguardian.com, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_TimAeppel, subsource_thegarance, subsource_memeorandum, subsource_sdonnan, subsource_lrozen, subsource_Frances_Coppola, subsource_maggieNYT, subsource_prchovanec,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_othersource, source_Marginal_Revolution, source_Incidental_Economist,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_politico.com, domain_youtube.com, domain_ft.com, domain_huffingtonpost.com, domain_vox.com,
Average loss at step 102000: 2.955046
Average loss at step 104000: 2.983352
Average loss at step 106000: 2.977136
Average loss at step 108000: 2.967884
Average loss at step 110000: 2.972482
Nearest to job: tenure, internship, career, apprenticeship, worker, employment, salary, payroll,
Nearest to go: roar, come, hell, crawl, stuff, turn, really, cry,
Nearest to make: making, give, made, steer, rake, remake, earn, convince,
Nearest to who: allegedly, able, also, prosecute, actually, dont, young, disabled,
Nearest to after: amid, despite, briefly, wake, over, prompt, during, earlier,
Nearest to from: onto, after, bound, sends, then, made, via, thanks,
Nearest to million: billion, mln, 1bn, 000, 2bn, bln, 5bn, billions,
Nearest to donald_trump: trump, trumps, barack_obama, marco_rubio, conway, alex_jones, franken, hillary_clinton,
Nearest to utc: 2017, shtml, htm, pdf, jpg, html, mp3, png,
Nearest to may: could, might, will, wont, cant, able, aim, probably,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_nytimes.com, domain_politico.com, domain_apnews.com, domain_bloomberg.com,
Nearest to first: 1st, 2nd, second, fourth, 7th, sixth, 4th, 5th,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_otherdomain, domain_www.cnn.com, domain_reuters.com, domain_yahoo.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_medium.com, domain_bloomberg.com, domain_reuters.com, domain_forbes.com, domain_washingtonpost.com, domain_wsj.com, domain_theguardian.com,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_instapundit, subsource_whpresscorps, subsource_deray, subsource_jonnajarian, subsource_memeorandum, subsource_vcutrader, subsource_valuewalk,
Nearest to domain_wsj.com: domain_reuters.com, domain_otherdomain, domain_nytimes.com, domain_ft.com, domain_washingtonpost.com, domain_bloomberg.com, domain_marketwatch.com, domain_cnbc.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_bbc.co.uk, domain_businessinsider.com, domain_marketwatch.com, domain_economist.com, domain_otherdomain,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_SaraEisen, subsource_TimAeppel, wary, subsource_JohnLothian, subsource_MESandbu, subsource_memeorandum, subsource_MaxBoot,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Incidental_Economist, source_othersource, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_apnews.com, domain_politico.com, domain_nymag.com, domain_huffingtonpost.com, domain_youtube.com,
Average loss at step 112000: 2.963653
Average loss at step 114000: 2.961494
Average loss at step 116000: 2.950636
Average loss at step 118000: 2.954810
Average loss at step 120000: 2.955435
Nearest to job: worker, workforce, career, employment, tenure, payroll, internship, apprenticeship,
Nearest to go: roar, come, enter, really, crawl, turn, anyone, sail,
Nearest to make: making, convince, steer, give, earn, rake, trim, swallow,
Nearest to who: allegedly, whose, able, young, convict, disabled, prosecute, dont,
Nearest to after: amid, briefly, despite, abruptly, unexpectedly, prompt, earlier, over,
Nearest to from: onto, after, thanks, attempt, via, into, hub, bound,
Nearest to million: billion, 000, mln, 1bn, 5bn, 2bn, bln, 7bn,
Nearest to donald_trump: trump, trumps, barack_obama, marco_rubio, conway, christie, bush, hillary_clinton,
Nearest to utc: 2017, shtml, htm, pdf, jpg, html, png, mp3,
Nearest to may: jul, might, apr, june, feb, could, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_gizmodo.com, domain_businessinsider.com, domain_cnbc.com, domain_nypost.com, domain_washingtonpost.com,
Nearest to first: 1st, fourth, 2nd, second, sixth, maiden, fifth, 7th,
Nearest to domain_washingtonpost.com: domain_www.cnn.com, domain_nytimes.com, domain_otherdomain, domain_wsj.com, domain_politico.com, domain_thehill.com, domain_reuters.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_theguardian.com, domain_washingtonpost.com, domain_medium.com, domain_reuters.com, domain_bloomberg.com, domain_nytimes.com, domain_telegraph.co.uk,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_maggieNYT, subsource_whpresscorps, subsource_memeorandum, subsource_aleks_kins, subsource_SteveKopack, subsource_nycjim, subsource_benshapiro,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_washingtonpost.com, domain_ft.com, domain_otherdomain, domain_politico.com, domain_bloomberg.com, domain_apnews.com,
Nearest to domain_ft.com: domain_wsj.com, domain_bbc.co.uk, domain_otherdomain, domain_bloomberg.com, domain_cnbc.com, domain_reuters.com, domain_economist.com, domain_stlouisfed.org,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, wary, subsource_SaraEisen, subsource_thegarance, subsource_AdamPosen, subsource_acrossthecurve, subsource_memeorandum, skeptical,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_Abnormal_Returns, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_othersource, owen,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_wsj.com, domain_otherdomain, domain_apnews.com, domain_nymag.com, domain_fortune.com, domain_youtube.com, domain_politico.com,
Average loss at step 122000: 2.949776
Average loss at step 124000: 2.946943
Average loss at step 126000: 2.947582
Average loss at step 128000: 2.951504
Average loss at step 130000: 2.945474
Nearest to job: worker, internship, career, workforce, tenure, payroll, apprenticeship, outsource,
Nearest to go: roar, come, crawl, trickle, sneak, run, enter, coming,
Nearest to make: give, rake, steer, convince, earn, making, enjoy, spare,
Nearest to who: allegedly, dont, able, young, innocent, portman, elderly, somehow,
Nearest to after: amid, despite, prompt, follow, castile, over, within, briefly,
Nearest to from: onto, via, after, then, into, sends, thanks, driven,
Nearest to million: billion, 000, mln, 5bn, 1bn, 2bn, bln, billions,
Nearest to donald_trump: trump, marco_rubio, trumps, conway, hillary_clinton, barack_obama, christie, kerry,
Nearest to utc: 2017, shtml, pdf, htm, jpg, html, png, mp3,
Nearest to may: might, feb, apr, jul, could, june, jan, november,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_dailymail.co.uk, domain_www.cnn.com, domain_nationalreview.com, domain_nypost.com,
Nearest to first: 1st, 2nd, fourth, second, fifth, sixth, maiden, 7th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_politico.com, domain_nytimes.com, domain_thehill.com, domain_www.cnn.com, domain_theatlantic.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_theguardian.com, domain_medium.com, domain_bloomberg.com, domain_reuters.com, domain_wsj.com,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_maggieNYT, subsource_memeorandum, subsource_instapundit, subsource_vcutrader, subsource_jonnajarian, subsource_aussietorres, subsource_whpresscorps,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_ft.com, domain_otherdomain, domain_bloomberg.com, domain_washingtonpost.com, domain_apnews.com, domain_cnbc.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bloomberg.com, domain_wsj.com, domain_bbc.co.uk, domain_otherdomain, domain_cnbc.com, domain_businessinsider.com, domain_nytimes.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_prchovanec, subsource_sdonnan, subsource_TimAeppel, subsource_thegarance, subsource_SaraEisen, subsource_AdamPosen, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Abnormal_Returns, source_Here_Is_The_City, source_Incidental_Economist, source_othersource, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_politico.com, domain_theguardian.com, domain_ft.com, domain_apnews.com,
Average loss at step 132000: 2.967756
Average loss at step 134000: 2.902855
Average loss at step 136000: 2.921197
Average loss at step 138000: 2.923829
Average loss at step 140000: 2.909218
Nearest to job: tenure, career, internship, worker, apprenticeship, employee, employment, payroll,
Nearest to go: roar, come, crawl, descend, enter, skip, cry, really,
Nearest to make: give, making, steer, create, convince, remake, enjoy, rake,
Nearest to who: able, allegedly, prosecute, young, somehow, portman, disabled, convict,
Nearest to after: amid, despite, over, briefly, wake, follow, prompt, abruptly,
Nearest to from: onto, via, after, sends, thanks, bound, with, about,
Nearest to million: billion, mln, 000, 2bn, 1bn, bln, 5bn, 7bn,
Nearest to donald_trump: trump, trumps, marco_rubio, barack_obama, everyone, franken, kerry, nixon,
Nearest to utc: 2017, shtml, htm, pdf, jpg, html, mp3, png,
Nearest to may: could, might, wont, able, arent, cant, will, aim,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_dailymail.co.uk, domain_businessinsider.com, domain_thedailybeast.com, domain_nytimes.com, domain_zerohedge.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, fourth, second, 7th, pivotal, sixth, maiden,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_www.cnn.com, domain_otherdomain, domain_vox.com, domain_theatlantic.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_reuters.com, domain_forbes.com, domain_theguardian.com, domain_wsj.com, domain_ft.com, domain_nytimes.com, domain_bloomberg.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_instapundit, subsource_whpresscorps, subsource_jonnajarian, subsource_AnnCoulter, subsource_maggieNYT, subsource_vcutrader,
Nearest to domain_wsj.com: domain_otherdomain, domain_reuters.com, domain_nytimes.com, domain_bloomberg.com, domain_ft.com, domain_forbes.com, domain_marketwatch.com, domain_washingtonpost.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_businessinsider.com, domain_reuters.com, domain_bbc.co.uk, domain_cnbc.com, domain_economist.com, domain_wsj.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_TimAeppel, subsource_lrozen, subsource_xeni, subsource_SaraEisen, domain_businessinsider.com, subsource_edwardnh, subsource_JohnLothian,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_apnews.com, domain_wsj.com, domain_nymag.com, domain_politico.com, domain_fortune.com, domain_huffingtonpost.com,
Average loss at step 142000: 2.916046
Average loss at step 144000: 2.905891
Average loss at step 146000: 2.894968
Average loss at step 148000: 2.902310
Average loss at step 150000: 2.883105
Nearest to job: workforce, employment, worker, payroll, internship, career, tenure, outsource,
Nearest to go: roar, enter, really, crawl, come, borrow, sneak, skip,
Nearest to make: give, steer, convince, making, rake, pave, swallow, remake,
Nearest to who: allegedly, able, dont, convict, portman, somehow, young, innocent,
Nearest to after: amid, despite, briefly, prompt, before, abruptly, wake, follow,
Nearest to from: onto, after, beneath, sends, attempt, upon, strip, via,
Nearest to million: billion, mln, 000, 1bn, 5bn, 2bn, bln, 7bn,
Nearest to donald_trump: trump, trumps, marco_rubio, barack_obama, hillary_clinton, franken, conway, christie,
Nearest to utc: 2017, shtml, htm, pdf, jpg, html, mp3, mp4,
Nearest to may: might, could, will, able, likely, wont, probably, already,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_washingtonpost.com, domain_businessinsider.com, domain_npr.org, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, fourth, 7th, second, sixth, maiden, groundbreaking,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_wsj.com, domain_npr.org, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_theguardian.com, domain_wsj.com, domain_npr.org, domain_reuters.com, domain_bloomberg.com,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_whpresscorps, subsource_vcutrader, subsource_memeorandum, subsource_instapundit, subsource_SteveKopack, subsource_maggieNYT, subsource_aleks_kins,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_otherdomain, domain_washingtonpost.com, domain_politico.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_bbc.co.uk, domain_otherdomain, domain_reuters.com, domain_wsj.com, domain_cnbc.com, domain_economist.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_JohnLothian, subsource_TimAeppel, subsource_AdamPosen, subsource_SaraEisen, subsource_edwardnh, subsource_jessefelder,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_othersource, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_youtube.com, domain_nymag.com, domain_apnews.com, domain_politico.com, domain_fortune.com,
Average loss at step 152000: 2.899414
Average loss at step 154000: 2.887680
Average loss at step 156000: 2.900116
Average loss at step 158000: 2.888251
Average loss at step 160000: 2.896964
Nearest to job: worker, internship, payroll, workforce, career, employment, tenure, apprenticeship,
Nearest to go: roar, come, enter, crawl, sneak, turn, really, lean,
Nearest to make: rake, steer, give, convince, pave, making, undercut, earn,
Nearest to who: allegedly, prosecute, young, elderly, unable, able, dont, portman,
Nearest to after: amid, despite, briefly, spark, during, over, prompt, follow,
Nearest to from: onto, via, thanks, then, attempt, about, after, sends,
Nearest to million: billion, mln, 000, 1bn, bln, 2bn, 5bn, billions,
Nearest to donald_trump: trump, trumps, barack_obama, hillary_clinton, marco_rubio, kerry, conway, franken,
Nearest to utc: 2017, shtml, htm, pdf, jpg, html, mp4, png,
Nearest to may: feb, apr, jul, might, june, mar, jan, could,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_nationalreview.com, domain_dailymail.co.uk, domain_businessinsider.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, fourth, maiden, second, 7th, sixth, pivotal,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_politico.com, domain_nytimes.com, domain_thehill.com, domain_www.cnn.com, domain_vox.com, domain_reuters.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_businessinsider.com, domain_medium.com, domain_telegraph.co.uk,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_instapundit, subsource_maggieNYT, subsource_whpresscorps, subsource_jonnajarian, subsource_SteveKopack, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_otherdomain, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_forbes.com, domain_economist.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bbc.co.uk, domain_bloomberg.com, domain_wsj.com, domain_otherdomain, domain_businessinsider.com, domain_cnbc.com, domain_economist.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_TimAeppel, subsource_blakehounshell, subsource_Frances_Coppola, subsource_SaraEisen, subsource_GlennKesslerWP, subsource_xeni, subsource_lrozen,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Abnormal_Returns, source_Here_Is_The_City, source_Incidental_Economist, source_othersource, source_Marginal_Revolution,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_huffingtonpost.com, domain_theatlantic.com,
Average loss at step 162000: 2.881076
Average loss at step 164000: 2.910515
Average loss at step 166000: 2.881075
Average loss at step 168000: 2.867449
Average loss at step 170000: 2.867426
Nearest to job: payroll, employment, workforce, career, internship, apprenticeship, employee, outsource,
Nearest to go: roar, really, come, crawl, sneak, trickle, lean, enter,
Nearest to make: making, pave, give, rake, steer, convince, reshape, remake,
Nearest to who: allegedly, able, prosecute, unable, somehow, disabled, convict, young,
Nearest to after: amid, despite, over, briefly, because, wake, from, mistakenly,
Nearest to from: after, onto, via, hailing, about, driven, regard, bound,
Nearest to million: billion, 000, mln, 2bn, bln, 5bn, 1bn, 7bn,
Nearest to donald_trump: trump, trumps, marco_rubio, barack_obama, everyone, hillary_clinton, franken, kerry,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, mp3, png,
Nearest to may: might, could, will, wont, hammond, able, likely, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_www.cnn.com, domain_apnews.com, domain_businessinsider.com, domain_bloomberg.com, domain_foxnews.com,
Nearest to first: 1st, 2nd, 7th, fourth, second, maiden, fifth, groundbreaking,
Nearest to domain_washingtonpost.com: domain_www.cnn.com, domain_politico.com, domain_nytimes.com, domain_otherdomain, domain_thehill.com, domain_thedailybeast.com, domain_vox.com, domain_npr.org,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_wsj.com, domain_bloomberg.com, domain_theguardian.com, domain_medium.com, domain_reuters.com, domain_washingtonpost.com,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_AnnCoulter, subsource_SteveKopack, subsource_nycjim, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_economist.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_bbc.co.uk, domain_otherdomain, domain_reuters.com, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_SaraEisen, subsource_thegarance, subsource_TimAeppel, subsource_edwardnh, domain_businessinsider.com, subsource_jessefelder, subsource_AdamPosen,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_othersource, source_Incidental_Economist, owen,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_bloomberg.com, domain_apnews.com, domain_nymag.com,
Average loss at step 172000: 2.860737
Average loss at step 174000: 2.862627
Average loss at step 176000: 2.864311
Average loss at step 178000: 2.846651
Average loss at step 180000: 2.852244
Nearest to job: payroll, workforce, internship, employment, apprenticeship, tenure, worker, career,
Nearest to go: roar, come, crawl, borrow, really, enter, trickle, sneak,
Nearest to make: remake, rake, steer, making, pave, give, convince, enjoy,
Nearest to who: allegedly, able, young, prosecute, convict, unable, somehow, wrongly,
Nearest to after: amid, despite, briefly, prompt, over, abruptly, because, before,
Nearest to from: onto, via, after, sends, receives, upon, attempt, iconic,
Nearest to million: billion, 000, mln, 5bn, bln, 1bn, 2bn, 7bn,
Nearest to donald_trump: trump, trumps, franken, marco_rubio, hillary_clinton, barack_obama, christie, kerry,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, mp3, deutschland,
Nearest to may: could, might, will, november, probably, likely, hammond, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_bloomberg.com, domain_nytimes.com, domain_thedailybeast.com, domain_businessinsider.com, domain_cnbc.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, second, fourth, sixth, maiden, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_wsj.com, domain_vox.com, domain_nbcnews.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_whpresscorps, subsource_thegarance, subsource_memeorandum, subsource_instapundit, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_AnnCoulter,
Nearest to domain_wsj.com: domain_reuters.com, domain_otherdomain, domain_nytimes.com, domain_ft.com, domain_washingtonpost.com, domain_bloomberg.com, domain_cnbc.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_businessinsider.com, domain_wsj.com, domain_cnbc.com, domain_economist.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_JohnLothian, subsource_SaraEisen, subsource_TimAeppel, subsource_lrozen, subsource_jessefelder, subsource_prashantrao,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_othersource, source_Capital_Spectator,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_nymag.com, domain_youtube.com, domain_politico.com, domain_apnews.com, domain_vox.com,
Average loss at step 182000: 2.840337
Average loss at step 184000: 2.854223
Average loss at step 186000: 2.844261
Average loss at step 188000: 2.845083
Average loss at step 190000: 2.834999
Nearest to job: internship, tenure, payroll, worker, career, workforce, profession, apprenticeship,
Nearest to go: roar, come, crawl, trickle, really, stuff, mess, enter,
Nearest to make: rake, enjoy, steer, convince, making, give, pave, earn,
Nearest to who: allegedly, young, dont, convict, portman, able, prosecute, somehow,
Nearest to after: amid, despite, briefly, because, unexpectedly, before, over, abruptly,
Nearest to from: onto, attempt, after, about, via, thanks, into, driven,
Nearest to million: billion, mln, 000, bln, 5bn, 1bn, 2bn, billions,
Nearest to donald_trump: trump, trumps, hillary_clinton, barack_obama, marco_rubio, franken, bush, republican,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, html, png,
Nearest to may: jul, apr, feb, mar, jun, jan, june, might,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_dailymail.co.uk, domain_businessinsider.com, domain_www.cnn.com, domain_nationalreview.com, domain_theguardian.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, second, sixth, fifth,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_vox.com, domain_wsj.com, domain_theguardian.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_theguardian.com, domain_bloomberg.com, domain_medium.com, domain_nytimes.com, domain_businessinsider.com, domain_reuters.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_SteveKopack, subsource_DividendMaster, subsource_nycjim, subsource_maggieNYT, subsource_vcutrader,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_otherdomain, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_forbes.com, domain_nydailynews.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bbc.co.uk, domain_otherdomain, domain_bloomberg.com, domain_cnbc.com, domain_wsj.com, domain_economist.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_xeni, subsource_memeorandum, subsource_Frances_Coppola, subsource_TimAeppel, subsource_GlennKesslerWP, subsource_blakehounshell, subsource_SaraEisen,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Incidental_Economist, source_othersource, owen,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_theguardian.com, domain_politico.com,
Average loss at step 192000: 2.845431
Average loss at step 194000: 2.839780
Average loss at step 196000: 2.844117
Average loss at step 198000: 2.871805
Average loss at step 200000: 2.808517
Nearest to job: career, internship, workforce, apprenticeship, payroll, employment, employee, gig,
Nearest to go: roar, trickle, come, crawl, turn, sneak, really, walk,
Nearest to make: steer, remake, rake, making, convince, pave, redefine, give,
Nearest to who: allegedly, able, unable, portman, dont, somehow, prosecute, convict,
Nearest to after: amid, despite, over, briefly, follow, abruptly, before, during,
Nearest to from: onto, into, via, after, infamous, sends, regard, hailing,
Nearest to million: billion, 000, mln, bln, 5bn, 1bn, 2bn, 7bn,
Nearest to donald_trump: trump, trumps, marco_rubio, barack_obama, kerry, christie, conway, imperial,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, png, deutschland,
Nearest to may: could, might, feb, hammond, will, wont, jul, november,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_businessinsider.com, domain_medium.com, domain_www.cnn.com, domain_theguardian.com, domain_washingtonpost.com, domain_bloomberg.com,
Nearest to first: 1st, 2nd, 7th, fourth, maiden, 5th, second, sixth,
Nearest to domain_washingtonpost.com: domain_politico.com, domain_www.cnn.com, domain_nytimes.com, domain_otherdomain, domain_thehill.com, domain_yahoo.com, domain_thedailybeast.com, domain_npr.org,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_washingtonpost.com, domain_medium.com, domain_forbes.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_AnnCoulter, subsource_thegarance, subsource_DividendMaster, subsource_nycjim,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_economist.com, domain_wsj.com, domain_bbc.co.uk, domain_businessinsider.com, domain_otherdomain, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_memeorandum, subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_TimAeppel, subsource_edwardnh, subsource_mehdirhasan, subsource_AmyResnick,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Incidental_Economist, source_othersource, source_Capital_Spectator,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_youtube.com, domain_wsj.com, domain_politico.com, domain_ft.com, domain_fortune.com, domain_bloomberg.com,
Average loss at step 202000: 2.829050
Average loss at step 204000: 2.834911
Average loss at step 206000: 2.808803
Average loss at step 208000: 2.823247
Average loss at step 210000: 2.809838
Nearest to job: payroll, career, tenure, employment, apprenticeship, internship, disappointing, workforce,
Nearest to go: roar, come, crawl, trickle, really, enter, lean, borrow,
Nearest to make: remake, steer, give, convince, rake, pave, making, made,
Nearest to who: allegedly, dont, able, unable, disabled, somehow, cant, convict,
Nearest to after: briefly, amid, despite, abruptly, castile, mistakenly, over, unexpectedly,
Nearest to from: onto, after, via, sends, infamous, with, regard, harsh,
Nearest to million: billion, mln, 000, bln, 1bn, 5bn, 2bn, 7bn,
Nearest to donald_trump: trump, trumps, marco_rubio, hillary_clinton, franken, barack_obama, alex_jones, kerry,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, mp3, deutschland,
Nearest to may: could, might, will, wont, arent, aim, able, probably,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_apnews.com, domain_businessinsider.com, domain_washingtonpost.com, domain_bloomberg.com, domain_thedailybeast.com,
Nearest to first: 1st, 7th, 2nd, second, sixth, maiden, fourth, highest,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_thehill.com, domain_www.cnn.com, domain_yahoo.com, domain_wsj.com, domain_npr.org,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_wsj.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_medium.com, domain_theguardian.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_thegarance, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_AnnCoulter, subsource_nycjim,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_cnbc.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_otherdomain, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_TimAeppel, subsource_SaraEisen, subsource_memeorandum, subsource_JohnLothian, subsource_AdamPosen, subsource_dandrezner,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_othersource, source_Capital_Spectator,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_politico.com, domain_youtube.com, domain_apnews.com, domain_huffingtonpost.com, domain_bloomberg.com,
Average loss at step 212000: 2.804200
Average loss at step 214000: 2.809507
Average loss at step 216000: 2.795700
Average loss at step 218000: 2.811826
Average loss at step 220000: 2.798881
Nearest to job: payroll, career, workforce, internship, tenure, worker, apprenticeship, profession,
Nearest to go: roar, crawl, come, really, sneak, sail, turn, enter,
Nearest to make: rake, steer, making, convince, give, pave, remake, swallow,
Nearest to who: allegedly, able, dont, unable, young, convict, prosecute, somehow,
Nearest to after: amid, over, briefly, abruptly, unexpectedly, despite, from, before,
Nearest to from: onto, after, attempt, regard, via, contradict, because, seek,
Nearest to million: billion, 000, mln, 1bn, 5bn, billions, bln, 2bn,
Nearest to donald_trump: trump, trumps, marco_rubio, franken, barack_obama, hillary_clinton, everyone, bush,
Nearest to utc: shtml, 2017, htm, pdf, jpg, mp4, html, mp3,
Nearest to may: jul, apr, mar, feb, jan, jun, might, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_businessinsider.com, domain_thedailybeast.com, domain_nationalreview.com, domain_bloomberg.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, fourth, maiden, second, sixth, ninth,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_wsj.com, domain_www.cnn.com, domain_theatlantic.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_theguardian.com, domain_washingtonpost.com, domain_nytimes.com, domain_bloomberg.com, domain_medium.com, domain_reuters.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_benshapiro, subsource_whpresscorps, subsource_nycjim, subsource_vcutrader, subsource_SteveKopack, subsource_maggieNYT,
Nearest to domain_wsj.com: domain_reuters.com, domain_washingtonpost.com, domain_nytimes.com, domain_otherdomain, domain_ft.com, domain_bloomberg.com, domain_forbes.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_wsj.com, domain_cnbc.com, domain_bbc.co.uk, domain_economist.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_TimAeppel, subsource_xeni, subsource_djrothkopf, subsource_maggieNYT,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, owen, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_apnews.com, domain_fortune.com, domain_nymag.com, domain_youtube.com, domain_politico.com,
Average loss at step 222000: 2.812851
Average loss at step 224000: 2.793018
Average loss at step 226000: 2.805579
Average loss at step 228000: 2.794623
Average loss at step 230000: 2.832155
Nearest to job: payroll, internship, career, employment, workforce, tenure, apprenticeship, worker,
Nearest to go: roar, come, crawl, trickle, turn, sneak, get, really,
Nearest to make: steer, rake, give, remake, swallow, enjoy, making, convince,
Nearest to who: allegedly, unable, dont, portman, somehow, able, wrongly, young,
Nearest to after: amid, despite, follow, prompt, over, briefly, castile, since,
Nearest to from: into, onto, after, via, infamous, sends, over, since,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, 2bn,
Nearest to donald_trump: trump, marco_rubio, trumps, hillary_clinton, kerry, barack_obama, conway, christie,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, deutschland, png,
Nearest to may: might, jul, feb, could, will, apr, november, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_businessinsider.com, domain_medium.com, domain_apnews.com, domain_thedailybeast.com, domain_www.cnn.com, domain_washingtonpost.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, second, sixth, worst,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_reuters.com, domain_thehill.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_theguardian.com, domain_businessinsider.com, domain_wsj.com, domain_reuters.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_jonnajarian, subsource_vcutrader, subsource_SteveKopack, subsource_AnnCoulter, subsource_thegarance,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_economist.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_otherdomain, domain_businessinsider.com, domain_bbc.co.uk, domain_economist.com, domain_nytimes.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_TimAeppel, subsource_Frances_Coppola, subsource_memeorandum, subsource_sdonnan, subsource_xeni, subsource_SaraEisen, weighs,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Abnormal_Returns, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_othersource, source_Capital_Spectator,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_politico.com, domain_ft.com, domain_fortune.com, domain_apnews.com,
Average loss at step 232000: 2.791251
Average loss at step 234000: 2.790502
Average loss at step 236000: 2.791830
Average loss at step 238000: 2.788871
Average loss at step 240000: 2.781051
Nearest to job: career, tenure, apprenticeship, payroll, internship, incentive, employed, employment,
Nearest to go: roar, come, really, trickle, crawl, honeymoon, turn, enter,
Nearest to make: rake, remake, give, steer, convince, making, pave, swallow,
Nearest to who: allegedly, portman, unable, prosecute, dont, disabled, able, young,
Nearest to after: amid, despite, wake, briefly, when, over, because, abruptly,
Nearest to from: onto, regard, via, has, sends, contradict, after, into,
Nearest to million: billion, 000, mln, bln, 5bn, 1bn, 2bn, 7bn,
Nearest to donald_trump: trump, trumps, marco_rubio, hillary_clinton, franken, kerry, barack_obama, rex_tillerson,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, deutschland, mp3,
Nearest to may: could, might, will, wont, november, arent, june, jul,
Nearest to domain_youtube.com: domain_otherdomain, domain_businessinsider.com, domain_thedailybeast.com, domain_nytimes.com, domain_medium.com, domain_www.cnn.com, domain_dailymail.co.uk, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, groundbreaking, sixth, second, fourth,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_politico.com, domain_otherdomain, domain_www.cnn.com, domain_thehill.com, domain_vox.com, domain_npr.org, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_nytimes.com, domain_telegraph.co.uk, domain_theguardian.com, domain_forbes.com,
Nearest to subsource_DRUDGE_REPORT: subsource_whpresscorps, subsource_thegarance, subsource_memeorandum, subsource_deray, subsource_instapundit, subsource_jonnajarian, subsource_AnnCoulter, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_forbes.com, domain_ft.com, domain_marketwatch.com, domain_washingtonpost.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_businessinsider.com, domain_bbc.co.uk, domain_economist.com, domain_cnbc.com, domain_wsj.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_TimAeppel, weighs, subsource_memeorandum, subsource_xeni, subsource_mehdirhasan,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_wsj.com, domain_otherdomain, domain_politico.com, domain_bloomberg.com, domain_apnews.com, domain_fortune.com, domain_youtube.com,
Average loss at step 242000: 2.782159
Average loss at step 244000: 2.768504
Average loss at step 246000: 2.768937
Average loss at step 248000: 2.757875
Average loss at step 250000: 2.778126
Nearest to job: payroll, career, workforce, tenure, employment, profession, apprenticeship, internship,
Nearest to go: roar, come, trickle, crawl, enter, really, borrow, honeymoon,
Nearest to make: give, rake, steer, convince, remake, making, pave, swallow,
Nearest to who: allegedly, dont, young, able, somehow, convict, cant, portman,
Nearest to after: amid, over, despite, briefly, from, abruptly, during, before,
Nearest to from: onto, after, into, hub, attempt, via, regard, sends,
Nearest to million: billion, mln, 000, 1bn, 5bn, bln, 7bn, 2bn,
Nearest to donald_trump: trump, trumps, marco_rubio, hillary_clinton, franken, barack_obama, kerry, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, mp3, png,
Nearest to may: could, might, will, november, likely, wont, able, considers,
Nearest to domain_youtube.com: domain_otherdomain, domain_thedailybeast.com, domain_medium.com, domain_nytimes.com, domain_bloomberg.com, domain_wsj.com, domain_businessinsider.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, sixth, maiden, second, fourth, worst,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_yahoo.com, domain_money.cnn.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_bloomberg.com, domain_telegraph.co.uk, domain_wsj.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_maggieNYT, subsource_vcutrader, subsource_deray, subsource_SteveKopack, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_otherdomain, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_politico.com, domain_forbes.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_cnbc.com, domain_economist.com, domain_wsj.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_memeorandum, subsource_TimAeppel, subsource_jessefelder, subsource_sdonnan, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_othersource, source_Capital_Spectator,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_fortune.com, domain_apnews.com, domain_nymag.com, domain_youtube.com, domain_politico.com,
Average loss at step 252000: 2.777596
Average loss at step 254000: 2.768734
Average loss at step 256000: 2.762662
Average loss at step 258000: 2.767156
Average loss at step 260000: 2.759440
Nearest to job: career, internship, tenure, payroll, workforce, apprenticeship, profession, employment,
Nearest to go: roar, come, crawl, sneak, turn, honeymoon, trickle, enter,
Nearest to make: rake, give, convince, steer, enjoy, pave, remake, swallow,
Nearest to who: allegedly, somehow, portman, unable, able, dont, prosecute, young,
Nearest to after: amid, over, despite, follow, briefly, during, castile, prompt,
Nearest to from: onto, about, into, after, via, sends, attempt, during,
Nearest to million: billion, mln, 000, bln, 5bn, 1bn, 7bn, 2bn,
Nearest to donald_trump: trump, trumps, barack_obama, hillary_clinton, kerry, marco_rubio, conway, shameful,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, aspx, png,
Nearest to may: jul, apr, feb, could, might, mar, jan, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_nationalreview.com, domain_washingtonpost.com, domain_www.cnn.com, domain_businessinsider.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, sixth, worst, second,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_politico.com, domain_nytimes.com, domain_www.cnn.com, domain_thehill.com, domain_vox.com, domain_slate.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_washingtonpost.com, domain_youtube.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_DividendMaster, subsource_instapundit, subsource_SteveKopack, subsource_maggieNYT,
Nearest to domain_wsj.com: domain_nytimes.com, domain_otherdomain, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_apnews.com, domain_economist.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bloomberg.com, domain_bbc.co.uk, domain_otherdomain, domain_wsj.com, domain_economist.com, domain_theguardian.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_TimAeppel, subsource_blakehounshell, subsource_mehdirhasan, subsource_maggieNYT, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Incidental_Economist, source_othersource, source_Capital_Spectator,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_bloomberg.com, domain_ft.com, domain_politico.com,
Average loss at step 262000: 2.773011
Average loss at step 264000: 2.800329
Average loss at step 266000: 2.743600
Average loss at step 268000: 2.765050
Average loss at step 270000: 2.757155
Nearest to job: career, apprenticeship, tenure, payroll, internship, employment, gig, workforce,
Nearest to go: roar, really, crawl, come, sneak, turn, honeymoon, enter,
Nearest to make: rake, give, remake, making, convince, steer, depict, pave,
Nearest to who: allegedly, dont, somehow, unable, portman, prosecute, able, young,
Nearest to after: amid, despite, over, briefly, follow, abruptly, when, from,
Nearest to from: about, after, onto, sends, via, infamous, with, regard,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, 2bn,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, barack_obama, kerry, marco_rubio, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, aspx, mp3,
Nearest to may: could, might, will, november, arent, jul, wont, able,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_businessinsider.com, domain_thedailybeast.com, domain_www.cnn.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, second, maiden, fourth, 5th, worst,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_thedailybeast.com, domain_thehill.com, domain_vox.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_wsj.com, domain_washingtonpost.com, domain_medium.com, domain_reuters.com, domain_theguardian.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_DividendMaster, subsource_SteveKopack, subsource_vcutrader, subsource_aussietorres,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_forbes.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_bbc.co.uk, domain_reuters.com, domain_wsj.com, domain_economist.com, domain_cnbc.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, weighs, subsource_edwardnh, subsource_jessefelder, subsource_TimAeppel, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Abnormal_Returns, source_Here_Is_The_City, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_yahoo.com, domain_bloomberg.com,
Average loss at step 272000: 2.744638
Average loss at step 274000: 2.755464
Average loss at step 276000: 2.746091
Average loss at step 278000: 2.737553
Average loss at step 280000: 2.735345
Nearest to job: payroll, employment, career, workforce, tenure, apprenticeship, internship, percent,
Nearest to go: roar, really, come, crawl, enter, trickle, borrow, turn,
Nearest to make: give, rake, steer, remake, convince, pave, enjoy, making,
Nearest to who: allegedly, dont, somehow, cant, unable, young, prosecute, able,
Nearest to after: amid, despite, follow, briefly, over, before, abruptly, castile,
Nearest to from: onto, after, via, into, sends, infamous, attempt, bound,
Nearest to million: billion, mln, 000, bln, 5bn, 1bn, 7bn, 2bn,
Nearest to donald_trump: trump, trumps, marco_rubio, hillary_clinton, kerry, franken, shameful, dubious,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, mp3, share_idea_stocktwits,
Nearest to may: could, might, will, november, probably, considers, likely, formally,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_dailymail.co.uk,
Nearest to first: 1st, 7th, 2nd, groundbreaking, sixth, second, maiden, fourth,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_politico.com, domain_otherdomain, domain_www.cnn.com, domain_thehill.com, domain_npr.org, domain_wsj.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_reuters.com, domain_nytimes.com, domain_ft.com, domain_wsj.com, domain_washingtonpost.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_instapundit, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_reuters.com, domain_otherdomain, domain_nytimes.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_politico.com, domain_apnews.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_cnbc.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_memeorandum, subsource_TimAeppel, why, subsource_dandrezner, subsource_JohnLothian,
Nearest to source_Huffington_Post: source_NYT_Dealbook, source_Business_Insider, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_othersource, source_Capital_Spectator,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_politico.com, domain_apnews.com, domain_nymag.com, domain_youtube.com, domain_fortune.com,
Average loss at step 282000: 2.744910
Average loss at step 284000: 2.745078
Average loss at step 286000: 2.739627
Average loss at step 288000: 2.729780
Average loss at step 290000: 2.739169
Nearest to job: career, payroll, workforce, worker, internship, tenure, employment, apprenticeship,
Nearest to go: roar, come, crawl, really, enter, turn, sneak, trickle,
Nearest to make: rake, steer, pave, give, remake, convince, making, evade,
Nearest to who: allegedly, dont, young, unable, somehow, portman, prosecute, convict,
Nearest to after: amid, despite, over, during, briefly, follow, kabul, wake,
Nearest to from: onto, attempt, after, into, receives, via, sends, hub,
Nearest to million: billion, mln, 000, bln, 5bn, 1bn, 2bn, 7bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, barack_obama, marco_rubio, everyone, republican, shameful,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: jul, apr, feb, jan, mar, jun, june, might,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_ft.com, domain_businessinsider.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, sixth, second, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_thedailybeast.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_washingtonpost.com, domain_nytimes.com, domain_theguardian.com, domain_medium.com, domain_wsj.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_SteveKopack, subsource_instapundit, subsource_DividendMaster, subsource_whpresscorps, subsource_maggieNYT, subsource_nycjim,
Nearest to domain_wsj.com: domain_reuters.com, domain_nytimes.com, domain_otherdomain, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_forbes.com, domain_economist.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_TimAeppel, subsource_SaraEisen, subsource_memeorandum, subsource_xeni, subsource_maggieNYT, weighs,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_bloomberg.com, domain_apnews.com, domain_theatlantic.com,
Average loss at step 292000: 2.735342
Average loss at step 294000: 2.737334
Average loss at step 296000: 2.770035
Average loss at step 298000: 2.722000
Average loss at step 300000: 2.736214
Nearest to job: career, apprenticeship, internship, tenure, payroll, incentive, workforce, employment,
Nearest to go: roar, come, really, crawl, turn, trickle, enter, sneak,
Nearest to make: rake, remake, give, convince, swallow, steer, making, pave,
Nearest to who: dont, allegedly, portman, somehow, able, unable, prosecute, young,
Nearest to after: amid, despite, briefly, over, during, before, abruptly, since,
Nearest to from: via, after, about, into, onto, regard, has, sends,
Nearest to million: billion, mln, 000, 5bn, 7bn, bln, 1bn, 2bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, marco_rubio, imperial, kerry, barack_obama, entirely,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: could, might, will, november, jul, feb, apr, formally,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_bloomberg.com, domain_medium.com, domain_theguardian.com, domain_businessinsider.com, domain_wsj.com, domain_thedailybeast.com,
Nearest to first: 1st, 2nd, 7th, second, maiden, fourth, 5th, worst,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thedailybeast.com, domain_theguardian.com, domain_thehill.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_theguardian.com, domain_bloomberg.com, domain_wsj.com, domain_reuters.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_nycjim, subsource_SteveKopack, subsource_george_chen, subsource_vcutrader,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_forbes.com, domain_economist.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_theguardian.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_memeorandum, subsource_TimAeppel, subsource_SaraEisen, subsource_AdamPosen, subsource_edwardnh, weighs,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_bloomberg.com, domain_apnews.com,
Average loss at step 302000: 2.739846
Average loss at step 304000: 2.718798
Average loss at step 306000: 2.732809
Average loss at step 308000: 2.718662
Average loss at step 310000: 2.712476
Nearest to job: career, payroll, apprenticeship, tenure, employment, workforce, profession, incentive,
Nearest to go: roar, come, crawl, really, enter, borrow, trickle, honeymoon,
Nearest to make: remake, rake, give, steer, pave, convince, reassure, tweak,
Nearest to who: allegedly, dont, unable, somehow, portman, able, cant, prosecute,
Nearest to after: amid, despite, briefly, before, over, from, during, abruptly,
Nearest to from: via, after, onto, sends, with, infamous, bound, into,
Nearest to million: billion, mln, 000, 5bn, bln, 1bn, 2bn, 7bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, marco_rubio, franken, shameful, christie, barack_obama,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, aspx, share_idea_stocktwits,
Nearest to may: could, might, will, november, probably, jul, already, wont,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_bloomberg.com, domain_nytimes.com, domain_thedailybeast.com, domain_apnews.com, domain_businessinsider.com, grover,
Nearest to first: 1st, 7th, 2nd, second, maiden, 5th, sixth, fourth,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_yahoo.com, domain_wsj.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_reuters.com, domain_nytimes.com, domain_theguardian.com, domain_medium.com, domain_wsj.com, domain_washingtonpost.com,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_nycjim, subsource_DowdEdward,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_washingtonpost.com, domain_bloomberg.com, domain_forbes.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_reuters.com, domain_businessinsider.com, domain_marketwatch.com, domain_economist.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_JohnLothian, subsource_SaraEisen, subsource_TimAeppel, subsource_dandrezner, subsource_memeorandum, subsource_sdonnan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_politico.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 312000: 2.714676
Average loss at step 314000: 2.703302
Average loss at step 316000: 2.721689
Average loss at step 318000: 2.707643
Average loss at step 320000: 2.717005
Nearest to job: career, payroll, tenure, worker, workforce, employment, incentive, profession,
Nearest to go: roar, come, crawl, really, enter, honeymoon, borrow, trickle,
Nearest to make: rake, steer, give, convince, pave, remake, swallow, reassure,
Nearest to who: allegedly, portman, young, convict, prosecute, unable, dont, yell,
Nearest to after: amid, despite, briefly, over, abruptly, from, before, unexpectedly,
Nearest to from: onto, after, attempt, via, with, into, about, infamous,
Nearest to million: billion, mln, 000, bln, 5bn, 1bn, billions, 7bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, republican, everyone, dubious, franken, barack_obama,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, aspx, share_idea_stocktwits,
Nearest to may: jul, apr, jan, feb, jun, mar, june, might,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_thedailybeast.com, domain_washingtonpost.com, domain_businessinsider.com, domain_bloomberg.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, sixth, fourth, second, ninth,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_wsj.com, domain_vox.com, domain_thehill.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_medium.com, domain_wsj.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_thegarance, subsource_memeorandum, subsource_SteveKopack, subsource_whpresscorps, subsource_nycjim, subsource_vcutrader, subsource_benshapiro, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_otherdomain, domain_reuters.com, domain_nytimes.com, domain_washingtonpost.com, domain_ft.com, domain_bloomberg.com, domain_apnews.com, domain_economist.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_bbc.co.uk, domain_cnbc.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_xeni, subsource_mehdirhasan, subsource_TimAeppel,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_apnews.com, domain_youtube.com, domain_fortune.com, domain_theatlantic.com, domain_politico.com,
Average loss at step 322000: 2.705323
Average loss at step 324000: 2.715907
Average loss at step 326000: 2.697558
Average loss at step 328000: 2.729638
Average loss at step 330000: 2.719721
Nearest to job: career, apprenticeship, payroll, internship, tenure, workforce, employment, worker,
Nearest to go: roar, come, really, trickle, crawl, turn, sneak, enter,
Nearest to make: rake, remake, give, steer, evade, convince, making, pave,
Nearest to who: allegedly, portman, unable, dont, somehow, young, able, wrongly,
Nearest to after: amid, follow, despite, during, from, abruptly, over, briefly,
Nearest to from: after, onto, into, via, about, sends, infamous, celebrates,
Nearest to million: billion, 000, mln, bln, 5bn, 1bn, 7bn, 2bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, marco_rubio, republican, everyone, shameful, barack_obama,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, aspx, share_idea_stocktwits,
Nearest to may: could, might, november, corbyn, jul, apr, will, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_washingtonpost.com, domain_bloomberg.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, 5th, second, sixth,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_www.cnn.com, domain_theatlantic.com, domain_thehill.com, domain_vox.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_medium.com, domain_telegraph.co.uk,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_thegarance, subsource_DividendMaster, subsource_ylanmui,
Nearest to domain_wsj.com: domain_nytimes.com, domain_reuters.com, domain_otherdomain, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bloomberg.com, domain_bbc.co.uk, domain_wsj.com, domain_theguardian.com, domain_economist.com, domain_businessinsider.com, domain_otherdomain,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_memeorandum, subsource_SaraEisen, subsource_Frances_Coppola, subsource_TimAeppel, subsource_mehdirhasan, subsource_blakehounshell, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_theguardian.com, domain_fortune.com, domain_apnews.com,
Average loss at step 332000: 2.714465
Average loss at step 334000: 2.713625
Average loss at step 336000: 2.703242
Average loss at step 338000: 2.701758
Average loss at step 340000: 2.703396
Nearest to job: career, payroll, tenure, apprenticeship, workforce, employment, incentive, internship,
Nearest to go: roar, come, crawl, really, turn, get, borrow, trickle,
Nearest to make: remake, rake, give, steer, convince, making, pave, depict,
Nearest to who: allegedly, unable, able, young, somehow, portman, dont, prosecute,
Nearest to after: amid, despite, briefly, over, during, from, when, castile,
Nearest to from: onto, via, after, with, infamous, sends, regard, contradict,
Nearest to million: billion, mln, 000, bln, 5bn, 7bn, 1bn, 2bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, everyone, barack_obama, franken, rex_tillerson, marco_rubio,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, aspx, share_idea_stocktwits,
Nearest to may: could, might, will, november, wont, arent, probably, formally,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_businessinsider.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_apnews.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, 5th, second, sixth, maiden, fourth,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_thedailybeast.com, domain_vox.com, domain_nbcnews.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_wsj.com, domain_reuters.com, domain_washingtonpost.com, domain_nytimes.com, domain_medium.com, domain_forbes.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_DowdEdward, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_otherdomain, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_SaraEisen, subsource_Frances_Coppola, subsource_memeorandum, subsource_dandrezner, subsource_TimAeppel, subsource_blakehounshell, subsource_sdonnan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_politico.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 342000: 2.684947
Average loss at step 344000: 2.692293
Average loss at step 346000: 2.679166
Average loss at step 348000: 2.697361
Average loss at step 350000: 2.694357
Nearest to job: career, payroll, workforce, employment, tenure, apprenticeship, incentive, worker,
Nearest to go: roar, come, crawl, really, borrow, sneak, turn, trickle,
Nearest to make: rake, convince, remake, making, give, steer, swallow, transform,
Nearest to who: allegedly, young, somehow, convict, dont, prosecute, wan, able,
Nearest to after: amid, briefly, over, despite, from, abruptly, before, unexpectedly,
Nearest to from: onto, after, attempt, hub, via, with, infamous, sends,
Nearest to million: billion, mln, 000, 5bn, bln, 1bn, 7bn, billions,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, republican, obamas, marco_rubio, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: jul, could, might, apr, november, mar, jan, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, domain_wsj.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, fourth, worst, sixth,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_wsj.com, domain_yahoo.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_maggieNYT, subsource_deray, subsource_benshapiro,
Nearest to domain_wsj.com: domain_otherdomain, domain_reuters.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_politico.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_bbc.co.uk, domain_reuters.com, domain_wsj.com, domain_cnbc.com, domain_economist.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_memeorandum, subsource_TimAeppel, subsource_dandrezner, subsource_jessefelder, subsource_prashantrao,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_apnews.com, domain_fortune.com, domain_youtube.com, domain_nymag.com, domain_politico.com,
Average loss at step 352000: 2.686509
Average loss at step 354000: 2.677975
Average loss at step 356000: 2.689983
Average loss at step 358000: 2.676109
Average loss at step 360000: 2.688398
Nearest to job: career, internship, workforce, tenure, payroll, worker, apprenticeship, incentive,
Nearest to go: roar, come, crawl, turn, sneak, really, get, enter,
Nearest to make: rake, remake, give, steer, convince, making, evade, become,
Nearest to who: allegedly, portman, unable, somehow, young, dont, wan, able,
Nearest to after: amid, despite, follow, briefly, over, during, castile, when,
Nearest to from: onto, about, into, after, via, attempt, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, billions, thousand, 7bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, kerry, shameful, republican, marco_rubio, conway,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: might, could, jul, november, apr, will, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_ft.com, domain_nypost.com, domain_washingtonpost.com,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, fourth, groundbreaking, second,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_nytimes.com, domain_theguardian.com, domain_bloomberg.com, domain_wsj.com, domain_reuters.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_SteveKopack, subsource_aussietorres, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bloomberg.com, domain_otherdomain, domain_wsj.com, domain_bbc.co.uk, domain_economist.com, domain_nytimes.com, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_TimAeppel, subsource_blakehounshell, subsource_GTCost, subsource_memeorandum, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Abnormal_Returns, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_ft.com, domain_apnews.com, domain_theguardian.com, domain_bloomberg.com,
Average loss at step 362000: 2.723917
Average loss at step 364000: 2.676127
Average loss at step 366000: 2.693577
Average loss at step 368000: 2.695001
Average loss at step 370000: 2.665959
Nearest to job: career, tenure, payroll, apprenticeship, internship, intern, incentive, employment,
Nearest to go: roar, come, really, crawl, turn, honeymoon, enter, get,
Nearest to make: remake, rake, give, making, steer, convince, depict, swallow,
Nearest to who: allegedly, portman, somehow, prosecute, unable, young, dont, able,
Nearest to after: amid, despite, over, briefly, from, follow, during, when,
Nearest to from: after, onto, via, about, with, has, regard, infamous,
Nearest to million: billion, mln, 000, bln, 5bn, 7bn, 1bn, 2bn,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, rex_tillerson, kerry, dubious, marco_rubio,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: could, might, will, november, jul, arent, june, apr,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_dailymail.co.uk, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, 5th, fourth, maiden, 8th, second,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_www.cnn.com, domain_otherdomain, domain_politico.com, domain_thehill.com, domain_vox.com, domain_thedailybeast.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_wsj.com, domain_nytimes.com, domain_reuters.com, domain_theguardian.com, domain_forbes.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_instapundit, subsource_DividendMaster, subsource_vcutrader, subsource_deray, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_forbes.com, domain_politico.com, domain_washingtonpost.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_businessinsider.com, domain_wsj.com, domain_economist.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_TimAeppel, subsource_memeorandum, subsource_blakehounshell, subsource_dandrezner, subsource_mehdirhasan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Capital_Spectator, source_Incidental_Economist, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_politico.com, domain_fortune.com, domain_bloomberg.com,
Average loss at step 372000: 2.686072
Average loss at step 374000: 2.672009
Average loss at step 376000: 2.667425
Average loss at step 378000: 2.662411
Average loss at step 380000: 2.668653
Nearest to job: career, workforce, payroll, employment, worker, tenure, incentive, outsource,
Nearest to go: roar, come, really, crawl, anyone, turn, trickle, borrow,
Nearest to make: give, remake, rake, steer, convince, making, swallow, pave,
Nearest to who: allegedly, dont, somehow, young, unable, cant, portman, able,
Nearest to after: amid, despite, over, from, briefly, follow, before, during,
Nearest to from: onto, after, with, via, infamous, sends, into, directly,
Nearest to million: billion, mln, 000, 5bn, bln, 1bn, 7bn, billions,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, marco_rubio, dubious, shameful, obamas,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: could, might, will, november, probably, already, arent, likely,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_theguardian.com, domain_dailymail.co.uk,
Nearest to first: 1st, 7th, 2nd, fourth, sixth, every, maiden, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_npr.org, domain_yahoo.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_wsj.com, domain_theguardian.com, domain_reuters.com, domain_bbc.co.uk,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_DividendMaster, subsource_maggieNYT,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_ft.com, domain_bloomberg.com, domain_yahoo.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_reuters.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_memeorandum, subsource_TimAeppel, subsource_jessefelder, subsource_rossfinley, subsource_sdonnan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_bloomberg.com, domain_fortune.com, domain_politico.com,
Average loss at step 382000: 2.676528
Average loss at step 384000: 2.660702
Average loss at step 386000: 2.675995
Average loss at step 388000: 2.653423
Average loss at step 390000: 2.671362
Nearest to job: career, payroll, internship, workforce, tenure, worker, incentive, apprenticeship,
Nearest to go: come, roar, crawl, turn, sneak, really, enter, trickle,
Nearest to make: rake, remake, steer, give, convince, become, pave, evade,
Nearest to who: allegedly, portman, young, somehow, able, prosecute, unable, dont,
Nearest to after: amid, despite, during, over, follow, briefly, since, before,
Nearest to from: onto, about, after, attempt, into, against, via, analyze,
Nearest to million: billion, mln, 000, 5bn, bln, 1bn, billions, 7bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, everyone, shameful, marco_rubio, obamas, kerry,
Nearest to utc: 2017, shtml, htm, pdf, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: apr, jul, might, jan, feb, could, mar, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_www.cnn.com, domain_businessinsider.com, domain_dailymail.co.uk, domain_nationalreview.com,
Nearest to first: 1st, 2nd, 7th, fourth, maiden, 5th, sixth, second,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_reuters.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_medium.com, domain_wsj.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_deray, subsource_maggieNYT, subsource_SteveKopack, subsource_instapundit, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_nytimes.com, domain_otherdomain, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_economist.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_wsj.com, domain_bbc.co.uk, domain_economist.com, domain_nytimes.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_TimAeppel, subsource_blakehounshell, subsource_xeni, subsource_memeorandum, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_theatlantic.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 392000: 2.654781
Average loss at step 394000: 2.695898
Average loss at step 396000: 2.668079
Average loss at step 398000: 2.669176
Average loss at step 400000: 2.670878
Nearest to job: career, incentive, payroll, apprenticeship, workforce, employment, internship, tenure,
Nearest to go: roar, come, really, crawl, sneak, get, turn, honeymoon,
Nearest to make: rake, remake, making, convince, give, depict, pave, steer,
Nearest to who: allegedly, portman, somehow, dont, young, unable, prosecute, able,
Nearest to after: amid, despite, over, briefly, from, follow, during, when,
Nearest to from: after, about, onto, via, into, sends, regard, infamous,
Nearest to million: billion, mln, 000, 5bn, bln, 7bn, 1bn, 2bn,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, kerry, marco_rubio, rex_tillerson, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: could, might, november, will, jul, corbyn, arent, apr,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_bloomberg.com, domain_thedailybeast.com, domain_www.cnn.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, 5th, groundbreaking, second,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_theguardian.com, domain_thehill.com, domain_vox.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_theguardian.com, domain_wsj.com, domain_washingtonpost.com, domain_reuters.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_deray, subsource_vcutrader, subsource_DividendMaster, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_economist.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_edwardnh, subsource_TimAeppel, subsource_rossfinley, subsource_jessefelder, subsource_mehdirhasan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_bloomberg.com, domain_apnews.com,
Average loss at step 402000: 2.661470
Average loss at step 404000: 2.655727
Average loss at step 406000: 2.658896
Average loss at step 408000: 2.645278
Average loss at step 410000: 2.649718
Nearest to job: career, tenure, payroll, workforce, apprenticeship, employment, worker, intern,
Nearest to go: roar, come, really, crawl, turn, trickle, enter, borrow,
Nearest to make: remake, rake, steer, give, convince, making, pave, reassure,
Nearest to who: allegedly, young, somehow, dont, unable, prosecute, cant, portman,
Nearest to after: amid, despite, briefly, before, over, from, during, follow,
Nearest to from: onto, after, via, infamous, with, sends, into, receives,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, 1bn,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, shameful, marco_rubio, kerry, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: could, might, november, will, jul, apr, arent, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_bloomberg.com, domain_thedailybeast.com, domain_apnews.com, domain_businessinsider.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, fourth, second, 5th, maiden, groundbreaking,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_washingtonpost.com, domain_ft.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_deray, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_DividendMaster, subsource_vcutrader, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_washingtonpost.com, domain_bloomberg.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_memeorandum, subsource_dandrezner, subsource_JohnLothian, subsource_TimAeppel, subsource_sdonnan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_apnews.com, domain_theguardian.com, domain_fortune.com,
Average loss at step 412000: 2.640718
Average loss at step 414000: 2.656596
Average loss at step 416000: 2.654352
Average loss at step 418000: 2.646180
Average loss at step 420000: 2.638513
Nearest to job: career, payroll, workforce, tenure, worker, internship, incentive, apprenticeship,
Nearest to go: roar, come, crawl, really, turn, honeymoon, sneak, enter,
Nearest to make: rake, give, steer, remake, convince, pave, ramp, reassure,
Nearest to who: allegedly, young, dont, portman, somehow, unable, prosecute, yell,
Nearest to after: amid, despite, from, over, briefly, before, when, during,
Nearest to from: onto, after, attempt, via, about, into, with, against,
Nearest to million: billion, 000, mln, 5bn, bln, billions, 7bn, 1bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, everyone, republican, obamas, shameful, kerry,
Nearest to utc: 2017, shtml, pdf, htm, jpg, mp4, share_idea_stocktwits, aspx,
Nearest to may: jul, apr, feb, mar, jan, jun, might, could,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_dailymail.co.uk, domain_nationalreview.com, domain_bloomberg.com, domain_washingtonpost.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, second, 5th, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_theguardian.com, domain_thehill.com, domain_wsj.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_instapundit, subsource_deray, subsource_whpresscorps, subsource_SteveKopack, subsource_DividendMaster, subsource_maggieNYT,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com, domain_forbes.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bloomberg.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_cnbc.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_maggieNYT, subsource_SaraEisen, subsource_TimAeppel, subsource_xeni, subsource_blakehounshell,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_politico.com, domain_fortune.com, domain_bloomberg.com,
Average loss at step 422000: 2.646894
Average loss at step 424000: 2.632294
Average loss at step 426000: 2.650923
Average loss at step 428000: 2.684859
Average loss at step 430000: 2.641447
Nearest to job: career, payroll, incentive, apprenticeship, employment, workforce, worker, internship,
Nearest to go: roar, come, really, crawl, turn, honeymoon, get, sneak,
Nearest to make: remake, rake, give, convince, steer, making, swallow, pave,
Nearest to who: allegedly, portman, young, somehow, dont, unable, able, prosecute,
Nearest to after: amid, despite, over, briefly, follow, abruptly, from, during,
Nearest to from: onto, via, after, into, about, with, infamous, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, shameful, barack_obama, marco_rubio, obamas,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, corbyn, jul, apr, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_www.cnn.com, domain_bloomberg.com, grover,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, fourth, groundbreaking, second,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_theguardian.com, domain_thehill.com, domain_thedailybeast.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_The_Real_Fly, subsource_deray, subsource_SteveKopack, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_otherdomain, domain_economist.com, domain_theguardian.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_edwardnh, subsource_SaraEisen, subsource_TimAeppel, subsource_blakehounshell, subsource_maggieNYT,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_bloomberg.com, domain_theguardian.com,
Average loss at step 432000: 2.660925
Average loss at step 434000: 2.645721
Average loss at step 436000: 2.637041
Average loss at step 438000: 2.645373
Average loss at step 440000: 2.635685
Nearest to job: career, payroll, tenure, employment, incentive, apprenticeship, workforce, intern,
Nearest to go: roar, come, crawl, really, honeymoon, turn, get, trickle,
Nearest to make: remake, rake, give, steer, convince, become, pave, reassure,
Nearest to who: allegedly, young, dont, somehow, portman, unable, wan, able,
Nearest to after: amid, despite, briefly, over, from, when, during, abruptly,
Nearest to from: after, via, onto, with, infamous, sends, into, about,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, hillary_clinton, everyone, kerry, shameful, dubious, obamas,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, already, corbyn, probably,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_thedailybeast.com, domain_medium.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, grover,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, second, 8th, groundbreaking,
Nearest to domain_washingtonpost.com: domain_nytimes.com, domain_otherdomain, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_yahoo.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_wsj.com, domain_reuters.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_sdonnan, subsource_dandrezner, subsource_mehdirhasan, subsource_blakehounshell,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_washingtonpost.com, domain_otherdomain, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 442000: 2.630092
Average loss at step 444000: 2.623097
Average loss at step 446000: 2.637626
Average loss at step 448000: 2.636755
Average loss at step 450000: 2.629735
Nearest to job: career, payroll, workforce, employment, incentive, tenure, apprenticeship, worker,
Nearest to go: roar, come, crawl, really, turn, honeymoon, sneak, get,
Nearest to make: rake, give, convince, remake, steer, pave, making, transform,
Nearest to who: allegedly, young, somehow, portman, dont, unable, able, wan,
Nearest to after: amid, over, from, despite, abruptly, briefly, before, during,
Nearest to from: after, onto, via, with, attempt, infamous, into, about,
Nearest to million: billion, 000, mln, 5bn, bln, 1bn, billions, 7bn,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, obamas, dubious, shameful, rex_tillerson,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, apr, mar, might, jan, could, jun, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_www.cnn.com, domain_washingtonexaminer.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, second, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_wsj.com, domain_thehill.com, domain_yahoo.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_SteveKopack, subsource_maggieNYT, subsource_vcutrader, subsource_whpresscorps, subsource_nycjim, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_ft.com, domain_bloomberg.com, domain_apnews.com, domain_forbes.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_businessinsider.com, domain_economist.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_maggieNYT, subsource_blakehounshell, subsource_TimAeppel, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_theatlantic.com,
Average loss at step 452000: 2.623280
Average loss at step 454000: 2.626291
Average loss at step 456000: 2.623932
Average loss at step 458000: 2.624294
Average loss at step 460000: 2.665273
Nearest to job: career, payroll, employment, tenure, internship, worker, apprenticeship, incentive,
Nearest to go: roar, come, crawl, really, turn, get, honeymoon, sneak,
Nearest to make: remake, rake, steer, give, convince, evade, become, swallow,
Nearest to who: allegedly, dont, portman, unable, young, somehow, wan, accuse,
Nearest to after: amid, despite, follow, over, during, from, when, briefly,
Nearest to from: onto, after, into, via, sends, about, against, with,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, hillary_clinton, everyone, shameful, kerry, rex_tillerson, trevor,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: might, could, jul, november, will, apr, mar, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, domain_ft.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, second, 8th, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_vox.com, domain_wsj.com, domain_theguardian.com, domain_thehill.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_thegarance, subsource_deray, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_wsj.com, domain_bbc.co.uk, domain_economist.com, domain_theguardian.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_sdonnan, subsource_SaraEisen, subsource_TimAeppel, subsource_blakehounshell, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com, domain_apnews.com,
Average loss at step 462000: 2.631927
Average loss at step 464000: 2.644507
Average loss at step 466000: 2.643092
Average loss at step 468000: 2.621721
Average loss at step 470000: 2.631907
Nearest to job: career, tenure, intern, payroll, apprenticeship, internship, employment, incentive,
Nearest to go: roar, really, come, honeymoon, crawl, get, turn, margaret,
Nearest to make: rake, remake, give, steer, depict, convince, making, become,
Nearest to who: allegedly, portman, dont, unable, somehow, able, prosecute, young,
Nearest to after: amid, despite, over, briefly, from, when, during, follow,
Nearest to from: after, onto, via, with, into, about, infamous, sends,
Nearest to million: billion, mln, 000, bln, 5bn, 7bn, billions, 1bn,
Nearest to donald_trump: trump, trumps, everyone, rex_tillerson, hillary_clinton, shameful, kerry, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, arent, june, corbyn,
Nearest to domain_youtube.com: domain_otherdomain, domain_businessinsider.com, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_www.cnn.com, grover,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, second, groundbreaking, fourth,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_vox.com, domain_thedailybeast.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_wsj.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_theguardian.com, domain_forbes.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_instapundit, subsource_deray, subsource_vcutrader, subsource_SteveKopack, subsource_maggieNYT,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_businessinsider.com, domain_economist.com, domain_wsj.com, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, deepens, subsource_maggieNYT, weighs,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Capital_Spectator, source_Incidental_Economist, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 472000: 2.617551
Average loss at step 474000: 2.617452
Average loss at step 476000: 2.613388
Average loss at step 478000: 2.608942
Average loss at step 480000: 2.623423
Nearest to job: career, payroll, workforce, employment, intern, incentive, tenure, outsource,
Nearest to go: roar, come, really, crawl, honeymoon, get, borrow, turn,
Nearest to make: give, remake, convince, rake, steer, depict, transform, become,
Nearest to who: allegedly, young, somehow, dont, portman, unable, able, cant,
Nearest to after: amid, from, despite, over, briefly, follow, before, during,
Nearest to from: onto, after, via, with, into, attempt, against, hub,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, everyone, republican, obamas, dubious, hillary_clinton, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, arent, already, likely,
Nearest to domain_youtube.com: domain_otherdomain, domain_medium.com, domain_nytimes.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, grover, domain_theguardian.com,
Nearest to first: 1st, 2nd, 7th, second, maiden, sixth, worst, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_wsj.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_forbes.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_vcutrader, subsource_maggieNYT, subsource_SteveKopack, subsource_deray, subsource_valuewalk,
Nearest to domain_wsj.com: domain_otherdomain, domain_reuters.com, domain_nytimes.com, domain_washingtonpost.com, domain_ft.com, domain_bloomberg.com, domain_yahoo.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_cnbc.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_memeorandum, subsource_rossfinley, subsource_alisterbull1, subsource_edwardnh, why,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_nymag.com,
Average loss at step 482000: 2.606725
Average loss at step 484000: 2.626320
Average loss at step 486000: 2.608218
Average loss at step 488000: 2.614199
Average loss at step 490000: 2.598403
Nearest to job: career, workforce, payroll, worker, employment, tenure, internship, incentive,
Nearest to go: roar, come, crawl, really, turn, sneak, honeymoon, brawl,
Nearest to make: rake, give, convince, remake, steer, become, depict, evade,
Nearest to who: allegedly, portman, somehow, young, unable, dont, able, prosecute,
Nearest to after: amid, despite, over, during, from, briefly, follow, when,
Nearest to from: after, onto, via, about, into, attempt, with, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, 1bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, everyone, rex_tillerson, obamas, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, could, might, apr, november, mar, jan, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_washingtonpost.com, domain_bloomberg.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, fourth, second, every,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_theguardian.com, domain_nytimes.com, domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_vcutrader, subsource_thegarance, subsource_whpresscorps, subsource_SteveKopack, subsource_instapundit, subsource_maggieNYT, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_nytimes.com, domain_otherdomain, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_forbes.com, domain_economist.com,
Nearest to domain_ft.com: domain_reuters.com, domain_bloomberg.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_theguardian.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_SaraEisen, subsource_memeorandum, subsource_TimAeppel, subsource_maggieNYT, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_politico.com, domain_apnews.com, domain_theatlantic.com,
Average loss at step 492000: 2.637371
Average loss at step 494000: 2.625830
Average loss at step 496000: 2.625912
Average loss at step 498000: 2.625891
Average loss at step 500000: 2.615072
Nearest to job: career, tenure, payroll, intern, incentive, employment, workforce, apprenticeship,
Nearest to go: roar, come, really, crawl, honeymoon, get, turn, sneak,
Nearest to make: rake, give, remake, depict, convince, steer, become, making,
Nearest to who: allegedly, portman, somehow, unable, young, dont, prosecute, able,
Nearest to after: amid, despite, over, from, when, briefly, follow, during,
Nearest to from: after, onto, about, via, into, infamous, with, attempt,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, rex_tillerson, everyone, hillary_clinton, obamas, shameful, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, arent, apr, corbyn,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_dailymail.co.uk, grover,
Nearest to first: 1st, 2nd, 7th, 5th, maiden, second, fourth, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_thedailybeast.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_wsj.com, domain_theguardian.com, domain_reuters.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_DividendMaster, subsource_SteveKopack, subsource_aussietorres,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_theguardian.com, domain_businessinsider.com, domain_wsj.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_memeorandum, subsource_SaraEisen, subsource_jessefelder, subsource_blakehounshell, subsource_rossfinley, subsource_maggieNYT,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_politico.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 502000: 2.610399
Average loss at step 504000: 2.609968
Average loss at step 506000: 2.594368
Average loss at step 508000: 2.604072
Average loss at step 510000: 2.587614
Nearest to job: career, payroll, workforce, employment, tenure, incentive, intern, worker,
Nearest to go: really, roar, come, crawl, borrow, get, sneak, honeymoon,
Nearest to make: give, convince, remake, rake, steer, pave, become, depict,
Nearest to who: allegedly, dont, somehow, young, unable, portman, accuse, prosecute,
Nearest to after: amid, over, despite, from, follow, briefly, during, before,
Nearest to from: after, onto, via, with, into, infamous, against, sends,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, obamas, rex_tillerson, shameful, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, arent, formally, already,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, every, second, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_politico.com, domain_www.cnn.com, domain_thehill.com, domain_yahoo.com, domain_theguardian.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_wsj.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_vcutrader, subsource_instapundit, subsource_SteveKopack, subsource_DividendMaster, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_politico.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_pierre, subsource_xeni, subsource_dandrezner,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_apnews.com, domain_fortune.com, domain_nymag.com,
Average loss at step 512000: 2.611253
Average loss at step 514000: 2.609089
Average loss at step 516000: 2.599101
Average loss at step 518000: 2.590197
Average loss at step 520000: 2.601430
Nearest to job: career, payroll, workforce, tenure, incentive, employment, worker, intern,
Nearest to go: roar, come, crawl, really, turn, sneak, get, honeymoon,
Nearest to make: rake, convince, give, remake, steer, become, pave, evade,
Nearest to who: allegedly, young, dont, portman, somehow, unable, prosecute, desperately,
Nearest to after: amid, despite, over, from, follow, during, briefly, with,
Nearest to from: after, onto, attempt, via, with, into, against, about,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, shameful, rex_tillerson, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, apr, could, might, mar, feb, jan, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_dailymail.co.uk, domain_nationalreview.com, domain_bloomberg.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, maiden, fourth, second, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_yahoo.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_washingtonpost.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_SteveKopack, subsource_whpresscorps, subsource_maggieNYT, subsource_vcutrader, subsource_margbrennan, subsource_instapundit,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_TimAeppel, subsource_maggieNYT, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_theatlantic.com, domain_politico.com,
Average loss at step 522000: 2.587718
Average loss at step 524000: 2.600749
Average loss at step 526000: 2.640502
Average loss at step 528000: 2.598290
Average loss at step 530000: 2.614857
Nearest to job: career, payroll, employment, incentive, workforce, apprenticeship, internship, employed,
Nearest to go: roar, come, really, crawl, honeymoon, get, sneak, turn,
Nearest to make: remake, give, convince, rake, depict, making, steer, swallow,
Nearest to who: allegedly, portman, somehow, unable, dont, young, desperately, prosecute,
Nearest to after: amid, despite, over, from, follow, during, briefly, since,
Nearest to from: after, via, into, onto, about, with, against, sends,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, rex_tillerson, hillary_clinton, obamas, everyone, kerry, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, apr, corbyn, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_businessinsider.com, domain_thedailybeast.com, grover, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, second, fourth, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_theguardian.com, domain_thehill.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_margbrennan, subsource_DividendMaster, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_rossfinley, subsource_edwardnh, subsource_TimAeppel,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_bloomberg.com, domain_apnews.com,
Average loss at step 532000: 2.611949
Average loss at step 534000: 2.587972
Average loss at step 536000: 2.604253
Average loss at step 538000: 2.590987
Average loss at step 540000: 2.583577
Nearest to job: career, payroll, employment, incentive, workforce, tenure, apprenticeship, intern,
Nearest to go: come, roar, really, crawl, get, honeymoon, turn, sneak,
Nearest to make: remake, give, rake, convince, steer, depict, become, pave,
Nearest to who: allegedly, somehow, dont, portman, unable, cant, young, able,
Nearest to after: amid, despite, from, over, briefly, during, when, follow,
Nearest to from: after, via, onto, with, sends, infamous, about, into,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, hillary_clinton, everyone, obamas, rex_tillerson, shameful, kerry,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, arent, jan, corbyn,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, grover, domain_apnews.com, domain_businessinsider.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, 5th, 8th, every,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_wsj.com, domain_yahoo.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_wsj.com, domain_reuters.com, domain_theguardian.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_aussietorres, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_wsj.com, domain_bbc.co.uk, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_blakehounshell, subsource_memeorandum, subsource_rossfinley, subsource_dandrezner, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 542000: 2.583151
Average loss at step 544000: 2.590448
Average loss at step 546000: 2.597538
Average loss at step 548000: 2.582347
Average loss at step 550000: 2.593091
Nearest to job: career, payroll, tenure, worker, intern, incentive, workforce, employment,
Nearest to go: come, roar, crawl, really, get, honeymoon, turn, borrow,
Nearest to make: rake, convince, give, remake, steer, become, making, pave,
Nearest to who: allegedly, young, portman, dont, somehow, unable, able, prosecute,
Nearest to after: amid, despite, from, over, briefly, follow, when, before,
Nearest to from: after, onto, via, into, attempt, with, about, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, 1bn,
Nearest to donald_trump: trump, trumps, hillary_clinton, obamas, everyone, rex_tillerson, dubious, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, apr, mar, jan, feb, might, could, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, 5th, 8th, every,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_wsj.com, domain_theguardian.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_theguardian.com, domain_bloomberg.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_SteveKopack, subsource_vcutrader, subsource_instapundit, subsource_nycjim, subsource_maggieNYT,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com, domain_cnbc.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_xeni, subsource_maggieNYT, subsource_mehdirhasan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_theatlantic.com,
Average loss at step 552000: 2.574249
Average loss at step 554000: 2.590324
Average loss at step 556000: 2.572890
Average loss at step 558000: 2.619246
Average loss at step 560000: 2.594699
Nearest to job: career, employment, payroll, workforce, worker, incentive, tenure, apprenticeship,
Nearest to go: come, roar, crawl, really, get, sneak, honeymoon, turn,
Nearest to make: rake, remake, convince, give, steer, become, making, swallow,
Nearest to who: allegedly, portman, young, unable, somehow, able, wrongly, accuse,
Nearest to after: amid, despite, from, follow, since, over, briefly, during,
Nearest to from: after, onto, via, into, about, infamous, sends, with,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, everyone, rex_tillerson, hillary_clinton, shameful, obamas, kerry,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, jul, apr, will, feb, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_washingtonpost.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, 5th, every, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_theguardian.com, domain_thehill.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_SteveKopack, subsource_The_Real_Fly, subsource_thegarance, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_theguardian.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_rossfinley, subsource_mehdirhasan, subsource_TimAeppel,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 562000: 2.602000
Average loss at step 564000: 2.599402
Average loss at step 566000: 2.587932
Average loss at step 568000: 2.584504
Average loss at step 570000: 2.581906
Nearest to job: career, tenure, payroll, employment, intern, incentive, workforce, apprenticeship,
Nearest to go: roar, come, really, crawl, get, honeymoon, turn, margaret,
Nearest to make: remake, rake, give, convince, steer, become, making, depict,
Nearest to who: allegedly, portman, young, somehow, dont, desperately, unable, prosecute,
Nearest to after: amid, despite, over, from, briefly, when, during, follow,
Nearest to from: after, via, onto, with, into, about, infamous, has,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, rex_tillerson, hillary_clinton, obamas, shameful, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, apr, arent, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, domain_apnews.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, maiden, second, 5th, every, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_theguardian.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_WallStManeet, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_forbes.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_dandrezner, subsource_pierre, subsource_mehdirhasan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 572000: 2.569083
Average loss at step 574000: 2.573419
Average loss at step 576000: 2.565935
Average loss at step 578000: 2.583704
Average loss at step 580000: 2.580707
Nearest to job: career, payroll, employment, workforce, worker, tenure, incentive, intern,
Nearest to go: come, roar, really, crawl, get, turn, margaret, honeymoon,
Nearest to make: rake, give, convince, remake, steer, become, making, swallow,
Nearest to who: allegedly, young, dont, somehow, portman, cant, able, desperately,
Nearest to after: amid, from, despite, over, briefly, follow, before, abruptly,
Nearest to from: after, onto, via, with, into, against, infamous, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, dubious, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, could, might, apr, november, mar, jan, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, grover,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, fourth, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_yahoo.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_instapundit, subsource_nycjim,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_memeorandum, subsource_blakehounshell, subsource_rossfinley, subsource_dandrezner, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 582000: 2.579781
Average loss at step 584000: 2.566017
Average loss at step 586000: 2.576688
Average loss at step 588000: 2.566132
Average loss at step 590000: 2.579438
Nearest to job: career, worker, tenure, employment, payroll, intern, incentive, workforce,
Nearest to go: come, roar, really, crawl, get, turn, sneak, honeymoon,
Nearest to make: give, remake, rake, convince, become, steer, swallow, evade,
Nearest to who: allegedly, portman, young, dont, somehow, unable, able, desperately,
Nearest to after: amid, despite, follow, from, over, during, briefly, when,
Nearest to from: after, onto, into, via, with, about, against, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, 1bn,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, republican, rex_tillerson, shameful, obamas,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, jul, november, apr, mar, jan, will,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, grover, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, 5th, second, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_nytimes.com, domain_theguardian.com, domain_bloomberg.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_SteveKopack, subsource_WallStManeet, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_economist.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_edwardnh, subsource_maggieNYT, subsource_mehdirhasan,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 592000: 2.611485
Average loss at step 594000: 2.579753
Average loss at step 596000: 2.598360
Average loss at step 598000: 2.578969
Average loss at step 600000: 2.567674
Nearest to job: career, tenure, employment, payroll, worker, intern, incentive, workforce,
Nearest to go: come, roar, really, crawl, honeymoon, get, turn, sneak,
Nearest to make: remake, give, rake, convince, become, making, depict, steer,
Nearest to who: allegedly, young, portman, dont, somehow, unable, cant, able,
Nearest to after: amid, despite, over, from, when, briefly, during, follow,
Nearest to from: after, via, with, about, onto, into, sends, against,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, billions, 1bn,
Nearest to donald_trump: trump, trumps, everyone, rex_tillerson, hillary_clinton, obamas, shameful, entirely,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, aspx, mp4,
Nearest to may: could, might, will, november, jul, apr, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_www.cnn.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, second, 8th, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_DividendMaster, subsource_WallStManeet, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_memeorandum, subsource_SaraEisen, subsource_mehdirhasan, subsource_blakehounshell, subsource_pierre, subsource_edwardnh,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_politico.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 602000: 2.579871
Average loss at step 604000: 2.564754
Average loss at step 606000: 2.565221
Average loss at step 608000: 2.556990
Average loss at step 610000: 2.571287
Nearest to job: career, payroll, employment, worker, workforce, tenure, renewables, intern,
Nearest to go: roar, really, come, crawl, borrow, turn, honeymoon, sneak,
Nearest to make: give, remake, convince, rake, steer, become, making, transform,
Nearest to who: allegedly, dont, young, cant, somehow, portman, able, unable,
Nearest to after: amid, despite, over, from, before, when, follow, during,
Nearest to from: after, onto, via, with, into, about, against, sends,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, 1bn, billions,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, obamas, republican, rex_tillerson, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, apr, arent, already,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, groundbreaking, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_yahoo.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_wsj.com, domain_washingtonpost.com, domain_reuters.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_DividendMaster, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_washingtonpost.com, domain_bloomberg.com, domain_yahoo.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_cnbc.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_memeorandum, subsource_rossfinley, subsource_jessefelder, subsource_blakehounshell, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 612000: 2.569725
Average loss at step 614000: 2.565259
Average loss at step 616000: 2.555442
Average loss at step 618000: 2.563063
Average loss at step 620000: 2.557633
Nearest to job: career, worker, payroll, employment, workforce, incentive, intern, tenure,
Nearest to go: come, roar, really, crawl, turn, sneak, get, honeymoon,
Nearest to make: convince, give, rake, remake, steer, become, evade, depict,
Nearest to who: allegedly, portman, young, somehow, dont, unable, able, accuse,
Nearest to after: amid, despite, over, follow, during, from, briefly, when,
Nearest to from: after, onto, via, about, into, with, against, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, obamas, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, jul, apr, november, mar, jan, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_ft.com, domain_businessinsider.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, 5th, every, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_wsj.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_washingtonpost.com, domain_theguardian.com, domain_bloomberg.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_SteveKopack, subsource_maggieNYT, subsource_deray, subsource_instapundit, subsource_vcutrader,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_ft.com, domain_bloomberg.com, domain_economist.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_pierre, subsource_mehdirhasan, subsource_maggieNYT,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 622000: 2.559265
Average loss at step 624000: 2.602596
Average loss at step 626000: 2.563020
Average loss at step 628000: 2.581325
Average loss at step 630000: 2.579182
Nearest to job: career, payroll, employment, worker, incentive, workforce, tenure, apprenticeship,
Nearest to go: come, roar, really, get, crawl, sneak, honeymoon, turn,
Nearest to make: remake, rake, convince, give, depict, making, become, steer,
Nearest to who: allegedly, portman, dont, young, somehow, unable, able, cant,
Nearest to after: amid, despite, over, from, briefly, when, follow, during,
Nearest to from: after, via, about, onto, into, with, infamous, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, rex_tillerson, obamas, hillary_clinton, republican, shameful,
Nearest to utc: 2017, shtml, pdf, jpg, htm, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, corbyn, apr, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_bloomberg.com, domain_thedailybeast.com, domain_www.cnn.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, 5th, second, 8th, every,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_theguardian.com, domain_thehill.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_The_Real_Fly, subsource_DividendMaster, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_SaraEisen, subsource_blakehounshell, subsource_memeorandum, subsource_rossfinley, subsource_mehdirhasan, subsource_maggieNYT,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com, domain_apnews.com,
Average loss at step 632000: 2.557896
Average loss at step 634000: 2.564868
Average loss at step 636000: 2.556174
Average loss at step 638000: 2.555697
Average loss at step 640000: 2.548846
Nearest to job: career, payroll, worker, employment, workforce, tenure, intern, incentive,
Nearest to go: come, really, roar, crawl, get, honeymoon, borrow, turn,
Nearest to make: remake, give, convince, rake, steer, making, become, depict,
Nearest to who: allegedly, dont, young, somehow, portman, cant, unable, able,
Nearest to after: amid, despite, over, from, follow, briefly, during, before,
Nearest to from: after, via, onto, with, infamous, into, sends, about,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, republican, hillary_clinton, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, apr, corbyn, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, grover,
Nearest to first: 1st, 2nd, 7th, maiden, every, 5th, second, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_yahoo.com, domain_vox.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_reuters.com, domain_theguardian.com, domain_wsj.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_DividendMaster, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_blakehounshell, subsource_memeorandum, subsource_pierre, subsource_rossfinley, subsource_dandrezner,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com,
Average loss at step 642000: 2.549239
Average loss at step 644000: 2.563802
Average loss at step 646000: 2.547884
Average loss at step 648000: 2.564500
Average loss at step 650000: 2.541582
Nearest to job: career, payroll, workforce, worker, employment, intern, tenure, incentive,
Nearest to go: come, really, roar, crawl, margaret, get, turn, honeymoon,
Nearest to make: give, convince, rake, remake, become, steer, making, depict,
Nearest to who: allegedly, young, somehow, portman, dont, unable, formally, prosecute,
Nearest to after: amid, despite, over, from, during, follow, when, briefly,
Nearest to from: after, onto, via, about, attempt, with, into, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, apr, might, mar, jun, could, jan, feb,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_dailymail.co.uk, domain_businessinsider.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_yahoo.com, domain_wsj.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_washingtonpost.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_instapundit, subsource_deray, subsource_margbrennan,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_politico.com, domain_apnews.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_xeni, subsource_pierre, subsource_maggieNYT,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_theatlantic.com,
Average loss at step 652000: 2.554584
Average loss at step 654000: 2.540766
Average loss at step 656000: 2.576229
Average loss at step 658000: 2.569098
Average loss at step 660000: 2.574305
Nearest to job: career, payroll, workforce, worker, employment, tenure, intern, incentive,
Nearest to go: come, really, roar, crawl, get, turn, honeymoon, sneak,
Nearest to make: convince, give, remake, rake, become, steer, making, swallow,
Nearest to who: allegedly, portman, somehow, young, unable, accuse, dont, formally,
Nearest to after: amid, despite, over, during, from, follow, briefly, abruptly,
Nearest to from: after, via, onto, into, about, infamous, with, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, obamas, everyone, rex_tillerson, republican, hillary_clinton, shameful,
Nearest to utc: 2017, shtml, pdf, jpg, htm, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, corbyn, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, grover, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, 5th, second, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_washingtonpost.com, domain_bloomberg.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_thegarance, subsource_margbrennan,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_politico.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_theguardian.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_rossfinley, subsource_pierre, subsource_maggieNYT,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 662000: 2.569160
Average loss at step 664000: 2.558729
Average loss at step 666000: 2.555354
Average loss at step 668000: 2.554735
Average loss at step 670000: 2.536373
Nearest to job: career, payroll, employment, tenure, workforce, intern, incentive, worker,
Nearest to go: come, really, roar, crawl, get, margaret, fall, honeymoon,
Nearest to make: remake, give, convince, rake, steer, become, depict, making,
Nearest to who: allegedly, young, somehow, portman, dont, unable, able, formally,
Nearest to after: amid, despite, over, from, during, when, briefly, follow,
Nearest to from: after, via, onto, with, into, about, infamous, sends,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, shameful, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, will, november, jul, apr, arent, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, grover, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, 5th, 8th, every,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_yahoo.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_wsj.com, domain_reuters.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_instapundit, subsource_deray, subsource_The_Real_Fly, subsource_Bencjacobs,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_politico.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 672000: 2.547798
Average loss at step 674000: 2.533266
Average loss at step 676000: 2.554710
Average loss at step 678000: 2.550882
Average loss at step 680000: 2.545891
Nearest to job: career, payroll, workforce, employment, worker, tenure, incentive, intern,
Nearest to go: come, roar, really, crawl, margaret, get, turn, borrow,
Nearest to make: rake, convince, give, remake, become, steer, making, swallow,
Nearest to who: allegedly, young, somehow, portman, dont, wan, formally, able,
Nearest to after: amid, despite, over, from, during, follow, before, briefly,
Nearest to from: after, onto, via, with, about, into, infamous, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, shameful, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, apr, might, could, november, jun, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, grover, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_yahoo.com, domain_theguardian.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_theguardian.com, domain_bloomberg.com, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_vcutrader, subsource_whpresscorps, subsource_thegarance, subsource_instapundit, subsource_SteveKopack, subsource_deray, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_marketwatch.com, domain_economist.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_maggieNYT, subsource_pierre, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_politico.com, domain_fortune.com, domain_theatlantic.com,
Average loss at step 682000: 2.536196
Average loss at step 684000: 2.546082
Average loss at step 686000: 2.532411
Average loss at step 688000: 2.547434
Average loss at step 690000: 2.588790
Nearest to job: career, payroll, employment, workforce, worker, incentive, tenure, intern,
Nearest to go: come, roar, really, crawl, get, sneak, margaret, turn,
Nearest to make: remake, give, rake, convince, steer, become, making, swallow,
Nearest to who: allegedly, young, portman, somehow, accuse, dont, wan, unable,
Nearest to after: amid, despite, over, from, during, follow, before, when,
Nearest to from: after, onto, into, via, about, with, against, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, obamas, republican, shameful, rex_tillerson,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: might, could, november, jul, apr, will, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_theguardian.com, domain_thehill.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_washingtonpost.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_The_Real_Fly, subsource_thegarance,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_rossfinley, subsource_xeni, subsource_TimAeppel,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 692000: 2.549663
Average loss at step 694000: 2.566678
Average loss at step 696000: 2.556380
Average loss at step 698000: 2.541609
Average loss at step 700000: 2.545268
Nearest to job: career, payroll, employment, tenure, intern, workforce, incentive, worker,
Nearest to go: come, roar, really, crawl, get, honeymoon, margaret, turn,
Nearest to make: remake, rake, give, convince, become, depict, making, steer,
Nearest to who: allegedly, young, portman, somehow, desperately, dont, unable, cant,
Nearest to after: amid, despite, over, from, when, during, briefly, before,
Nearest to from: after, via, onto, into, with, about, infamous, against,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, obamas, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, corbyn, mull,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_thedailybeast.com, domain_medium.com, domain_businessinsider.com, domain_bloomberg.com, grover, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, 5th, every, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_wsj.com, domain_washingtonpost.com, domain_reuters.com, domain_theguardian.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_WallStManeet, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_economist.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_roburban, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_politico.com, domain_apnews.com, domain_bloomberg.com,
Average loss at step 702000: 2.536314
Average loss at step 704000: 2.536146
Average loss at step 706000: 2.530986
Average loss at step 708000: 2.535958
Average loss at step 710000: 2.544014
Nearest to job: career, payroll, workforce, employment, tenure, incentive, worker, intern,
Nearest to go: come, really, roar, crawl, get, margaret, borrow, fall,
Nearest to make: give, convince, rake, remake, become, steer, making, depict,
Nearest to who: allegedly, young, somehow, dont, portman, accuse, able, cant,
Nearest to after: amid, from, despite, over, briefly, during, follow, before,
Nearest to from: after, onto, with, via, into, against, about, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, republican, rex_tillerson, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, april,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, second, every, maiden, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_yahoo.com, domain_vox.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_vcutrader, subsource_instapundit, subsource_thegarance, subsource_deray, subsource_SteveKopack, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_ft.com, domain_bloomberg.com, domain_washingtonpost.com, domain_yahoo.com, domain_politico.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_memeorandum, subsource_blakehounshell, subsource_rossfinley, subsource_pierre, subsource_jessefelder,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_theatlantic.com,
Average loss at step 712000: 2.532491
Average loss at step 714000: 2.534696
Average loss at step 716000: 2.522027
Average loss at step 718000: 2.540141
Average loss at step 720000: 2.523477
Nearest to job: career, payroll, employment, worker, workforce, tenure, incentive, intern,
Nearest to go: come, roar, really, crawl, get, margaret, sneak, turn,
Nearest to make: give, rake, convince, remake, become, steer, depict, evade,
Nearest to who: allegedly, young, portman, somehow, dont, accuse, desperately, unable,
Nearest to after: amid, despite, over, during, from, follow, when, briefly,
Nearest to from: after, onto, via, into, with, about, against, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, billions, thousand,
Nearest to donald_trump: trump, trumps, everyone, hillary_clinton, obamas, republican, rex_tillerson, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: might, could, jul, november, apr, mar, jan, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_apnews.com, domain_businessinsider.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_washingtonpost.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_instapundit, subsource_SteveKopack, subsource_deray, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_theguardian.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_xeni, subsource_pierre, subsource_TimAeppel,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 722000: 2.571974
Average loss at step 724000: 2.544734
Average loss at step 726000: 2.558118
Average loss at step 728000: 2.553431
Average loss at step 730000: 2.540827
Nearest to job: career, payroll, incentive, employment, tenure, worker, intern, workforce,
Nearest to go: roar, really, come, get, crawl, honeymoon, sneak, margaret,
Nearest to make: remake, give, convince, rake, become, depict, steer, making,
Nearest to who: allegedly, young, portman, somehow, dont, unable, desperately, able,
Nearest to after: amid, despite, over, from, follow, when, during, briefly,
Nearest to from: after, via, about, onto, into, with, against, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, dubious, entirely,
Nearest to utc: 2017, shtml, pdf, jpg, htm, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, corbyn, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_www.cnn.com, grover,
Nearest to first: 1st, 2nd, 7th, second, maiden, every, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_WallStManeet, subsource_deray, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_rossfinley, subsource_pierre, subsource_dandrezner,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 732000: 2.535756
Average loss at step 734000: 2.535988
Average loss at step 736000: 2.525470
Average loss at step 738000: 2.522866
Average loss at step 740000: 2.522107
Nearest to job: career, payroll, workforce, employment, tenure, incentive, worker, intern,
Nearest to go: really, come, roar, crawl, get, honeymoon, margaret, sneak,
Nearest to make: give, remake, convince, rake, become, steer, depict, making,
Nearest to who: allegedly, young, somehow, dont, portman, unable, cant, accuse,
Nearest to after: amid, despite, over, from, during, before, follow, briefly,
Nearest to from: after, onto, via, with, into, infamous, about, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, dubious, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, arent, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_apnews.com, domain_businessinsider.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, groundbreaking, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_yahoo.com, domain_theguardian.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_wsj.com, domain_reuters.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_SteveKopack, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_pierre, subsource_dandrezner, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 742000: 2.535486
Average loss at step 744000: 2.532352
Average loss at step 746000: 2.529095
Average loss at step 748000: 2.517870
Average loss at step 750000: 2.530032
Nearest to job: career, payroll, employment, workforce, worker, incentive, tenure, intern,
Nearest to go: come, roar, really, crawl, fall, margaret, honeymoon, get,
Nearest to make: give, rake, convince, remake, become, steer, depict, evade,
Nearest to who: allegedly, young, somehow, portman, dont, unable, wan, accuse,
Nearest to after: amid, despite, over, from, during, before, briefly, when,
Nearest to from: after, onto, via, into, with, about, against, attempt,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: might, could, jul, apr, november, mar, jan, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_yahoo.com, domain_vox.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_instapundit, subsource_SteveKopack, subsource_WallStManeet, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_pierre, subsource_TimAeppel, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 752000: 2.514948
Average loss at step 754000: 2.534370
Average loss at step 756000: 2.564555
Average loss at step 758000: 2.540995
Average loss at step 760000: 2.552779
Nearest to job: career, payroll, employment, incentive, workforce, worker, tenure, intern,
Nearest to go: roar, really, come, crawl, get, margaret, honeymoon, sneak,
Nearest to make: remake, give, convince, rake, become, making, depict, steer,
Nearest to who: allegedly, young, somehow, portman, accuse, dont, unable, desperately,
Nearest to after: amid, despite, over, from, during, when, briefly, follow,
Nearest to from: after, into, via, about, onto, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_theguardian.com, domain_thehill.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_DividendMaster, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_rossfinley, subsource_TimAeppel, subsource_dandrezner,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_reddit/Economics, source_Here_Is_The_City, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 762000: 2.532365
Average loss at step 764000: 2.525389
Average loss at step 766000: 2.536320
Average loss at step 768000: 2.516938
Average loss at step 770000: 2.518137
Nearest to job: career, payroll, employment, incentive, tenure, workforce, intern, worker,
Nearest to go: come, really, roar, crawl, get, fall, honeymoon, margaret,
Nearest to make: give, remake, rake, convince, steer, become, depict, making,
Nearest to who: allegedly, young, somehow, portman, dont, unable, accuse, prosecute,
Nearest to after: amid, despite, over, from, when, briefly, during, before,
Nearest to from: after, via, onto, into, about, with, infamous, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, shameful, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, corbyn, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_apnews.com, domain_businessinsider.com, grover,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 5th, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_yahoo.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_washingtonpost.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_DividendMaster,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_marketwatch.com, domain_economist.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 772000: 2.513143
Average loss at step 774000: 2.527284
Average loss at step 776000: 2.525368
Average loss at step 778000: 2.519004
Average loss at step 780000: 2.512639
Nearest to job: career, payroll, workforce, worker, incentive, intern, tenure, employment,
Nearest to go: come, roar, really, crawl, get, margaret, borrow, fall,
Nearest to make: give, rake, convince, remake, become, steer, making, ramp,
Nearest to who: allegedly, young, somehow, dont, portman, accuse, wan, want,
Nearest to after: amid, despite, over, from, during, when, before, briefly,
Nearest to from: after, onto, via, into, about, with, attempt, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, shameful, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: jul, apr, might, could, jun, jan, november, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_www.cnn.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, groundbreaking, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_yahoo.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_SteveKopack, subsource_The_Real_Fly, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_pierre, subsource_dandrezner, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 782000: 2.519445
Average loss at step 784000: 2.514756
Average loss at step 786000: 2.515869
Average loss at step 788000: 2.559624
Average loss at step 790000: 2.531409
Nearest to job: career, payroll, worker, incentive, employment, workforce, tenure, intern,
Nearest to go: come, roar, really, get, crawl, sneak, honeymoon, fall,
Nearest to make: give, remake, rake, convince, steer, become, swallow, making,
Nearest to who: allegedly, young, somehow, portman, dont, accuse, wan, prosecute,
Nearest to after: amid, despite, over, from, follow, during, when, briefly,
Nearest to from: after, onto, into, via, about, with, attempt, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, shameful, republican,
Nearest to utc: 2017, shtml, pdf, jpg, share_idea_stocktwits, htm, mp4, aspx,
Nearest to may: might, could, november, jul, will, apr, jan, corbyn,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_www.cnn.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_SteveKopack, subsource_thegarance, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_theguardian.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_rossfinley, subsource_pierre, subsource_dandrezner,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_bloomberg.com, domain_apnews.com, domain_politico.com,
Average loss at step 792000: 2.547666
Average loss at step 794000: 2.540987
Average loss at step 796000: 2.514748
Average loss at step 798000: 2.527583
Average loss at step 800000: 2.512939
Nearest to job: career, payroll, employment, tenure, incentive, workforce, worker, intern,
Nearest to go: come, roar, really, crawl, get, margaret, fall, sneak,
Nearest to make: give, remake, rake, convince, become, steer, depict, swallow,
Nearest to who: allegedly, young, portman, somehow, dont, wan, desperately, unable,
Nearest to after: amid, despite, over, from, when, briefly, during, follow,
Nearest to from: after, via, onto, into, with, about, infamous, against,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, shameful, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, corbyn, mull,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, second, maiden, every, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_theguardian.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_ft.com, domain_washingtonpost.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 802000: 2.511637
Average loss at step 804000: 2.510808
Average loss at step 806000: 2.512236
Average loss at step 808000: 2.523640
Average loss at step 810000: 2.510293
Nearest to job: career, payroll, worker, workforce, employment, incentive, tenure, intern,
Nearest to go: come, really, roar, get, crawl, margaret, sneak, borrow,
Nearest to make: give, convince, rake, remake, steer, become, depict, making,
Nearest to who: allegedly, young, somehow, dont, portman, cant, accuse, wan,
Nearest to after: amid, despite, over, from, before, during, briefly, follow,
Nearest to from: after, onto, via, into, with, against, about, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, dubious,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, jul, november, apr, jun, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 8th, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_SteveKopack, subsource_deray, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_pierre, subsource_dandrezner, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 812000: 2.522078
Average loss at step 814000: 2.503170
Average loss at step 816000: 2.512911
Average loss at step 818000: 2.497842
Average loss at step 820000: 2.538898
Nearest to job: career, payroll, worker, workforce, employment, incentive, tenure, intern,
Nearest to go: come, really, roar, crawl, get, margaret, sneak, borrow,
Nearest to make: give, remake, rake, convince, steer, become, swallow, evade,
Nearest to who: allegedly, young, portman, somehow, dont, wan, accuse, anybody,
Nearest to after: amid, despite, over, from, follow, during, when, before,
Nearest to from: after, onto, into, via, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, jpg, share_idea_stocktwits, htm, mp4, aspx,
Nearest to may: could, might, jul, november, apr, jun, will, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com, domain_bloomberg.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_washingtonpost.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_The_Real_Fly, subsource_SteveKopack, subsource_thegarance, subsource_deray,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_pierre, subsource_dandrezner, subsource_TimAeppel,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 822000: 2.527471
Average loss at step 824000: 2.537516
Average loss at step 826000: 2.540530
Average loss at step 828000: 2.524134
Average loss at step 830000: 2.512621
Nearest to job: career, payroll, incentive, worker, intern, employment, tenure, workforce,
Nearest to go: really, come, roar, get, crawl, margaret, honeymoon, fall,
Nearest to make: give, remake, rake, convince, become, steer, depict, making,
Nearest to who: allegedly, young, somehow, portman, unable, desperately, dont, accuse,
Nearest to after: amid, despite, over, when, from, during, follow, briefly,
Nearest to from: after, via, about, onto, with, into, against, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, shameful, republican,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, corbyn, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, domain_www.cnn.com, grover,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_businessinsider.com, domain_wsj.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 832000: 2.513761
Average loss at step 834000: 2.500739
Average loss at step 836000: 2.511166
Average loss at step 838000: 2.496423
Average loss at step 840000: 2.516270
Nearest to job: career, payroll, employment, worker, workforce, incentive, intern, tenure,
Nearest to go: really, come, roar, crawl, get, margaret, borrow, fall,
Nearest to make: give, remake, rake, convince, steer, become, depict, transform,
Nearest to who: allegedly, young, somehow, portman, dont, cant, want, accuse,
Nearest to after: amid, despite, over, from, when, during, follow, before,
Nearest to from: after, onto, via, into, with, about, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, mull, arent, corbyn,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, grover,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 5th, groundbreaking,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_wsj.com, domain_reuters.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_memeorandum, subsource_blakehounshell, subsource_dandrezner, subsource_jessefelder, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Abnormal_Returns, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 842000: 2.512506
Average loss at step 844000: 2.508104
Average loss at step 846000: 2.496203
Average loss at step 848000: 2.509677
Average loss at step 850000: 2.494613
Nearest to job: career, payroll, worker, workforce, employment, incentive, tenure, intern,
Nearest to go: come, really, roar, crawl, get, fall, margaret, sneak,
Nearest to make: give, rake, convince, remake, become, steer, prove, depict,
Nearest to who: allegedly, young, portman, somehow, accuse, dont, formally, wan,
Nearest to after: amid, despite, over, from, during, when, follow, briefly,
Nearest to from: after, onto, about, into, via, against, with, attempt,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: might, could, jul, november, apr, mar, jun, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 5th, 8th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_washingtonpost.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_thegarance, subsource_instapundit, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_theguardian.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_SaraEisen, subsource_memeorandum, subsource_dandrezner, subsource_TimAeppel, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 852000: 2.508043
Average loss at step 854000: 2.550428
Average loss at step 856000: 2.512432
Average loss at step 858000: 2.537835
Average loss at step 860000: 2.522201
Nearest to job: career, payroll, employment, incentive, worker, tenure, intern, workforce,
Nearest to go: really, come, roar, crawl, get, margaret, sneak, honeymoon,
Nearest to make: give, remake, convince, rake, become, depict, steer, making,
Nearest to who: allegedly, young, portman, somehow, accuse, desperately, want, prosecute,
Nearest to after: amid, despite, over, from, when, during, briefly, follow,
Nearest to from: after, about, into, onto, via, with, against, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, shameful, republican,
Nearest to utc: 2017, shtml, pdf, jpg, htm, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_businessinsider.com, domain_thedailybeast.com, domain_apnews.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, second, every, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_thegarance, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_Frances_Coppola, subsource_thegarance, subsource_blakehounshell, subsource_SaraEisen, subsource_memeorandum, subsource_dandrezner, subsource_TimAeppel, subsource_jessefelder,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 862000: 2.504172
Average loss at step 864000: 2.509317
Average loss at step 866000: 2.502642
Average loss at step 868000: 2.499637
Average loss at step 870000: 2.495192
Nearest to job: career, payroll, employment, incentive, worker, workforce, intern, tenure,
Nearest to go: really, come, roar, crawl, get, margaret, fall, sneak,
Nearest to make: give, remake, convince, rake, become, steer, depict, making,
Nearest to who: allegedly, young, somehow, portman, cant, dont, accuse, prosecute,
Nearest to after: amid, despite, from, over, when, briefly, follow, during,
Nearest to from: after, via, onto, into, with, infamous, about, against,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, will, november, jul, apr, mar, corbyn,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_wsj.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 872000: 2.505891
Average loss at step 874000: 2.509569
Average loss at step 876000: 2.501751
Average loss at step 878000: 2.498781
Average loss at step 880000: 2.489886
Nearest to job: career, payroll, employment, worker, incentive, workforce, intern, tenure,
Nearest to go: come, really, roar, crawl, get, margaret, fall, html,
Nearest to make: give, convince, rake, remake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, dont, accuse, desperately, wan,
Nearest to after: amid, despite, from, over, during, when, follow, briefly,
Nearest to from: after, onto, via, into, about, with, against, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: jul, might, could, apr, jun, november, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_yahoo.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com, domain_economist.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 882000: 2.503353
Average loss at step 884000: 2.488281
Average loss at step 886000: 2.537603
Average loss at step 888000: 2.514668
Average loss at step 890000: 2.523815
Nearest to job: career, payroll, incentive, worker, workforce, employment, intern, tenure,
Nearest to go: really, come, roar, get, crawl, margaret, honeymoon, fall,
Nearest to make: give, remake, convince, rake, become, steer, depict, swallow,
Nearest to who: allegedly, somehow, young, portman, accuse, desperately, prosecute, dont,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, via, onto, into, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, will, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_washingtonpost.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_The_Real_Fly, subsource_thegarance,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_theguardian.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_TimAeppel,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 892000: 2.519958
Average loss at step 894000: 2.506254
Average loss at step 896000: 2.504931
Average loss at step 898000: 2.497990
Average loss at step 900000: 2.491816
Nearest to job: career, payroll, employment, incentive, tenure, intern, workforce, worker,
Nearest to go: really, come, roar, get, crawl, fall, honeymoon, margaret,
Nearest to make: give, remake, convince, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, cant, anybody, dont, accuse,
Nearest to after: amid, despite, over, from, when, briefly, during, follow,
Nearest to from: after, via, onto, about, with, into, infamous, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, grover, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_wsj.com, domain_reuters.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_dandrezner, subsource_SaraEisen, subsource_roburban, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 902000: 2.493629
Average loss at step 904000: 2.488521
Average loss at step 906000: 2.505968
Average loss at step 908000: 2.499968
Average loss at step 910000: 2.496632
Nearest to job: career, payroll, incentive, intern, workforce, worker, tenure, employment,
Nearest to go: come, really, roar, crawl, get, fall, margaret, honeymoon,
Nearest to make: give, convince, rake, remake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, dont, anybody, formally, cant,
Nearest to after: amid, despite, from, over, when, briefly, follow, during,
Nearest to from: after, onto, via, into, with, about, infamous, against,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, republican, rex_tillerson, dubious,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: might, jul, could, apr, november, jun, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, grover, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_theguardian.com, domain_bloomberg.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_thegarance, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_otherdomain, domain_bloomberg.com, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_marketwatch.com, domain_economist.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 912000: 2.486808
Average loss at step 914000: 2.498943
Average loss at step 916000: 2.481429
Average loss at step 918000: 2.504211
Average loss at step 920000: 2.532417
Nearest to job: career, payroll, incentive, worker, workforce, intern, employment, tenure,
Nearest to go: come, really, roar, crawl, get, fall, honeymoon, margaret,
Nearest to make: give, remake, rake, convince, become, steer, evade, ramp,
Nearest to who: allegedly, young, somehow, portman, anybody, accuse, dont, cant,
Nearest to after: amid, despite, from, over, when, follow, during, briefly,
Nearest to from: after, onto, into, via, about, against, with, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, republican, rex_tillerson, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, apr, will, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_washingtonpost.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_SteveKopack, subsource_thegarance, subsource_deray, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_theguardian.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 922000: 2.512524
Average loss at step 924000: 2.526813
Average loss at step 926000: 2.499750
Average loss at step 928000: 2.493635
Average loss at step 930000: 2.504364
Nearest to job: career, payroll, tenure, incentive, intern, employment, workforce, worker,
Nearest to go: come, really, roar, get, crawl, fall, honeymoon, margaret,
Nearest to make: give, remake, rake, convince, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, anybody, desperately, prosecute, formally,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, via, onto, into, about, with, against, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, republican, dubious,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, june,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, grover, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_washingtonpost.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 932000: 2.483395
Average loss at step 934000: 2.492681
Average loss at step 936000: 2.483597
Average loss at step 938000: 2.498727
Average loss at step 940000: 2.496197
Nearest to job: career, payroll, incentive, workforce, tenure, worker, intern, employment,
Nearest to go: come, really, roar, get, crawl, margaret, fall, honeymoon,
Nearest to make: give, convince, remake, rake, become, steer, depict, making,
Nearest to who: allegedly, young, somehow, portman, dont, convict, formally, cant,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, onto, into, via, with, against, about, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, republican, dubious,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, apr, will, mar, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, grover,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_wsj.com, domain_theguardian.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_SteveKopack, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_SaraEisen, subsource_memeorandum, subsource_blakehounshell, subsource_dandrezner, subsource_pierre, subsource_jessefelder,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 942000: 2.491313
Average loss at step 944000: 2.481379
Average loss at step 946000: 2.493901
Average loss at step 948000: 2.485029
Average loss at step 950000: 2.488218
Nearest to job: career, payroll, incentive, worker, tenure, workforce, intern, employment,
Nearest to go: come, really, roar, crawl, get, margaret, fall, html,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, formally, dont, anybody, accuse,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, onto, into, about, via, against, with, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, jpg, share_idea_stocktwits, mp4, aspx,
Nearest to may: could, might, jul, november, apr, mar, will, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_washingtonpost.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_SteveKopack, subsource_deray, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_SaraEisen, subsource_memeorandum, subsource_dandrezner, subsource_pierre, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 952000: 2.531454
Average loss at step 954000: 2.504489
Average loss at step 956000: 2.518839
Average loss at step 958000: 2.506578
Average loss at step 960000: 2.485481
Nearest to job: career, payroll, incentive, tenure, intern, worker, employment, workforce,
Nearest to go: come, really, roar, get, crawl, honeymoon, margaret, fall,
Nearest to make: give, convince, remake, rake, become, depict, steer, making,
Nearest to who: allegedly, young, somehow, portman, desperately, formally, dont, prosecute,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, onto, via, about, into, against, with, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, dubious, republican,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_businessinsider.com, domain_thedailybeast.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, maiden, second, every, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 962000: 2.501130
Average loss at step 964000: 2.481816
Average loss at step 966000: 2.486767
Average loss at step 968000: 2.478915
Average loss at step 970000: 2.480829
Nearest to job: career, payroll, incentive, workforce, employment, tenure, worker, intern,
Nearest to go: come, really, roar, get, crawl, margaret, fall, honeymoon,
Nearest to make: give, convince, remake, rake, become, steer, depict, swallow,
Nearest to who: allegedly, young, somehow, portman, dont, accuse, desperately, formally,
Nearest to after: amid, despite, from, over, when, during, follow, before,
Nearest to from: after, onto, via, into, with, against, about, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, republican, hillary_clinton, dubious,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, mull,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_yahoo.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_wsj.com, domain_washingtonpost.com, domain_reuters.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 972000: 2.499335
Average loss at step 974000: 2.483671
Average loss at step 976000: 2.495283
Average loss at step 978000: 2.475458
Average loss at step 980000: 2.487483
Nearest to job: career, payroll, incentive, worker, workforce, employment, tenure, intern,
Nearest to go: come, really, roar, crawl, get, margaret, fall, html,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, dont, formally, accuse, desperately,
Nearest to after: amid, despite, when, from, over, during, follow, briefly,
Nearest to from: after, onto, via, about, into, against, with, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, dubious,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, jul, apr, november, jun, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, maiden, every, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_SteveKopack, subsource_deray, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 982000: 2.475226
Average loss at step 984000: 2.514452
Average loss at step 986000: 2.503713
Average loss at step 988000: 2.516529
Average loss at step 990000: 2.508645
Nearest to job: career, payroll, incentive, employment, workforce, worker, tenure, intern,
Nearest to go: really, come, roar, get, crawl, margaret, honeymoon, html,
Nearest to make: give, remake, convince, rake, become, steer, depict, making,
Nearest to who: allegedly, young, portman, somehow, dont, accuse, desperately, formally,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, about, via, into, onto, against, with, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_bloomberg.com, domain_thedailybeast.com, domain_businessinsider.com, domain_apnews.com, domain_www.cnn.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_nytimes.com, domain_bloomberg.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_thegarance, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_theguardian.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Abnormal_Returns, source_Incidental_Economist, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 992000: 2.493213
Average loss at step 994000: 2.489041
Average loss at step 996000: 2.487321
Average loss at step 998000: 2.475691
Average loss at step 1000000: 2.480706
Nearest to job: career, payroll, incentive, employment, tenure, worker, workforce, intern,
Nearest to go: come, really, roar, get, crawl, honeymoon, margaret, fall,
Nearest to make: give, remake, convince, rake, become, steer, depict, transform,
Nearest to who: allegedly, young, somehow, portman, dont, prosecute, accuse, cant,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, via, onto, into, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, republican, rex_tillerson, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_vox.com, domain_theguardian.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_wsj.com, domain_washingtonpost.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_Bencjacobs,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1002000: 2.469362
Average loss at step 1004000: 2.490689
Average loss at step 1006000: 2.488858
Average loss at step 1008000: 2.482673
Average loss at step 1010000: 2.470905
Nearest to job: career, payroll, worker, workforce, incentive, employment, intern, tenure,
Nearest to go: come, really, roar, crawl, get, margaret, fall, html,
Nearest to make: give, convince, rake, become, remake, steer, depict, making,
Nearest to who: allegedly, young, somehow, portman, dont, formally, desperately, wan,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, onto, via, into, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, republican, rex_tillerson, dubious,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: jul, might, could, apr, november, jun, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_thegarance, subsource_SteveKopack, subsource_deray, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com, domain_yahoo.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_xeni,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1012000: 2.484018
Average loss at step 1014000: 2.466669
Average loss at step 1016000: 2.484273
Average loss at step 1018000: 2.523911
Average loss at step 1020000: 2.492182
Nearest to job: career, payroll, worker, workforce, incentive, employment, intern, tenure,
Nearest to go: come, really, roar, get, crawl, honeymoon, fall, margaret,
Nearest to make: give, convince, remake, become, rake, steer, depict, swallow,
Nearest to who: allegedly, young, portman, somehow, accuse, dont, formally, desperately,
Nearest to after: amid, despite, from, when, over, during, follow, before,
Nearest to from: after, onto, into, via, about, against, with, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 1bn,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, apr, will, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1022000: 2.516011
Average loss at step 1024000: 2.491135
Average loss at step 1026000: 2.477583
Average loss at step 1028000: 2.487172
Average loss at step 1030000: 2.475835
Nearest to job: career, payroll, employment, incentive, worker, workforce, intern, tenure,
Nearest to go: really, come, roar, get, crawl, honeymoon, margaret, fall,
Nearest to make: give, remake, convince, rake, become, steer, depict, transform,
Nearest to who: allegedly, young, portman, somehow, formally, cant, accuse, desperately,
Nearest to after: amid, despite, over, when, from, during, follow, briefly,
Nearest to from: after, via, into, onto, with, about, against, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_forbes.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1032000: 2.472936
Average loss at step 1034000: 2.469363
Average loss at step 1036000: 2.479815
Average loss at step 1038000: 2.489386
Average loss at step 1040000: 2.476729
Nearest to job: career, payroll, workforce, worker, employment, incentive, intern, tenure,
Nearest to go: come, really, roar, get, crawl, margaret, html, honeymoon,
Nearest to make: give, convince, rake, remake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, accuse, cant, formally, dont,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, onto, via, into, with, about, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, republican, rex_tillerson, dubious,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, jul, november, apr, mar, jun, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, memorable,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_theguardian.com, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_SteveKopack, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1042000: 2.471565
Average loss at step 1044000: 2.471141
Average loss at step 1046000: 2.478100
Average loss at step 1048000: 2.466778
Average loss at step 1050000: 2.513651
Nearest to job: career, payroll, worker, workforce, incentive, employment, tenure, intern,
Nearest to go: come, really, roar, get, crawl, fall, html, margaret,
Nearest to make: give, convince, remake, rake, become, steer, ramp, depict,
Nearest to who: allegedly, young, portman, somehow, accuse, dont, formally, cant,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, onto, into, via, about, against, with, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, jul, november, apr, will, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, maiden, second, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_thegarance, subsource_SteveKopack, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_wsj.com, domain_economist.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1052000: 2.489450
Average loss at step 1054000: 2.506505
Average loss at step 1056000: 2.497590
Average loss at step 1058000: 2.480927
Average loss at step 1060000: 2.484596
Nearest to job: career, payroll, incentive, tenure, worker, intern, workforce, employment,
Nearest to go: come, really, roar, get, crawl, honeymoon, margaret, fall,
Nearest to make: give, convince, remake, rake, become, steer, depict, making,
Nearest to who: allegedly, young, portman, somehow, accuse, formally, cant, dont,
Nearest to after: amid, despite, over, when, from, during, follow, briefly,
Nearest to from: after, via, onto, into, about, with, against, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, will, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_businessinsider.com, domain_thedailybeast.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_WallStManeet, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1062000: 2.474398
Average loss at step 1064000: 2.472422
Average loss at step 1066000: 2.471356
Average loss at step 1068000: 2.464193
Average loss at step 1070000: 2.480070
Nearest to job: career, payroll, workforce, worker, incentive, employment, intern, tenure,
Nearest to go: come, really, roar, get, crawl, fall, honeymoon, html,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, accuse, cant, dont, formally,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, onto, into, via, with, about, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, republican, rex_tillerson, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, mull,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_washingtonpost.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Abnormal_Returns, source_Capital_Spectator, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1072000: 2.476455
Average loss at step 1074000: 2.477456
Average loss at step 1076000: 2.465251
Average loss at step 1078000: 2.477508
Average loss at step 1080000: 2.457591
Nearest to job: career, payroll, worker, workforce, incentive, intern, employment, tenure,
Nearest to go: come, really, roar, get, crawl, fall, margaret, html,
Nearest to make: give, convince, rake, remake, become, steer, ramp, depict,
Nearest to who: allegedly, young, portman, somehow, accuse, formally, dont, anybody,
Nearest to after: amid, despite, over, from, when, during, follow, before,
Nearest to from: after, onto, into, via, about, with, against, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 100m,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, jeff_session,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, jul, november, apr, mar, jan, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_ft.com, domain_apnews.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_SteveKopack, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_marketwatch.com, domain_businessinsider.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1082000: 2.482999
Average loss at step 1084000: 2.509826
Average loss at step 1086000: 2.495417
Average loss at step 1088000: 2.503261
Average loss at step 1090000: 2.478793
Nearest to job: career, payroll, worker, incentive, employment, workforce, intern, tenure,
Nearest to go: come, really, roar, get, crawl, honeymoon, margaret, html,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, accuse, formally, dont, cant,
Nearest to after: amid, despite, over, when, from, during, follow, briefly,
Nearest to from: after, into, about, via, onto, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 100m,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, will, apr, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_dailymail.co.uk,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_bloomberg.com, domain_washingtonpost.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1092000: 2.473624
Average loss at step 1094000: 2.481731
Average loss at step 1096000: 2.460742
Average loss at step 1098000: 2.470006
Average loss at step 1100000: 2.461539
Nearest to job: career, payroll, worker, employment, workforce, incentive, intern, tenure,
Nearest to go: really, come, roar, get, crawl, margaret, honeymoon, fall,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, accuse, dont, cant, formally,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, via, onto, into, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 100m,
Nearest to donald_trump: trump, trumps, obamas, everyone, rex_tillerson, hillary_clinton, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_nytimes.com, domain_www.cnn.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_reuters.com, domain_washingtonpost.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1102000: 2.478735
Average loss at step 1104000: 2.478267
Average loss at step 1106000: 2.473993
Average loss at step 1108000: 2.460641
Average loss at step 1110000: 2.472709
Nearest to job: career, payroll, worker, workforce, incentive, employment, intern, tenure,
Nearest to go: come, really, roar, get, crawl, margaret, honeymoon, fall,
Nearest to make: give, convince, rake, remake, become, steer, ramp, depict,
Nearest to who: allegedly, young, portman, somehow, accuse, formally, dont, desperately,
Nearest to after: despite, amid, from, over, when, during, follow, briefly,
Nearest to from: after, into, onto, via, about, with, against, attempt,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, htm, share_idea_stocktwits, jpg, mp4, aspx,
Nearest to may: could, might, jul, apr, november, jun, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_SteveKopack, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1112000: 2.464891
Average loss at step 1114000: 2.465068
Average loss at step 1116000: 2.510571
Average loss at step 1118000: 2.486018
Average loss at step 1120000: 2.499250
Nearest to job: career, payroll, worker, incentive, employment, workforce, intern, tenure,
Nearest to go: come, really, roar, get, crawl, honeymoon, margaret, fall,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, accuse, formally, dont, desperately,
Nearest to after: despite, amid, over, from, when, during, follow, abruptly,
Nearest to from: after, into, via, onto, about, with, against, infamous,
Nearest to million: billion, 000, mln, bln, 5bn, 7bn, thousand, 100m,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, will, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_businessinsider.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_fortune.com, domain_apnews.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1122000: 2.487131
Average loss at step 1124000: 2.465830
Average loss at step 1126000: 2.483477
Average loss at step 1128000: 2.465351
Average loss at step 1130000: 2.464550
Nearest to job: career, payroll, employment, incentive, worker, intern, workforce, tenure,
Nearest to go: come, really, roar, get, crawl, honeymoon, margaret, fall,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, accuse, dont, formally, cant,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, via, into, onto, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 100m,
Nearest to donald_trump: trump, trumps, obamas, everyone, rex_tillerson, hillary_clinton, republican, shameful,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, grover,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_thegarance, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_roburban, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1132000: 2.462047
Average loss at step 1134000: 2.461273
Average loss at step 1136000: 2.475597
Average loss at step 1138000: 2.461733
Average loss at step 1140000: 2.471727
Nearest to job: career, payroll, worker, intern, workforce, incentive, employment, tenure,
Nearest to go: come, really, roar, get, crawl, margaret, honeymoon, fall,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, dont, want, accuse, formally,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, onto, into, via, with, about, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, obamas, everyone, hillary_clinton, rex_tillerson, republican, dubious,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, jul, apr, november, jun, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_SteveKopack, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_roburban, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1142000: 2.457589
Average loss at step 1144000: 2.467696
Average loss at step 1146000: 2.454560
Average loss at step 1148000: 2.497363
Average loss at step 1150000: 2.483403
Nearest to job: career, payroll, worker, incentive, intern, workforce, employment, tenure,
Nearest to go: come, really, roar, get, crawl, margaret, honeymoon, html,
Nearest to make: give, convince, remake, rake, become, steer, ramp, depict,
Nearest to who: allegedly, young, portman, somehow, accuse, dont, anybody, formally,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, onto, into, via, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, republican, jeff_session,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, apr, will, jan, mar,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_wsj.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_thegarance, subsource_The_Real_Fly, subsource_SteveKopack,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_reddit/Economics, source_Incidental_Economist, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1152000: 2.497727
Average loss at step 1154000: 2.492989
Average loss at step 1156000: 2.476528
Average loss at step 1158000: 2.467064
Average loss at step 1160000: 2.467825
Nearest to job: career, payroll, intern, incentive, employment, tenure, workforce, worker,
Nearest to go: come, really, roar, get, crawl, margaret, html, honeymoon,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, dont, anybody, cant, want,
Nearest to after: amid, despite, over, when, from, during, follow, briefly,
Nearest to from: after, via, into, onto, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, republican, entirely,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_thehill.com, domain_politico.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_thegarance, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1162000: 2.453918
Average loss at step 1164000: 2.463072
Average loss at step 1166000: 2.455838
Average loss at step 1168000: 2.472165
Average loss at step 1170000: 2.471235
Nearest to job: career, payroll, workforce, incentive, worker, intern, employment, tenure,
Nearest to go: come, really, roar, get, crawl, margaret, html, fall,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, somehow, portman, dont, want, anybody, accuse,
Nearest to after: amid, despite, from, over, when, during, follow, briefly,
Nearest to from: after, onto, into, via, with, about, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, republican, hillary_clinton, dubious,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, apr, will, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, memorable,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_whpresscorps, subsource_instapundit, subsource_thegarance, subsource_vcutrader, subsource_deray, subsource_SteveKopack, subsource_The_Real_Fly,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_SaraEisen, subsource_blakehounshell, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1172000: 2.464597
Average loss at step 1174000: 2.455603
Average loss at step 1176000: 2.468493
Average loss at step 1178000: 2.448910
Average loss at step 1180000: 2.466251
Nearest to job: career, payroll, worker, incentive, workforce, intern, tenure, employment,
Nearest to go: come, really, roar, get, crawl, fall, margaret, html,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, anybody, dont, accuse, formally,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, onto, into, via, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, hillary_clinton, rex_tillerson, republican, jeff_session,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, jul, november, apr, mar, will, jun,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_businessinsider.com, domain_bloomberg.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_roburban,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_wsj.com, domain_washingtonpost.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,
Average loss at step 1182000: 2.505152
Average loss at step 1184000: 2.476614
Average loss at step 1186000: 2.500059
Average loss at step 1188000: 2.473854
Average loss at step 1190000: 2.457510
Nearest to job: career, payroll, incentive, intern, worker, tenure, workforce, employment,
Nearest to go: really, come, roar, get, crawl, margaret, fall, honeymoon,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, anybody, dont, accuse, want,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, via, into, onto, about, with, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, 100m,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, hillary_clinton, republican, entirely,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, jul, will, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_thehill.com, domain_politico.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_deray, subsource_The_Real_Fly, subsource_thegarance, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_reuters.com, domain_otherdomain, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_blakehounshell, subsource_memeorandum, subsource_SaraEisen, subsource_dandrezner, subsource_rossfinley, subsource_pierre,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_bloomberg.com, domain_politico.com,
Average loss at step 1192000: 2.471102
Average loss at step 1194000: 2.457776
Average loss at step 1196000: 2.459769
Average loss at step 1198000: 2.452461
Average loss at step 1200000: 2.464022
Nearest to job: career, payroll, incentive, intern, worker, workforce, employment, tenure,
Nearest to go: come, really, roar, get, crawl, margaret, html, fall,
Nearest to make: give, convince, remake, rake, become, steer, depict, ramp,
Nearest to who: allegedly, young, portman, somehow, dont, anybody, want, accuse,
Nearest to after: amid, despite, over, from, when, during, follow, briefly,
Nearest to from: after, onto, via, into, with, about, against, infamous,
Nearest to million: billion, 000, mln, 5bn, bln, 7bn, thousand, billions,
Nearest to donald_trump: trump, trumps, everyone, obamas, rex_tillerson, republican, hillary_clinton, jeff_session,
Nearest to utc: 2017, shtml, pdf, share_idea_stocktwits, htm, jpg, mp4, aspx,
Nearest to may: could, might, november, will, jul, apr, mar, jan,
Nearest to domain_youtube.com: domain_otherdomain, domain_nytimes.com, domain_medium.com, domain_thedailybeast.com, domain_bloomberg.com, domain_businessinsider.com, domain_apnews.com, domain_ft.com,
Nearest to first: 1st, 2nd, 7th, every, second, maiden, 8th, 5th,
Nearest to domain_washingtonpost.com: domain_otherdomain, domain_www.cnn.com, domain_nytimes.com, domain_politico.com, domain_thehill.com, domain_theguardian.com, domain_vox.com, domain_yahoo.com,
Nearest to domain_otherdomain: domain_youtube.com, domain_bloomberg.com, domain_nytimes.com, domain_theguardian.com, domain_washingtonpost.com, domain_reuters.com, domain_wsj.com, domain_medium.com,
Nearest to subsource_DRUDGE_REPORT: subsource_memeorandum, subsource_instapundit, subsource_whpresscorps, subsource_vcutrader, subsource_thegarance, subsource_deray, subsource_The_Real_Fly, subsource_WallStManeet,
Nearest to domain_wsj.com: domain_otherdomain, domain_nytimes.com, domain_reuters.com, domain_washingtonpost.com, domain_bloomberg.com, domain_ft.com, domain_yahoo.com, domain_apnews.com,
Nearest to domain_ft.com: domain_bloomberg.com, domain_otherdomain, domain_reuters.com, domain_bbc.co.uk, domain_economist.com, domain_wsj.com, domain_businessinsider.com, domain_marketwatch.com,
Nearest to subsource_pdacosta: subsource_thegarance, subsource_Frances_Coppola, subsource_memeorandum, subsource_blakehounshell, subsource_SaraEisen, subsource_dandrezner, subsource_pierre, subsource_rossfinley,
Nearest to source_Huffington_Post: source_Business_Insider, source_NYT_Dealbook, source_Here_Is_The_City, source_Incidental_Economist, source_reddit/Economics, source_Capital_Spectator, source_Abnormal_Returns, source_othersource,
Nearest to domain_nytimes.com: domain_otherdomain, domain_washingtonpost.com, domain_wsj.com, domain_youtube.com, domain_apnews.com, domain_fortune.com, domain_politico.com, domain_bloomberg.com,

In [13]:

    

############################################################
# plot in 2D, after transforming 300-vectors with TSNE
############################################################


num_points = 300

tsne = TSNE(perplexity=30,
            n_components=2,
            init='pca',
            n_iter=5000)

two_d_embeddings = tsne.fit_transform(final_embeddings[1:num_points+1, :])

def plot(embeddings, labels):
  assert embeddings.shape[0] >= len(labels), 'More labels than embeddings'
  pylab.figure(figsize=(15,15))  # in inches
  for i, label in enumerate(labels):
    x, y = embeddings[i,:]
    pylab.scatter(x, y)
    pylab.annotate(label, xy=(x, y), xytext=(5, 2), textcoords='offset points',
                   ha='right', va='bottom')
  pylab.show()

words = [reverse_dictionary[i] for i in range(1, num_points+1)]
plot(two_d_embeddings, words)


    

In [12]:

    

###########################################################
# list collocated words
# allows us to add more mappings for common phrases
############################################################

import operator
import nltk
from nltk.collocations import *
with open(filename, 'r') as myfile:
        data = myfile.read()
# trigrams with at least 100 occurrences       
tokens = nltk.word_tokenize(data)
finder = TrigramCollocationFinder.from_words(tokens)
finder.apply_freq_filter(100)
z = finder.ngram_fd.viewitems()
print(sorted(z, key=operator.itemgetter(1), reverse=True))

# bigrams with at least 200 occurrences       
finder = BigramCollocationFinder.from_words(tokens)
finder.apply_freq_filter(200)
z = finder.ngram_fd.viewitems()
print(sorted(z, key=operator.itemgetter(1), reverse=True))


    

    




[(('subsource_othersubsource', 'subsource_othersubsource', 'subsource_othersubsource'), 7074), (('domain_otherdomain', 'subsource_othersubsource', 'source_Huffington_Post'), 6938), (('domain_otherdomain', 'subsource_othersubsource', 'subsource_othersubsource'), 5546), (('domain_otherdomain', 'subsource_othersubsource', 'source_reddit/Economics'), 2120), (('domain_otherdomain', 'subsource_othersubsource', 'source_NYT_Dealbook'), 2063), (('domain_otherdomain', 'subsource_othersubsource', 'source_Business_Insider'), 1905), (('pdf', 'domain_otherdomain', 'subsource_othersubsource'), 1674), (('domain_otherdomain', 'subsource_othersubsource', 'instagram_post'), 1332), (('domain_otherdomain', 'subsource_othersubsource', 'source_Here_Is_The_City'), 1308), (('subsource_EllevateNtwk', 'subsource_EllevateNtwk', 'subsource_EllevateNtwk'), 1307), (('domain_otherdomain', 'subsource_othersubsource', 'source_Abnormal_Returns'), 1258), (('domain_otherdomain', 'subsource_doctorow', 'danismm'), 1203), (('domain_otherdomain', 'subsource_othersubsource', 'source_Marginal_Revolution'), 1045), (('amazon_com', 'book', 'domain_otherdomain'), 856), (('domain_otherdomain', 'subsource_EllevateNtwk', 'subsource_EllevateNtwk'), 851), (('subsource_doctorow', 'subsource_doctorow', 'subsource_doctorow'), 787), (('utc', 'domain_otherdomain', 'subsource_othersubsource'), 764), (('domain_otherdomain', 'subsource_othersubsource', 'how'), 722), (('subsource_pdacosta', 'subsource_pdacosta', 'subsource_pdacosta'), 712), (('subsource_doctorow', 'danismm', 'domain_otherdomain'), 677), (('via', 'dailycaller', 'domain_otherdomain'), 670), (('domain_otherdomain', 'subsource_othersubsource', 'source_Incidental_Economist'), 653), (('domain_youtube.com', 'subsource_othersubsource', 'subsource_othersubsource'), 605), (('2017', 'domain_otherdomain', 'subsource_othersubsource'), 574), (('domain_otherdomain', 'subsource_doctorow', 'gameraboy'), 572), (('domain_otherdomain', 'subsource_othersubsource', 'source_othersource'), 490), (('subsource_hedge_funds', 'subsource_hedge_funds', 'subsource_hedge_funds'), 490), (('book', 'domain_otherdomain', 'subsource_othersubsource'), 480), (('domain_theguardian.com', 'subsource_othersubsource', 'source_Here_Is_The_City'), 473), (('domain_cnbc.com', 'subsource_othersubsource', 'source_Here_Is_The_City'), 439), (('subsource_pewresearch', 'subsource_pewresearch', 'subsource_pewresearch'), 413), (('domain_otherdomain', 'subsource_hedge_funds', 'hedge_fund'), 412), (('domain_otherdomain', 'subsource_SpirosMargaris', 'how'), 404), (('domain_ap.org', 'subsource_DRUDGE_REPORT', 'news_from_associate_press'), 389), (('trump', 'domain_otherdomain', 'subsource_othersubsource'), 365), (('domain_otherdomain', 'subsource_othersubsource', 'instagram_photo'), 364), (('first', '100', 'day'), 351), (('say', 'domain_otherdomain', 'subsource_othersubsource'), 350), (('domain_otherdomain', 'subsource_othersubsource', 'why'), 350), (('domain_otherdomain', 'subsource_SpirosMargaris', 'subsource_SpirosMargaris'), 335), (('domain_otherdomain', 'subsource_othersubsource', 'source_EconLog'), 331), (('domain_otherdomain', 'subsource_othersubsource', 'source_Capital_Spectator'), 329), (('domain_otherdomain', 'subsource_SpirosMargaris', 'subsource_JimMarous'), 322), (('domain_otherdomain', 'subsource_othersubsource', 'what'), 317), (('subsource_doctorow', 'gameraboy', 'domain_otherdomain'), 315), (('domain_otherdomain', 'subsource_EllevateNtwk', 'instagram_post'), 315), (('domain_otherdomain', 'subsource_DRUDGE_REPORT', 'subsource_memeorandum'), 310), (('domain_otherdomain', 'subsource_othersubsource', 'redirect'), 308), (('subsource_EllevateNtwk', 'instagram_post', 'ellevate'), 301), (('instagram_post', 'ellevate', 'network'), 300), (('domain_otherdomain', 'subsource_JimMarous', 'subsource_SpirosMargaris'), 296), (('subsource_othersubsource', 'source_Huffington_Post', 'trump'), 287), (('year', 'domain_otherdomain', 'subsource_othersubsource'), 287), (('domain_businessinsider.com', 'subsource_pdacosta', 'trump'), 285), (('subsource_DougKass', 'subsource_DougKass', 'subsource_DougKass'), 284), (('domain_otherdomain', 'subsource_aleks_kins', 'how'), 283), (('subsource_othersubsource', 'subsource_othersubsource', 'source_Huffington_Post'), 280), (('domain_businessinsider.com', 'subsource_pdacosta', 'subsource_pdacosta'), 271), (('domain_otherdomain', 'subsource_othersubsource', 'source_VOX'), 256), (('subsource_othersubsource', 'subsource_othersubsource', 'source_Abnormal_Returns'), 252), (('subsource_NoahCRothman', 'subsource_NoahCRothman', 'subsource_NoahCRothman'), 251), (('trump', 'immigration', 'order'), 251), (('domain_otherdomain', 'subsource_othersubsource', 'trump'), 244), (('domain_otherdomain', 'subsource_jonnajarian', 'instagram_post'), 243), (('instagram_post', 'jon', 'najarian'), 240), (('report', 'domain_otherdomain', 'subsource_othersubsource'), 238), (('domain_otherdomain', 'subsource_MissTrade', 'screen'), 234), (('domain_otherdomain', 'subsource_doctorow', 'subsource_doctorow'), 232), (('domain_otherdomain', 'subsource_JimMarous', 'subsource_JimMarous'), 232), (('subsource_jonnajarian', 'instagram_post', 'jon'), 232), (('subsource_jonallendc', 'sidewire', 'inc'), 232), (('domain_otherdomain', 'subsource_othersubsource', 'new'), 224), (('apple', 'podcasts', 'domain_otherdomain'), 223), (('subsource_othersubsource', 'source_Huffington_Post', 'how'), 223), (('subsource_mediatwit', 'subsource_mediatwit', 'subsource_mediatwit'), 221), (('instagram_post', 'erick', 'erickson'), 221), (('home', 'domain_otherdomain', 'subsource_othersubsource'), 219), (('subsource_EWErickson', 'instagram_post', 'erick'), 218), (('domain_otherdomain', 'subsource_othersubsource', 'this'), 217), (('domain_otherdomain', 'subsource_othersubsource', 'source_Crossing_Wall_Street'), 217), (('subsource_brianstelter', 'subsource_brianstelter', 'subsource_brianstelter'), 217), (('joy', 'ann', 'reid'), 215), (('subsource_johnmaeda', 'subsource_johnmaeda', 'subsource_johnmaeda'), 215), (('domain_reuters.com', 'subsource_othersubsource', 'source_Here_Is_The_City'), 212), (('subsource_othersubsource', 'subsource_RobinJPowell', 'source_Abnormal_Returns'), 211), (('domain_otherdomain', 'subsource_EWErickson', 'instagram_post'), 210), (('domain_otherdomain', 'subsource_valuewalk', '2017'), 210), (('2016', 'domain_otherdomain', 'subsource_othersubsource'), 210), (('subsource_JimMarous', 'subsource_JimMarous', 'subsource_JimMarous'), 203), (('domain_otherdomain', 'subsource_NoahCRothman', 'subsource_NoahCRothman'), 199), (('president', 'trump', 'domain_otherdomain'), 198), (('domain_otherdomain', 'subsource_jonallendc', 'sidewire'), 197), (('bbc', 'radio', 'domain_otherdomain'), 196), (('2017', 'pdf', 'domain_otherdomain'), 195), (('office', 'national', 'statistic'), 195), (('trump', 'travel_ban', 'domain_otherdomain'), 194), (('domain_otherdomain', 'subsource_benshapiro', 'subsource_benshapiro'), 192), (('kindle', 'store', 'domain_otherdomain'), 192), (('domain_otherdomain', 'subsource_othersubsource', '2017'), 191), (('subsource_JimMarous', 'subsource_SpirosMargaris', 'subsource_JimMarous'), 190), (('subsource_DRUDGE_REPORT', 'news_from_associate_press', 'domain_otherdomain'), 189), (('subsource_SpirosMargaris', 'subsource_SpirosMargaris', 'subsource_SpirosMargaris'), 189), (('us', 'domain_otherdomain', 'subsource_othersubsource'), 189), (('subsource_ValaAfshar', 'subsource_ValaAfshar', 'subsource_ValaAfshar'), 189), (('domain_otherdomain', 'subsource_SpirosMargaris', 'why'), 187), (('subsource_MaMoMVPY', 'subsource_MaMoMVPY', 'subsource_MaMoMVPY'), 186), (('subsource_othersubsource', 'subsource_othersubsource', 'source_Business_Insider'), 185), (('domain_nytimes.com', 'subsource_othersubsource', 'source_Huffington_Post'), 185), (('hedge_fund_news_from_hedgeco_net', 'domain_otherdomain', 'subsource_hedge_funds'), 184), (('domain_bloomberg.com', 'subsource_othersubsource', 'source_Crossing_Wall_Street'), 184), (('domain_otherdomain', 'subsource_SpirosMargaris', 'subsource_leimer'), 184), (('subsource_Richard_Florida', 'subsource_Richard_Florida', 'subsource_Richard_Florida'), 184), (('trump', 'administration', 'domain_otherdomain'), 182), (('day', 'domain_otherdomain', 'subsource_othersubsource'), 180), (('domain_bloomberg.com', 'subsource_othersubsource', 'source_Here_Is_The_City'), 179), (('subsource_JeffDSachs', 'subsource_JeffDSachs', 'subsource_JeffDSachs'), 178), (('domain_otherdomain', 'subsource_othersubsource', 'source_Calculated_Risk'), 178), (('right', 'now', 'domain_otherdomain'), 176), (('domain_otherdomain', 'subsource_ACLU', 'subsource_ACLU'), 176), (('can', 'learn', 'from'), 175), (('world', 'domain_otherdomain', 'subsource_othersubsource'), 175), (('domain_otherdomain', 'subsource_jonnajarian', 'subsource_othersubsource'), 174), (('subsource_othersubsource', 'redirect', 'domain_otherdomain'), 174), (('domain_otherdomain', 'subsource_moorehn', 'instagram_post'), 174), (('domain_otherdomain', 'subsource_aleks_kins', 'blockchain'), 174), (('subsource_othersubsource', 'source_Abnormal_Returns', 'how'), 172), (('may', '2017', 'domain_otherdomain'), 172), (('domain_otherdomain', 'subsource_mranti', 'domain_otherdomain'), 172), (('domain_otherdomain', 'subsource_loic', 'instagram_post'), 170), (('instagram_post', 'loic', 'meur'), 170), (('russia', 'probe', 'domain_otherdomain'), 169), (('subsource_loic', 'instagram_post', 'loic'), 169), (('domain_otherdomain', 'subsource_hedge_funds', 'subsource_hedge_funds'), 169), (('plan', 'domain_otherdomain', 'subsource_othersubsource'), 169), (('domain_youtube.com', 'subsource_othersubsource', 'source_Huffington_Post'), 168), (('with', 'trump', 'domain_otherdomain'), 167), (('subsource_shadihamid', 'subsource_shadihamid', 'subsource_shadihamid'), 167), (('brexit', 'domain_otherdomain', 'subsource_othersubsource'), 167), (('than', 'you', 'think'), 167), (('subsource_djrothkopf', 'subsource_djrothkopf', 'subsource_djrothkopf'), 166), (('dakota', 'access', 'pipeline'), 166), (('domain_otherdomain', 'subsource_DrPippaM', 'instagram_post'), 163), (('under', 'trump', 'domain_otherdomain'), 163), (('video', 'domain_otherdomain', 'subsource_othersubsource'), 163), (('america', 'domain_otherdomain', 'subsource_othersubsource'), 163), (('domain_wsj.com', 'subsource_aleks_kins', 'subsource_hedge_funds'), 162), (('subsource_SpirosMargaris', 'subsource_JimMarous', 'subsource_SpirosMargaris'), 162), (('domain_otherdomain', 'subsource_Livesquawk', 'press_release'), 162), (('bbc', 'world', 'service'), 162), (('kindle', 'ebooks', 'amazon_com'), 161), (('subsource_ProfSteveKeen', 'subsource_ProfSteveKeen', 'subsource_ProfSteveKeen'), 160), (('domain_otherdomain', 'subsource_moorehn', 'subsource_moorehn'), 160), (('subsource_DrPippaM', 'instagram_post', 'drpippam'), 159), (('subsource_othersubsource', 'subsource_othersubsource', 'source_NYT_Dealbook'), 158), (('trump', 'first', '100'), 157), (('domain_otherdomain', 'subsource_ValaAfshar', 'subsource_ValaAfshar'), 156), (('work', 'domain_otherdomain', 'subsource_othersubsource'), 155), (('instagram_post', 'joy', 'ann'), 154), (('domain_otherdomain', 'subsource_JoyAnnReid', 'instagram_post'), 153), (('subsource_othersubsource', 'subsource_othersubsource', 'how'), 153), (('march', '2017', 'domain_otherdomain'), 152), (('job', 'domain_otherdomain', 'subsource_othersubsource'), 151), (('product', 'hunt', 'domain_otherdomain'), 150), (('shoot', '2017', 'png'), 150), (('screen', 'shoot', '2017'), 150), (('you', 'think', 'domain_otherdomain'), 149), (('domain_medium.com', 'subsource_othersubsource', 'subsource_othersubsource'), 149), (('white_house', 'correspondent', 'dinner'), 149), (('domain_otherdomain', 'subsource_SpirosMargaris', 'artificial_intelligence'), 148), (('subsource_MissTrade', 'screen', 'shoot'), 148), (('subsource_ShaunKing', 'subsource_ShaunKing', 'subsource_ShaunKing'), 148), (('domain_otherdomain', 'subsource_othersubsource', 'watch'), 147), (('june', '2017', 'domain_otherdomain'), 147), (('subsource_MissTrade', 'subsource_MissTrade', 'subsource_MissTrade'), 147), (('saturday', 'night', 'live'), 147), (('market', 'domain_otherdomain', 'subsource_othersubsource'), 147), (('how', 'do', 'you'), 146), (('subsource_JoyAnnReid', 'instagram_post', 'joy'), 146), (('kevin', 'williamson', 'national_review'), 145), (('subsource_mattklewis', 'subsource_mattklewis', 'subsource_mattklewis'), 145), (('jonah', 'goldberg', 'national_review'), 145), (('tell', 'us', 'about'), 145), (('trump', 'era', 'domain_otherdomain'), 142), (('business', 'domain_otherdomain', 'subsource_othersubsource'), 142), (('election', 'domain_otherdomain', 'subsource_othersubsource'), 142), (('subsource_marketminute', 'subsource_marketminute', 'subsource_marketminute'), 141), (('subsource_othersubsource', 'source_Business_Insider', 'trump'), 140), (('official', 'say', 'domain_otherdomain'), 140), (('general', 'election', '2017'), 140), (('subsource_GlennKesslerWP', 'share', 'fact'), 140), (('study', 'find', 'domain_otherdomain'), 140), (('domain_otherdomain', 'subsource_GlennKesslerWP', 'share'), 140), (('subsource_othersubsource', 'scott', 'adam'), 140), (('after', 'brexit', 'domain_otherdomain'), 140), (('subsource_sarahkendzior', 'subsource_sarahkendzior', 'subsource_sarahkendzior'), 139), (('subsource_JasonLeopold', 'subsource_JasonLeopold', 'subsource_JasonLeopold'), 139), (('last', 'year', 'domain_otherdomain'), 139), (('life', 'domain_otherdomain', 'subsource_othersubsource'), 138), (('subsource_othersubsource', 'subsource_othersubsource', 'source_reddit/Economics'), 138), (('trump', 'tax', 'return'), 138), (('domain_otherdomain', 'subsource_othersubsource', 'source_AVC'), 138), (('up', 'domain_otherdomain', 'subsource_othersubsource'), 138), (('david', 'french', 'national_review'), 137), (('instapundit', 'blog', 'archive'), 137), (('domain_otherdomain', 'subsource_leimer', 'subsource_SpirosMargaris'), 137), (('domain_otherdomain', 'subsource_othersubsource', 'subsource_RobinJPowell'), 137), (('domain_otherdomain', 'subsource_marketminute', 'subsource_marketminute'), 135), (('obamacare', 'repeal', 'bill'), 135), (('subsource_hedge_funds', 'hedge_fund', 'implode_meter'), 135), (('subsource_JimMarous', 'subsource_JimMarous', 'subsource_SpirosMargaris'), 135), (('implode_meter', 'news', 'pick'), 135), (('news', 'pick', 'ups'), 135), (('hedge_fund', 'implode_meter', 'news'), 135), (('instagram_photo', 'video', 'domain_otherdomain'), 134), (('subsource_DanielJHannan', 'subsource_DanielJHannan', 'subsource_DanielJHannan'), 134), (('domain_otherdomain', 'subsource_jricole', 'instagram_post'), 133), (('everything', 'you_need_know', 'about'), 133), (('paris', 'climate', 'agreement'), 133), (('domain_otherdomain', 'subsource_econromesh', 'subsource_econromesh'), 133), (('domain_otherdomain', 'subsource_aleks_kins', 'subsource_SpirosMargaris'), 133), (('domain_otherdomain', 'subsource_seanmdav', 'subsource_seanmdav'), 132), (('domain_reuters.com', 'subsource_othersubsource', 'source_Crossing_Wall_Street'), 131), (('domain_otherdomain', 'subsource_othersubsource', 'sky'), 131), (('domain_imf.org', 'subsource_othersubsource', 'source_reddit/Economics'), 131), (('video', 'via', 'dailycaller'), 131), (('news', 'top', 'story'), 130), (('pre', 'existing', 'condition'), 129), (('domain_otherdomain', 'subsource_MichaelKitces', 'michael'), 129), (('domain_otherdomain', 'subsource_othersubsource', 'home'), 129), (('subsource_abrams', 'subsource_abrams', 'subsource_abrams'), 128), (('subsource_othersubsource', 'source_Huffington_Post', 'home'), 128), (('subsource_othersubsource', 'source_Abnormal_Returns', 'why'), 128), (('domain_otherdomain', 'subsource_JimMarous', 'subsource_leimer'), 128), (('domain_otherdomain', 'subsource_othersubsource', 'us'), 128), (('look', 'like', 'domain_otherdomain'), 128), (('subsource_othersubsource', 'source_Huffington_Post', 'why'), 127), (('you', 'domain_otherdomain', 'subsource_othersubsource'), 127), (('domain_otherdomain', 'subsource_doctorow', 'wilwheaton'), 127), (('subsource_aleks_kins', 'subsource_aleks_kins', 'subsource_aleks_kins'), 126), (('subsource_othersubsource', 'source_Huffington_Post', 'source_Huffington_Post'), 126), (('domain_otherdomain', 'subsource_othersubsource', 'best'), 126), (('time', 'domain_otherdomain', 'subsource_othersubsource'), 126), (('domain_bloomberg.com', 'subsource_othersubsource', 'subsource_othersubsource'), 125), (('policy', 'domain_otherdomain', 'subsource_othersubsource'), 125), (('deal', 'domain_otherdomain', 'subsource_othersubsource'), 124), (('subsource_SteveCase', 'subsource_SteveCase', 'subsource_SteveCase'), 124), (('make', 'america', 'great'), 124), (('domain_otherdomain', 'subsource_othersubsource', 'nasdaq'), 124), (('domain_otherdomain', 'subsource_doctorow', 'scificovers'), 124), (('national_review', 'domain_otherdomain', 'subsource_othersubsource'), 124), (('utc', 'domain_otherdomain', 'subsource_SpirosMargaris'), 123), (('domain_voxeu.org', 'subsource_FGoria', 'subsource_BaldwinRE'), 123), (('subsource_jowyang', 'subsource_jowyang', 'subsource_jowyang'), 123), (('company', 'domain_otherdomain', 'subsource_othersubsource'), 122), (('subsource_SpirosMargaris', 'subsource_JimMarous', 'subsource_JimMarous'), 122), (('sidewire', 'inc', 'domain_otherdomain'), 122), (('domain_otherdomain', 'subsource_activiststocks', 'subsource_activiststocks'), 122), (('subsource_othersubsource', 'sky', 'news'), 122), (('report', 'say', 'domain_otherdomain'), 121), (('domain_otherdomain', 'subsource_RichLowry', 'redirect'), 121), (('obamacare', 'repeal', 'domain_otherdomain'), 121), (('donorschoose', 'org', 'project'), 121), (('april', '2017', 'domain_otherdomain'), 120), (('redirect', 'domain_otherdomain', 'subsource_othersubsource'), 120), (('subsource_charlesarthur', 'start', 'up'), 120), (('georgia', 'special', 'election'), 120), (('domain_otherdomain', 'subsource_SpirosMargaris', 'what'), 119), (('rule', 'domain_otherdomain', 'subsource_othersubsource'), 119), (('international', 'woman', 'day'), 119), (('fund', 'domain_otherdomain', 'subsource_othersubsource'), 119), (('teach', 'us', 'about'), 119), (('economy', 'domain_otherdomain', 'subsource_othersubsource'), 119), (('domain_otherdomain', 'subsource_Livesquawk', 'subsource_AntBarton89'), 119), (('border', 'adjustment', 'tax'), 118), (('again', 'domain_otherdomain', 'subsource_othersubsource'), 118), (('immigration', 'order', 'domain_otherdomain'), 118), (('2017', '05pm', 'utc'), 118), (('store', 'domain_otherdomain', 'subsource_othersubsource'), 118), (('domain_otherdomain', 'subsource_Richard_Florida', 'subsource_Richard_Florida'), 118), (('domain_otherdomain', 'subsource_notayesmansecon', 'subsource_notayesmansecon'), 118), (('official', 'music', 'video'), 118), (('domain_washingtonpost.com', 'subsource_othersubsource', 'source_Huffington_Post'), 117), (('domain_reuters.com', 'subsource_othersubsource', 'source_reddit/Economics'), 117), (('instagram_post', 'heidi', 'moore'), 117), (('domain_thehill.com', 'subsource_memeorandum', 'trump'), 117), (('february', '2017', 'domain_otherdomain'), 117), (('subsource_moorehn', 'instagram_post', 'heidi'), 117), (('domain_otherdomain', 'subsource_hgetson', 'how'), 117), (('trump', 'inauguration', 'domain_otherdomain'), 116), (('next', 'week', 'domain_otherdomain'), 116), (('now', 'domain_otherdomain', 'subsource_othersubsource'), 116), (('domain_otherdomain', 'subsource_SpirosMargaris', 'fintech'), 116), (('subsource_othersubsource', 'source_Huffington_Post', 'what'), 115), (('domain_otherdomain', 'subsource_Livesquawk', 'moody'), 115), (('2017', '01pm', 'utc'), 115), (('domain_voxeu.org', 'subsource_FGoria', 'subsource_MarkThoma'), 114), (('2017', 'png', 'domain_otherdomain'), 114), (('instagram_post', 'juan', 'cole'), 114), (('subsource_jricole', 'instagram_post', 'juan'), 114), (('domain_otherdomain', 'subsource_othersubsource', 'domain_otherdomain'), 114), (('2017', '55pm', 'utc'), 114), (('paris', 'climate', 'accord'), 114), (('100', 'day', 'domain_otherdomain'), 113), (('greta', 'van', 'susteren'), 113), (('subsource_JimMarous', 'subsource_SpirosMargaris', 'subsource_leimer'), 113), (('2017', '22pm', 'utc'), 113), (('subsource_ivanhoff', 'market', 'wisdom'), 113), (('domain_otherdomain', 'subsource_othersubsource', 'amazon_com'), 113), (('danismm', 'domain_otherdomain', 'subsource_othersubsource'), 112), (('show', 'domain_otherdomain', 'subsource_othersubsource'), 112), (('martin', 'luther', 'king'), 112), (('2017', '26pm', 'utc'), 112), (('domain_otherdomain', 'subsource_alex', 'instagram_post'), 112), (('domain_otherdomain', 'subsource_ivanhoff', 'market'), 112), (('official', 'video', 'domain_otherdomain'), 112), (('woman', 'march', 'washington'), 111), (('2017', '09pm', 'utc'), 111), (('2017', '07pm', 'utc'), 111), (('subsource_instapundit', 'instapundit', 'blog'), 111), (('domain_otherdomain', 'subsource_toadmeister', 'subsource_toadmeister'), 110), (('domain_nytimes.com', 'subsource_othersubsource', 'source_Incidental_Economist'), 110), (('subsource_othersubsource', 'subsource_othersubsource', 'source_Marginal_Revolution'), 110), (('subsource_alex', 'instagram_post', 'alex'), 110), (('american', 'health_care', 'act'), 109), (('will', 'make', 'you'), 109), (('2017', '59pm', 'utc'), 109), (('domain_theguardian.com', 'subsource_othersubsource', 'source_Huffington_Post'), 109), (('2017', '30pm', 'utc'), 109), (('domain_forbes.com', 'subsource_othersubsource', 'source_Huffington_Post'), 109), (('2017', '11pm', 'utc'), 109), (('domain_otherdomain', 'subsource_instapundit', 'instapundit'), 109), (('2017', '39pm', 'utc'), 109), (('domain_otherdomain', 'subsource_fbonacci', 'instagram_post'), 108), (('2017', '50pm', 'utc'), 108), (('2017', '41pm', 'utc'), 108), (('national', 'statistic', 'domain_otherdomain'), 108), (('crisis', 'domain_otherdomain', 'subsource_othersubsource'), 108), (('2017', '34pm', 'utc'), 107), (('domain_reuters.com', 'subsource_othersubsource', 'source_Business_Insider'), 107), (('service', 'domain_otherdomain', 'subsource_othersubsource'), 107), (('2017', '43pm', 'utc'), 107), (('subsource_EconBrothers', 'daily', 'presidential_tracking_poll'), 107), (('subsource_JimMarous', 'subsource_leimer', 'subsource_SpirosMargaris'), 106), (('bureau', 'labor', 'statistic'), 106), (('2017', '38pm', 'utc'), 106), (('2017', '57pm', 'utc'), 105), (('subsource_othersubsource', 'subsource_othersubsource', 'source_Here_Is_The_City'), 105), (('app', 'store', 'domain_otherdomain'), 105), (('2017', '14pm', 'utc'), 105), (('subsource_activiststocks', 'subsource_activiststocks', 'subsource_activiststocks'), 105), (('domain_otherdomain', 'subsource_ProfSteveKeen', 'subsource_ProfSteveKeen'), 105), (('2017', '35pm', 'utc'), 105), (('domain_otherdomain', 'subsource_charlesarthur', 'start'), 104), (('subsource_TheEconomist', 'subsource_TheEconomist', 'subsource_TheEconomist'), 104), (('2017', '48pm', 'utc'), 104), (('2017', '19pm', 'utc'), 104), (('2017', '13pm', 'utc'), 104), (('subsource_KevinNR', 'subsource_KevinNR', 'subsource_KevinNR'), 104), (('subsource_othersubsource', 'instagram_post', 'tim'), 103), (('2017', '40pm', 'utc'), 103), (('subsource_OwenJones84', 'subsource_OwenJones84', 'subsource_OwenJones84'), 103), (('2017', '17pm', 'utc'), 103), (('instagram_post', 'tim', 'ferriss'), 103), (('2017', '31pm', 'utc'), 103), (('subsource_othersubsource', 'owen', 'thomas'), 103), (('kim', 'jong', 'nam'), 103), (('owen', 'thomas', 'workout'), 103), (('2017', '00pm', 'utc'), 103), (('2017', '47pm', 'utc'), 102), (('domain_otherdomain', 'subsource_othersubsource', 'do'), 101), (('great', 'again', 'domain_otherdomain'), 101), (('russia', 'domain_otherdomain', 'subsource_othersubsource'), 101), (('com', 'domain_otherdomain', 'subsource_othersubsource'), 101), (('domain_otherdomain', 'subsource_MissTrade', 'subsource_MissTrade'), 101), (('2017', '52pm', 'utc'), 100), (('2017', '44pm', 'utc'), 100), (('2017', '42pm', 'utc'), 100), (('mary', 'tyler', 'moore'), 100)]
[(('domain_otherdomain', 'subsource_othersubsource'), 63812), (('subsource_othersubsource', 'subsource_othersubsource'), 17776), (('subsource_othersubsource', 'source_Huffington_Post'), 10537), (('domain_otherdomain', 'subsource_SpirosMargaris'), 7230), (('domain_youtube.com', 'subsource_othersubsource'), 5404), (('pdf', 'domain_otherdomain'), 5370), (('utc', 'domain_otherdomain'), 4977), (('subsource_othersubsource', 'source_reddit/Economics'), 3959), (('subsource_othersubsource', 'source_Business_Insider'), 3943), (('domain_otherdomain', 'subsource_DRUDGE_REPORT'), 3729), (('subsource_othersubsource', 'source_Abnormal_Returns'), 3704), (('domain_otherdomain', 'subsource_doctorow'), 3606), (('domain_otherdomain', 'subsource_aleks_kins'), 3506), (('subsource_othersubsource', 'source_NYT_Dealbook'), 3386), (('subsource_othersubsource', 'source_Here_Is_The_City'), 3370), (('2017', 'domain_otherdomain'), 3327), (('domain_otherdomain', 'subsource_newsycombinator'), 3299), (('domain_otherdomain', 'subsource_JohnLothian'), 2851), (('domain_businessinsider.com', 'subsource_pdacosta'), 2704), (('domain_otherdomain', 'subsource_MissTrade'), 2664), (('domain_otherdomain', 'subsource_EllevateNtwk'), 2505), (('domain_otherdomain', 'subsource_hedge_funds'), 2449), (('president', 'trump'), 2430), (('domain_otherdomain', 'subsource_OpenOutcrier'), 2414), (('trump', 'domain_otherdomain'), 2308), (('subsource_EllevateNtwk', 'subsource_EllevateNtwk'), 2270), (('domain_otherdomain', 'subsource_moorehn'), 2202), (('domain_otherdomain', 'subsource_benshapiro'), 2078), (('say', 'domain_otherdomain'), 2064), (('domain_bloomberg.com', 'subsource_othersubsource'), 1924), (('subsource_othersubsource', 'source_Marginal_Revolution'), 1892), (('may', '2017'), 1815), (('domain_otherdomain', 'subsource_memeorandum'), 1771), (('mar', '2017'), 1651), (('trump', 'administration'), 1627), (('domain_otherdomain', 'subsource_firoozye'), 1609), (('apr', '2017'), 1609), (('domain_otherdomain', 'subsource_aussietorres'), 1603), (('domain_cnbc.com', 'subsource_othersubsource'), 1599), (('subsource_pdacosta', 'subsource_pdacosta'), 1597), (('domain_otherdomain', 'subsource_Livesquawk'), 1532), (('book', 'domain_otherdomain'), 1485), (('domain_otherdomain', 'subsource_AmyResnick'), 1483), (('year', 'domain_otherdomain'), 1483), (('domain_otherdomain', 'subsource_hgetson'), 1480), (('jan', '2017'), 1457), (('domain_otherdomain', 'subsource_instapundit'), 1446), (('feb', '2017'), 1425), (('domain_otherdomain', 'subsource_JimMarous'), 1409), (('jun', '2017'), 1391), (('amazon_com', 'book'), 1389), (('subsource_othersubsource', 'instagram_post'), 1388), (('subsource_othersubsource', 'source_Crossing_Wall_Street'), 1384), (('domain_otherdomain', 'subsource_whpresscorps'), 1379), (('domain_otherdomain', 'subsource_OccupyWallStNYC'), 1362), (('subsource_othersubsource', 'how'), 1353), (('domain_wsj.com', 'subsource_aleks_kins'), 1343), (('subsource_DRUDGE_REPORT', 'subsource_memeorandum'), 1313), (('report', 'domain_otherdomain'), 1308), (('domain_otherdomain', 'subsource_tunkuv'), 1296), (('domain_theguardian.com', 'subsource_othersubsource'), 1275), (('domain_nytimes.com', 'subsource_othersubsource'), 1269), (('domain_reuters.com', 'subsource_othersubsource'), 1266), (('subsource_doctorow', 'danismm'), 1204), (('via', 'dailycaller'), 1202), (('domain_otherdomain', 'subsource_leimer'), 1190), (('domain_otherdomain', 'subsource_monaeltahawy'), 1155), (('domain_otherdomain', 'subsource_sdonnan'), 1125), (('domain_otherdomain', 'subsource_lrozen'), 1120), (('domain_otherdomain', 'subsource_george_chen'), 1108), (('domain_otherdomain', 'subsource_WillauerProsky'), 1105), (('domain_otherdomain', 'subsource_raju'), 1105), (('domain_otherdomain', 'subsource_jricole'), 1097), (('domain_reuters.com', 'subsource_aleks_kins'), 1080), (('day', 'domain_otherdomain'), 1070), (('subsource_doctorow', 'subsource_doctorow'), 1053), (('domain_otherdomain', 'subsource_GreatDismal'), 1025), (('us', 'domain_otherdomain'), 1020), (('domain_otherdomain', 'subsource_craignewmark'), 1013), (('with', 'trump'), 1012), (('domain_otherdomain', 'subsource_JohnRentoul'), 1010), (('domain_otherdomain', 'subsource_seanmdav'), 1010), (('subsource_memeorandum', 'trump'), 988), (('domain_otherdomain', 'subsource_mranti'), 987), (('domain_otherdomain', 'subsource_HotlineJosh'), 985), (('domain_medium.com', 'subsource_othersubsource'), 982), (('election', 'domain_otherdomain'), 981), (('domain_otherdomain', 'subsource_chigrl'), 979), (('video', 'domain_otherdomain'), 967), (('domain_otherdomain', 'subsource_charlesornstein'), 959), (('domain_otherdomain', 'subsource_Techmeme'), 954), (('world', 'domain_otherdomain'), 953), (('domain_thehill.com', 'subsource_memeorandum'), 948), (('domain_yahoo.com', 'subsource_DRUDGE_REPORT'), 946), (('subsource_othersubsource', 'source_Incidental_Economist'), 936), (('domain_otherdomain', 'subsource_jonnajarian'), 936), (('domain_otherdomain', 'subsource_ACLU'), 923), (('domain_otherdomain', 'subsource_MichaelKitces'), 921), (('domain_otherdomain', 'subsource_ValaAfshar'), 914), (('domain_bloomberg.com', 'subsource_JohnLothian'), 905), (('domain_otherdomain', 'subsource_NoahCRothman'), 903), (('domain_otherdomain', 'subsource_deray'), 900), (('domain_otherdomain', 'subsource_peterwsinger'), 900), (('domain_wsj.com', 'subsource_othersubsource'), 885), (('domain_otherdomain', 'subsource_asibiza1'), 875), (('domain_otherdomain', 'subsource_ProfSteveKeen'), 866), (('market', 'domain_otherdomain'), 863), (('redirect', 'domain_otherdomain'), 855), (('work', 'domain_otherdomain'), 853), (('2016', 'domain_otherdomain'), 845), (('domain_otherdomain', 'subsource_NPR'), 837), (('domain_otherdomain', 'subsource_EWErickson'), 824), (('domain_otherdomain', 'subsource_noalpha_allbeta'), 822), (('domain_otherdomain', 'subsource_davewiner'), 820), (('plan', 'domain_otherdomain'), 817), (('national_review', 'domain_otherdomain'), 812), (('subsource_SpirosMargaris', 'subsource_SpirosMargaris'), 811), (('subsource_hedge_funds', 'subsource_hedge_funds'), 807), (('domain_otherdomain', 'subsource_Richard_Florida'), 806), (('hedge_fund_news_from_hedgeco_net', 'domain_otherdomain'), 806), (('domain_washingtonpost.com', 'subsource_GlennKesslerWP'), 803), (('domain_otherdomain', 'subsource_niubi'), 802), (('domain_otherdomain', 'subsource_PoliticsReid'), 801), (('trump', 'say'), 798), (('domain_otherdomain', 'subsource_PublishersWkly'), 798), (('domain_otherdomain', 'subsource_thegarance'), 793), (('brexit', 'domain_otherdomain'), 791), (('domain_otherdomain', 'subsource_LiquidNewsroom'), 789), (('obamacare', 'repeal'), 785), (('domain_otherdomain', 'subsource_RobinJPowell'), 783), (('deal', 'domain_otherdomain'), 778), (('job', 'domain_otherdomain'), 776), (('domain_otherdomain', 'subsource_JournalofValue'), 774), (('up', 'domain_otherdomain'), 773), (('america', 'domain_otherdomain'), 769), (('subsource_SpirosMargaris', 'subsource_JimMarous'), 760), (('home', 'domain_otherdomain'), 760), (('domain_otherdomain', 'subsource_AlanTonelson'), 757), (('domain_otherdomain', 'subsource_AntBarton89'), 756), (('business', 'domain_otherdomain'), 753), (('look', 'like'), 747), (('life', 'domain_otherdomain'), 740), (('how', 'do'), 739), (('domain_otherdomain', 'subsource_greta'), 738), (('interview', 'with'), 734), (('domain_otherdomain', 'subsource_DrPippaM'), 730), (('subsource_JimMarous', 'subsource_SpirosMargaris'), 719), (('time', 'domain_otherdomain'), 717), (('domain_otherdomain', 'subsource_ron_fournier'), 713), (('now', 'domain_otherdomain'), 711), (('domain_washingtonpost.com', 'subsource_othersubsource'), 711), (('trump', 'travel_ban'), 710), (('talk', 'about'), 707), (('subsource_pewresearch', 'subsource_pewresearch'), 699), (('trump', 'russia'), 697), (('subsource_othersubsource', 'why'), 692), (('domain_otherdomain', 'subsource_Kiffmeister'), 691), (('domain_otherdomain', 'subsource_Ian_Fraser'), 691), (('domain_otherdomain', 'subsource_maggieNYT'), 691), (('you', 'domain_otherdomain'), 689), (('danismm', 'domain_otherdomain'), 687), (('domain_otherdomain', 'subsource_econromesh'), 685), (('domain_otherdomain', 'subsource_mathewi'), 684), (('subsource_othersubsource', 'trump'), 683), (('again', 'domain_otherdomain'), 677), (('domain_otherdomain', 'subsource_20committee'), 677), (('domain_otherdomain', 'subsource_alex'), 676), (('domain_reuters.com', 'subsource_pdacosta'), 672), (('dailycaller', 'domain_otherdomain'), 670), (('subsource_pdacosta', 'trump'), 668), (('domain_otherdomain', 'subsource_valuewalk'), 653), (('domain_otherdomain', 'subsource_edwardlucas'), 651), (('domain_otherdomain', 'subsource_s_m_i'), 648), (('trump', 'budget'), 646), (('attack', 'domain_otherdomain'), 643), (('domain_otherdomain', 'subsource_BruceBartlett'), 643), (('html', 'domain_otherdomain'), 642), (('subsource_NoahCRothman', 'subsource_NoahCRothman'), 641), (('domain_otherdomain', 'subsource_volatilitysmile'), 635), (('domain_politico.com', 'subsource_memeorandum'), 635), (('domain_newyorker.com', 'subsource_cduhigg'), 635), (('domain_otherdomain', 'subsource_GreekAnalyst'), 634), (('domain_otherdomain', 'subsource_SteveCase'), 632), (('domain_otherdomain', 'subsource_JoyAnnReid'), 632), (('domain_otherdomain', 'subsource_JoeTrippi'), 631), (('domain_otherdomain', 'subsource_cfarivar'), 630), (('domain_otherdomain', 'subsource_stevesilberman'), 628), (('about', 'trump'), 627), (('100', 'day'), 626), (('subsource_JimMarous', 'subsource_JimMarous'), 623), (('domain_otherdomain', 'subsource_DividendMaster'), 618), (('domain_axios.com', 'subsource_inafried'), 617), (('domain_otherdomain', 'subsource_pointsnfigures'), 617), (('show', 'domain_otherdomain'), 614), (('domain_otherdomain', 'subsource_AnnPettifor'), 610), (('under', 'trump'), 608), (('domain_forbes.com', 'subsource_othersubsource'), 605), (('domain_otherdomain', 'subsource_MarkThoma'), 604), (('trump', 'have'), 604), (('subsource_othersubsource', 'what'), 604), (('say', 'trump'), 603), (('russia', 'domain_otherdomain'), 602), (('policy', 'domain_otherdomain'), 601), (('how', 'trump'), 600), (('domain_dealbreaker.com', 'subsource_hedge_funds'), 592), (('domain_otherdomain', 'subsource_GTCost'), 590), (('domain_otherdomain', 'subsource_ComfortablySmug'), 590), (('domain_otherdomain', 'subsource_toadmeister'), 589), (('elon', 'musk'), 589), (('subsource_thegarance', 'subsource_memeorandum'), 589), (('president', 'domain_otherdomain'), 588), (('domain_theguardian.com', 'subsource_GroomB'), 588), (('domain_otherdomain', 'subsource_abuaardvark'), 587), (('subsource_SpirosMargaris', 'how'), 586), (('woman', 'march'), 585), (('learn', 'from'), 585), (('after', 'trump'), 583), (('domain_otherdomain', 'subsource_Dvolatility'), 579), (('domain_otherdomain', 'subsource_davemcclure'), 577), (('economy', 'domain_otherdomain'), 576), (('subsource_doctorow', 'gameraboy'), 575), (('general', 'election'), 573), (('domain_otherdomain', 'subsource_pdacosta'), 572), (('domain_otherdomain', 'subsource_Wexboy_Value'), 566), (('rule', 'domain_otherdomain'), 565), (('russia', 'probe'), 565), (('future', 'domain_otherdomain'), 562), (('domain_otherdomain', 'subsource_hnshah'), 560), (('domain_otherdomain', 'subsource_SBarlow_ROB'), 560), (('domain_ft.com', 'subsource_othersubsource'), 559), (('domain_otherdomain', 'subsource_rmack'), 558), (('donald_trump', 'domain_otherdomain'), 558), (('anti', 'trump'), 557), (('subsource_hedge_funds', 'hedge_fund'), 553), (('crisis', 'domain_otherdomain'), 553), (('domain_otherdomain', 'subsource_Heritage'), 551), (('do', 'you'), 551), (('story', 'domain_otherdomain'), 550), (('more', 'domain_otherdomain'), 549), (('trump', 'tweet'), 547), (('domain_businessinsider.com', 'subsource_othersubsource'), 545), (('china', 'domain_otherdomain'), 543), (('domain_otherdomain', 'subsource_jaketapper'), 543), (('domain_otherdomain', 'subsource_RudyHavenstein'), 539), (('domain_otherdomain', 'subsource_robenfarzad'), 538), (('subsource_othersubsource', 'source_Capital_Spectator'), 538), (('subsource_othersubsource', 'source_othersource'), 538), (('domain_otherdomain', 'subsource_harrisj'), 533), (('will', 'not'), 533), (('news', 'domain_otherdomain'), 532), (('domain_otherdomain', 'subsource_mrtgr'), 531), (('subsource_othersubsource', 'source_EconLog'), 529), (('how', 'much'), 527), (('domain_otherdomain', 'subsource_dangillmor'), 526), (('domain_go.com', 'subsource_memeorandum'), 526), (('from', 'trump'), 525), (('domain_ap.org', 'subsource_DRUDGE_REPORT'), 524), (('fbi', 'director'), 523), (('domain_otherdomain', 'subsource_JasonHirschhorn'), 523), (('company', 'domain_otherdomain'), 521), (('domain_otherdomain', 'subsource_Longreads'), 520), (('do', 'domain_otherdomain'), 520), (('domain_otherdomain', 'subsource_NinjaEconomics'), 519), (('war', 'domain_otherdomain'), 516), (('bill', 'domain_otherdomain'), 516), (('subsource_ValaAfshar', 'subsource_ValaAfshar'), 515), (('domain_otherdomain', 'subsource_xeni'), 511), (('service', 'domain_otherdomain'), 510), (('right', 'domain_otherdomain'), 507), (('town', 'hall'), 503), (('lesson', 'from'), 502), (('law', 'domain_otherdomain'), 499), (('domain_otherdomain', 'subsource_pegobry'), 497), (('domain_otherdomain', 'subsource_Accendo_Mike'), 496), (('domain_theguardian.com', 'subsource_Bencjacobs'), 496), (('against', 'trump'), 496), (('problem', 'domain_otherdomain'), 495), (('you', 'think'), 495), (('domain_otherdomain', 'subsource_TheBubbleBubble'), 494), (('domain_otherdomain', 'subsource_JasonLeopold'), 493), (('domain_otherdomain', 'subsource_steve_hawkes'), 492), (('woman', 'domain_otherdomain'), 488), (('domain_otherdomain', 'subsource_KatrinaNation'), 488), (('domain_independent.co.uk', 'subsource_JohnRentoul'), 486), (('charge', 'with'), 484), (('domain_bbc.co.uk', 'subsource_othersubsource'), 484), (('domain_otherdomain', 'subsource_dabeard'), 484), (('domain_yahoo.com', 'subsource_othersubsource'), 483), (('domain_snappytv.com', 'subsource_othersubsource'), 483), (('fund', 'domain_otherdomain'), 480), (('domain_otherdomain', 'subsource_MOstwald1'), 480), (('store', 'domain_otherdomain'), 480), (('state', 'domain_otherdomain'), 478), (('com', 'domain_otherdomain'), 476), (('man', 'who'), 475), (('domain_otherdomain', 'subsource_jdmarkman'), 475), (('you', 'should'), 474), (('out', 'domain_otherdomain'), 473), (('worry', 'about'), 473), (('history', 'domain_otherdomain'), 472), (('domain_otherdomain', 'subsource_ClaraJeffery'), 471), (('domain_otherdomain', 'subsource_AnnCoulter'), 471), (('domain_otherdomain', 'subsource_Noahpinion'), 469), (('domain_otherdomain', 'subsource_benedictevans'), 468), (('domain_otherdomain', 'subsource_rcwhalen'), 468), (('why', 'trump'), 467), (('over', 'trump'), 467), (('domain_ft.com', 'subsource_JohnLothian'), 465), (('white_house', 'domain_otherdomain'), 465), (('french', 'election'), 463), (('domain_otherdomain', 'subsource_HamzeiAnalytics'), 463), (('tax', 'cut'), 462), (('trump', 'inauguration'), 461), (('its', 'own'), 461), (('domain_reuters.com', 'subsource_JohnLothian'), 461), (('domain_otherdomain', 'subsource_sarahcuda'), 459), (('meet', 'with'), 459), (('bbc', 'radio'), 459), (('domain_otherdomain', 'subsource_KevinNR'), 457), (('opinion', 'trump'), 456), (('source', 'domain_otherdomain'), 456), (('domain_otherdomain', 'subsource_IngrahamAngle'), 455), (('speech', 'domain_otherdomain'), 454), (('city', 'domain_otherdomain'), 454), (('that', 'will'), 453), (('subsource_memeorandum', 'subsource_othersubsource'), 453), (('domain_otherdomain', 'subsource_GuyKawasaki'), 452), (('domain_otherdomain', 'subsource_Mark_J_Perry'), 450), (('domain_otherdomain', 'subsource_DianeCoyle1859'), 448), (('domain_otherdomain', 'subsource_mikebutcher'), 447), (('domain_otherdomain', 'subsource_OpenSecretsDC'), 447), (('trump', 'first'), 447), (('subsource_othersubsource', 'this'), 447), (('case', 'domain_otherdomain'), 445), (('trump', 'white_house'), 445), (('vote', 'domain_otherdomain'), 445), (('domain_otherdomain', 'subsource_DowdEdward'), 443), (('meeting', 'with'), 440), (('trump', 'immigration'), 439), (('ellevate', 'network'), 439), (('money', 'domain_otherdomain'), 439), (('system', 'domain_otherdomain'), 438), (('domain_otherdomain', 'subsource_emptywheel'), 438), (('but', 'not'), 437), (('domain_otherdomain', 'subsource_cate_long'), 436), (('week', 'domain_otherdomain'), 436), (('after', 'brexit'), 436), (('conversation', 'with'), 435), (('up', 'with'), 435), (('them', 'domain_otherdomain'), 434), (('last', 'year'), 434), (('domain_otherdomain', 'subsource_vali_nasr'), 431), (('his', 'own'), 431), (('domain_forbes.com', 'subsource_SpirosMargaris'), 430), (('growth', 'domain_otherdomain'), 430), (('with', 'new'), 430), (('domain_otherdomain', 'subsource_alexisohanian'), 429), (('what', 'you'), 428), (('right', 'now'), 428), (('syria', 'domain_otherdomain'), 426), (('subsource_leimer', 'subsource_SpirosMargaris'), 426), (('domain_otherdomain', 'subsource_ebruenig'), 425), (('domain_otherdomain', 'subsource_ChristopherJM'), 422), (('domain_otherdomain', 'subsource_activiststocks'), 422), (('domain_youtube.com', 'subsource_RudyHavenstein'), 422), (('subsource_brianstelter', 'subsource_brianstelter'), 420), (('domain_otherdomain', 'subsource_ddayen'), 419), (('free', 'speech'), 419), (('data', 'domain_otherdomain'), 418), (('domain_otherdomain', 'subsource_abrams'), 418), (('find', 'domain_otherdomain'), 417), (('domain_qz.com', 'subsource_george_chen'), 417), (('domain_otherdomain', 'subsource_dannysullivan'), 416), (('industry', 'domain_otherdomain'), 416), (('domain_otherdomain', 'subsource_TonysAngle'), 416), (('march', '2017'), 416), (('people', 'domain_otherdomain'), 415), (('subsource_SpirosMargaris', 'subsource_leimer'), 415), (('domain_otherdomain', 'subsource_nickgillespie'), 414), (('probe', 'domain_otherdomain'), 413), (('utc', 'domain_youtube.com'), 410), (('country', 'domain_otherdomain'), 410), (('you', 'do'), 409), (('domain_dailymail.co.uk', 'subsource_jonnajarian'), 409), (('trump', 'era'), 408), (('apple', 'podcasts'), 407), (('domain_otherdomain', 'subsource_marketminute'), 407), (('domain_otherdomain', 'subsource_mediagazer'), 407), (('travel_ban', 'domain_otherdomain'), 407), (('trump', 'will'), 406), (('today', 'domain_otherdomain'), 406), (('risk', 'domain_otherdomain'), 403), (('domain_marketwatch.com', 'subsource_OpenOutcrier'), 402), (('betsy', 'devos'), 401), (('medium', 'domain_otherdomain'), 401), (('domain_otherdomain', 'subsource_BrankoMilan'), 400), (('subsource_JohnLothian', 'subsource_othersubsource'), 399), (('month', 'domain_otherdomain'), 395), (('office', 'domain_otherdomain'), 395), (('program', 'domain_otherdomain'), 394), (('domain_otherdomain', 'subsource_prchovanec'), 394), (('order', 'domain_otherdomain'), 394), (('domain_washingtonpost.com', 'subsource_thegarance'), 394), (('you', 'want'), 393), (('bank', 'domain_otherdomain'), 393), (('domain_otherdomain', 'subsource_sspencer_smb'), 392), (('review', 'domain_otherdomain'), 392), (('back', 'domain_otherdomain'), 392), (('domain_otherdomain', 'subsource_MaglanCapital'), 391), (('domain_dailymail.co.uk', 'subsource_DRUDGE_REPORT'), 391), (('subsource_othersubsource', 'subsource_memeorandum'), 391), (('domain_qz.com', 'subsource_cduhigg'), 390), (('subsource_DRUDGE_REPORT', 'news_from_associate_press'), 390), (('domain_otherdomain', 'subsource_SMTuffy'), 389), (('domain_otherdomain', 'subsource_beacongal'), 388), (('domain_reason.com', 'subsource_MattWelch'), 387), (('domain_otherdomain', 'subsource_pewresearch'), 386), (('away', 'from'), 386), (('domain_otherdomain', 'subsource_kimmaicutler'), 386), (('domain_otherdomain', 'subsource_W7VOA'), 386), (('voter', 'fraud'), 385), (('domain_arstechnica.com', 'subsource_cfarivar'), 385), (('domain_otherdomain', 'subsource_mehdirhasan'), 384), (('domain_otherdomain', 'subsource_digiphile'), 384), (('post', 'brexit'), 384), (('domain_ft.com', 'subsource_TabbFORUM'), 384), (('subsource_othersubsource', 'source_VOX'), 383), (('domain_otherdomain', 'subsource_MattGoldstein26'), 382), (('investor', 'domain_otherdomain'), 382), (('domain_huffingtonpost.com', 'subsource_lpolgreen'), 381), (('domain_otherdomain', 'subsource_andrewchen'), 381), (('that', 'could'), 380), (('domain_otherdomain', 'subsource_sunny_hundal'), 380), (('study', 'find'), 379), (('domain_otherdomain', 'subsource_anildash'), 379), (('domain_otherdomain', 'subsource_Revkin'), 378), (('make', 'you'), 378), (('subsource_ShaunKing', 'subsource_ShaunKing'), 377), (('sanctuary', 'city'), 377), (('result', 'domain_otherdomain'), 377), (('domain_bloomberg.com', 'subsource_MissTrade'), 377), (('subsource_JohnLothian', 'subsource_TabbFORUM'), 377), (('domain_wsj.com', 'subsource_WillauerProsky'), 376), (('game', 'domain_otherdomain'), 376), (('domain_otherdomain', 'subsource_tcarmody'), 375), (('domain_nypost.com', 'subsource_CGasparino'), 375), (('take', 'over'), 375), (('domain_www.cnn.com', 'subsource_CillizzaCNN'), 374), (('domain_otherdomain', 'subsource_stacyherbert'), 374), (('europe', 'domain_otherdomain'), 374), (('domain_otherdomain', 'subsource_joanwalsh'), 373), (('domain_otherdomain', 'subsource_pseudoerasmus'), 372), (('million', 'domain_otherdomain'), 372), (('subsource_Richard_Florida', 'subsource_Richard_Florida'), 372), (('trump', 'team'), 371), (('domain_otherdomain', 'subsource_Jason'), 371), (('jon', 'najarian'), 371), (('subsource_othersubsource', 'instagram_photo'), 370), (('think', 'domain_otherdomain'), 369), (('first', '100'), 369), (('him', 'domain_otherdomain'), 369), (('tax', 'return'), 368), (('official', 'say'), 368), (('immigration', 'order'), 368), (('sale', 'domain_otherdomain'), 367), (('reason', 'why'), 366), (('domain_otherdomain', 'subsource_quatremer'), 365), (('subsource_moorehn', 'subsource_moorehn'), 365), (('you_need_know', 'about'), 365), (('year', 'ago'), 364), (('tax', 'domain_otherdomain'), 364), (('trump', 'trade'), 363), (('domain_otherdomain', 'subsource_ShaunKing'), 363), (('claim', 'domain_otherdomain'), 362), (('trump', 'tax'), 362), (('domain_otherdomain', 'subsource_scottlincicome'), 362), (('rise', 'domain_otherdomain'), 362), (('domain_otherdomain', 'subsource_markos'), 361), (('wikipedia', 'domain_otherdomain'), 361), (('politics', 'domain_otherdomain'), 361), (('june', '2017'), 361), (('world', 'most'), 360), (('down', 'domain_otherdomain'), 360), (('power', 'domain_otherdomain'), 359), (('subsource_ProfSteveKeen', 'subsource_ProfSteveKeen'), 358), (('domain_wsj.com', 'subsource_JohnLothian'), 358), (('subsource_othersubsource', 'new'), 357), (('nuzzel', 'newsletter'), 356), (('domain_breitbart.com', 'subsource_memeorandum'), 355), (('fire', 'domain_otherdomain'), 355), (('domain_otherdomain', 'subsource_LWinthorpe'), 355), (('domain_otherdomain', 'subsource_johnmaeda'), 355), (('with', 'us'), 355), (('domain_www.cnn.com', 'subsource_othersubsource'), 354), (('domain_otherdomain', 'subsource_jackshafer'), 354), (('domain_otherdomain', 'subsource_charlesarthur'), 354), (('unite', 'airline'), 353), (('march', 'domain_otherdomain'), 352), (('step', 'down'), 352), (('problem', 'with'), 351), (('subsource_maggieNYT', 'subsource_memeorandum'), 351), (('domain_otherdomain', 'subsource_retheauditors'), 350), (('domain_marketwatch.com', 'subsource_othersubsource'), 350), (('jul', '2017'), 350), (('domain_otherdomain', 'subsource_amonck'), 349), (('news_from_associate_press', 'domain_otherdomain'), 349), (('domain_otherdomain', 'subsource_dinapomeranz'), 348), (('their', 'own'), 348), (('domain_ft.com', 'subsource_GroomB'), 348), (('trump', 'campaign'), 348), (('domain_otherdomain', 'subsource_DefenseBaron'), 348), (('you', 'have'), 346), (('talk', 'domain_otherdomain'), 346), (('domain_otherdomain', 'subsource_bfeld'), 346), (('domain_washingtonpost.com', 'subsource_memeorandum'), 346), (('special', 'election'), 345), (('stock', 'domain_otherdomain'), 345), (('domain_otherdomain', 'subsource_JonahNRO'), 344), (('domain_youtube.com', 'subsource_dougtee'), 344), (('domain_otherdomain', 'subsource_Dawn_Kopecki'), 344), (('start', 'up'), 344), (('fact', 'check'), 344), (('domain_otherdomain', 'subsource_jonallendc'), 344), (('domain_otherdomain', 'subsource_jbenton'), 344), (('domain_otherdomain', 'subsource_jdportes'), 343), (('grenfell', 'tower'), 343), (('work', 'with'), 342), (('may', 'not'), 342), (('what', 'trump'), 341), (('domain_washingtonpost.com', 'subsource_lrozen'), 341), (('wrong', 'domain_otherdomain'), 341), (('2017', 'pdf'), 341), (('domain_otherdomain', 'subsource_normative'), 340), (('domain_otherdomain', 'subsource_fbonacci'), 340), (('domain_fortune.com', 'subsource_othersubsource'), 340), (('group', 'domain_otherdomain'), 340), (('domain_otherdomain', 'subsource_stephenkinsella'), 340), (('cut', 'domain_otherdomain'), 339), (('win', 'domain_otherdomain'), 339), (('talk', 'with'), 339), (('record', 'domain_otherdomain'), 338), (('less', 'than'), 338), (('with', 'russia'), 338), (('source_Huffington_Post', 'trump'), 338), (('meeting', 'domain_otherdomain'), 338), (('trump', 'plan'), 337), (('domain_reuters.com', 'subsource_HowardArcherUK'), 337), (('subsource_benshapiro', 'subsource_benshapiro'), 337), (('subsource_othersubsource', 'subsource_RobinJPowell'), 335), (('woman', 'who'), 335), (('domain_otherdomain', 'subsource_arabist'), 334), (('price', 'domain_otherdomain'), 334), (('domain_voxeu.org', 'subsource_FGoria'), 334), (('leader', 'domain_otherdomain'), 333), (('domain_otherdomain', 'subsource_YanniKouts'), 332), (('domain_otherdomain', 'subsource_lizzieohreally'), 332), (('instagram_post', 'jon'), 331), (('trump', 'want'), 331), (('domain_reuters.com', 'subsource_FlitterOnFraud'), 330), (('will', 'make'), 330), (('domain_otherdomain', 'subsource_SonnyBunch'), 329), (('how', 'make'), 329), (('trump', 'order'), 329), (('its', 'time'), 329), (('subsource_aleks_kins', 'how'), 328), (('update', 'domain_otherdomain'), 327), (('domain_otherdomain', 'subsource_DavidBCollum'), 327), (('trump', 'voter'), 327), (('domain_otherdomain', 'subsource_carlzimmer'), 326), (('domain_otherdomain', 'subsource_johnvanreenen'), 326), (('gameraboy', 'domain_otherdomain'), 326), (('domain_otherdomain', 'subsource_jmartNYT'), 326), (('trump', 'call'), 325), (('domain_otherdomain', 'subsource_WilliamsJon'), 325), (('subsource_othersubsource', 'redirect'), 325), (('domain_huffingtonpost.com', 'subsource_NickBaumann'), 325), (('domain_reuters.com', 'subsource_TimAeppel'), 325), (('april', '2017'), 325), (('domain_buzzfeed.com', 'subsource_katherinemiller'), 324), (('subsource_SteveCase', 'subsource_SteveCase'), 324), (('live', 'domain_otherdomain'), 324), (('subsource_RobinJPowell', 'source_Abnormal_Returns'), 324), (('trump', 'presidency'), 324), (('subsource_othersubsource', 'us'), 324), (('erick', 'erickson'), 323), (('school', 'domain_otherdomain'), 323), (('domain_otherdomain', 'subsource_rachelsklar'), 323), (('tell', 'us'), 323), (('domain_otherdomain', 'subsource_jpodhoretz'), 323), (('too', 'much'), 322), (('paris', 'climate'), 322), (('domain_otherdomain', 'subsource_jonostrower'), 321), (('domain_otherdomain', 'subsource_BobBrinker'), 321), (('science', 'domain_otherdomain'), 320), (('domain_otherdomain', 'subsource_brianmlucey'), 320), (('subsource_mediatwit', 'subsource_mediatwit'), 319), (('domain_usatoday.com', 'subsource_jguynn'), 319), (('domain_otherdomain', 'subsource_brianstelter'), 318), (('year', 'after'), 318), (('matter', 'domain_otherdomain'), 318), (('pdf', 'domain_youtube.com'), 318), (('startup', 'domain_otherdomain'), 318), (('domain_nytimes.com', 'subsource_prashantrao'), 317), (('domain_washingtonexaminer.com', 'subsource_whpresscorps'), 317), (('subsource_EllevateNtwk', 'instagram_post'), 316), (('domain_otherdomain', 'subsource_ampressman'), 316), (('give', 'up'), 315), (('trade', 'domain_otherdomain'), 315), (('subsource_DougKass', 'subsource_DougKass'), 314), (('trump', 'adviser'), 314), (('domain_otherdomain', 'subsource_MaxBoot'), 314), (('obamacare', 'domain_otherdomain'), 314), (('part', 'domain_otherdomain'), 314), (('domain_otherdomain', 'subsource_derekwillis'), 314), (('house', 'republican'), 313), (('domain_otherdomain', 'subsource_ftrain'), 313), (('domain_otherdomain', 'subsource_daveweigel'), 312), (('thing', 'domain_otherdomain'), 312), (('far', 'right'), 312), (('threat', 'domain_otherdomain'), 312), (('change', 'domain_otherdomain'), 311), (('domain_otherdomain', 'subsource_Financial_Orbit'), 311), (('subsource_djrothkopf', 'subsource_djrothkopf'), 311), (('domain_bbc.com', 'subsource_othersubsource'), 311), (('domain_otherdomain', 'subsource_FlitterOnFraud'), 311), (('project', 'domain_otherdomain'), 311), (('one', 'domain_otherdomain'), 310), (('domain_otherdomain', 'subsource_matthewherper'), 310), (('domain_otherdomain', 'subsource_nycjim'), 310), (('domain_otherdomain', 'subsource_TheStalwart'), 310), (('next', 'week'), 309), (('know', 'domain_otherdomain'), 309), (('domain_otherdomain', 'subsource_brianbeutler'), 308), (('domain_huffingtonpost.com', 'subsource_CitizenCohn'), 308), (('subsource_mranti', 'domain_otherdomain'), 308), (('subsource_MissTrade', 'subsource_MissTrade'), 307), (('ever', 'domain_otherdomain'), 307), (('domain_otherdomain', 'subsource_jbouie'), 307), (('strategy', 'domain_otherdomain'), 306), (('age', 'domain_otherdomain'), 305), (('january', '2017'), 305), (('stand', 'up'), 305), (('campaign', 'domain_otherdomain'), 304), (('domain_www.cnn.com', 'subsource_memeorandum'), 304), (('kindle', 'store'), 304), (('subsource_memeorandum', 'subsource_davewiner'), 303), (('set', 'up'), 303), (('with', 'russian'), 302), (('domain_otherdomain', 'subsource_DougKass'), 302), (('subsource_ACLU', 'subsource_ACLU'), 301), (('domain_otherdomain', 'subsource_sarahkendzior'), 301), (('its', 'not'), 301), (('instagram_post', 'ellevate'), 301), (('congress', 'domain_otherdomain'), 301), (('domain_otherdomain', 'subsource_briansolis'), 301), (('subsource_othersubsource', '2017'), 300), (('what', 'happen'), 300), (('show', 'how'), 300), (('domain_nytimes.com', 'subsource_lrozen'), 299), (('trump', 'new'), 299), (('question', 'domain_otherdomain'), 299), (('subsource_shadihamid', 'subsource_shadihamid'), 298), (('domain_otherdomain', 'subsource_Dutch_Book'), 298), (('trump', 'supporter'), 298), (('domain_otherdomain', 'subsource_SultanAlQassemi'), 297), (('blog', 'domain_otherdomain'), 297), (('domain_otherdomain', 'subsource_mattklewis'), 297), (('domain_otherdomain', 'subsource_GlennKesslerWP'), 297), (('question', 'about'), 297), (('around', 'world'), 297), (('source_Abnormal_Returns', 'how'), 296), (('domain_otherdomain', 'subsource_christaylor_nyc'), 296), (('domain_otherdomain', 'subsource_jeffjarvis'), 296), (('after', 'year'), 296), (('domain_washingtonpost.com', 'subsource_HotlineJosh'), 296), (('domain_youtube.com', 'subsource_DavidClinchNews'), 296), (('domain_otherdomain', 'subsource_margarita'), 296), (('domain_otherdomain', 'subsource_mattdpearce'), 296), (('can', 'not'), 296), (('domain_otherdomain', 'subsource_TimHarford'), 296), (('domain_ft.com', 'subsource_Accendo_Mike'), 295), (('domain_ft.com', 'subsource_Frances_Coppola'), 295), (('domain_otherdomain', 'subsource_evanasmith'), 295), (('february', '2017'), 295), (('domain_otherdomain', 'subsource_saeedamenfx'), 294), (('domain_otherdomain', 'subsource_ddale8'), 294), (('go', 'domain_otherdomain'), 294), (('interview', 'domain_otherdomain'), 294), (('say', 'us'), 293), (('domain_reuters.com', 'subsource_lrozen'), 293), (('award', 'domain_otherdomain'), 293), (('not', 'just'), 293), (('mike', 'penny'), 293), (('domain_fortune.com', 'subsource_ampressman'), 293), (('domain_otherdomain', 'subsource_stevesi'), 292), (('fake_news', 'domain_otherdomain'), 292), (('record', 'high'), 292), (('domain_europa.eu', 'subsource_othersubsource'), 292), (('when', 'you'), 292), (('domain_otherdomain', 'subsource_davidfrum'), 291), (('us', 'about'), 291), (('subsource_activiststocks', 'subsource_activiststocks'), 291), (('gop', 'senator'), 291), (('shake', 'up'), 291), (('people', 'who'), 290), (('study', 'domain_otherdomain'), 290), (('domain_otherdomain', 'subsource_jessesingal'), 290), (('tweet', 'domain_otherdomain'), 290), (('subsource_lrozen', 'subsource_lrozen'), 289), (('subsource_JimMarous', 'subsource_leimer'), 289), (('subsource_lrozen', 'trump'), 289), (('civil', 'right'), 288), (('great', 'again'), 288), (('government', 'domain_otherdomain'), 288), (('domain_otherdomain', 'subsource_borzou'), 288), (('subsource_SpirosMargaris', 'why'), 288), (('truth', 'about'), 287), (('event', 'domain_otherdomain'), 287), (('about', 'domain_otherdomain'), 287), (('domain_otherdomain', 'subsource_Poynter'), 287), (('death', 'domain_otherdomain'), 286), (('come', 'from'), 286), (('test', 'domain_otherdomain'), 286), (('subsource_mattklewis', 'subsource_mattklewis'), 286), (('domain_otherdomain', 'subsource_carlbildt'), 285), (('can', 'help'), 284), (('party', 'domain_otherdomain'), 284), (('two', 'year'), 284), (('think', 'about'), 283), (('domain_otherdomain', 'subsource_Anthony'), 283), (('domain_otherdomain', 'subsource_maxkeiser'), 283), (('report', 'say'), 283), (('virtual', 'reality'), 282), (('subsource_BaldwinRE', 'subsource_BaldwinRE'), 282), (('subsource_JasonLeopold', 'subsource_JasonLeopold'), 282), (('domain_reuters.com', 'subsource_OpenOutcrier'), 282), (('domain_otherdomain', 'subsource_BretBaier'), 281), (('official', 'domain_otherdomain'), 281), (('all', 'domain_otherdomain'), 281), (('domain_otherdomain', 'subsource_nycsouthpaw'), 281), (('good', 'domain_otherdomain'), 281), (('domain_otherdomain', 'subsource_sanderwagner'), 281), (('this_year', 'domain_otherdomain'), 280), (('subsource_othersubsource', 'subsource_JohnLothian'), 280), (('subsource_marketminute', 'subsource_marketminute'), 279), (('administration', 'domain_otherdomain'), 279), (('way', 'domain_otherdomain'), 279), (('president', 'donald_trump'), 279), (('trump', 'pick'), 279), (('bill', 'reilly'), 278), (('say', 'about'), 278), (('domain_otherdomain', 'subsource_chrisgeidner'), 278), (('this', 'what'), 278), (('domain_otherdomain', 'subsource_inafried'), 278), (('obamacare', 'replacement'), 278), (('conference', 'domain_otherdomain'), 278), (('era', 'domain_otherdomain'), 277), (('domain_otherdomain', 'subsource_chrislhayes'), 277), (('subsource_monaeltahawy', 'subsource_monaeltahawy'), 277), (('protest', 'domain_otherdomain'), 276), (('election', '2017'), 276), (('domain_otherdomain', 'subsource_BrendanNyhan'), 276), (('america', 'first'), 276), (('dead', 'domain_otherdomain'), 276), (('domain_otherdomain', 'subsource_MarketChartist'), 275), (('domain_otherdomain', 'subsource_haroldpollack'), 275), (('subsource_JeffDSachs', 'subsource_JeffDSachs'), 274), (('domain_telegraph.co.uk', 'subsource_othersubsource'), 274), (('domain_otherdomain', 'subsource_JohnWaggoner'), 274), (('internet', 'thing'), 274), (('megyn', 'kelly'), 274), (('subsource_MaMoMVPY', 'subsource_MaMoMVPY'), 274), (('domain_otherdomain', 'subsource_loic'), 274), (('wonder', 'woman'), 274), (('domain_latimes.com', 'subsource_othersubsource'), 273), (('race', 'domain_otherdomain'), 273), (('say', 'they'), 273), (('car', 'domain_otherdomain'), 272), (('analysis', 'trump'), 272), (('alt', 'right'), 272), (('subsource_davemcclure', 'subsource_othersubsource'), 272), (('domain_telegraph.co.uk', 'subsource_DRUDGE_REPORT'), 272), (('subsource_johnmaeda', 'subsource_johnmaeda'), 272), (('domain_otherdomain', 'subsource_MartinSFP'), 271), (('just', 'get'), 271), (('george', 'bush'), 271), (('domain_otherdomain', 'subsource_asymco'), 271), (('domain_forbes.com', 'subsource_AntoineGara'), 270), (('domain_otherdomain', 'subsource_Frances_Coppola'), 270), (('jpg', 'domain_otherdomain'), 269), (('domain_otherdomain', 'subsource_GregMitch'), 269), (('its', 'first'), 269), (('domain_sec.gov', 'subsource_othersubsource'), 269), (('scandal', 'domain_otherdomain'), 269), (('domain_otherdomain', 'subsource_grossdm'), 269), (('subsource_MaxBoot', 'subsource_MaxBoot'), 269), (('domain_otherdomain', 'subsource_lachlan'), 269), (('climate_change', 'domain_otherdomain'), 268), (('billion', 'domain_otherdomain'), 268), (('inc', 'domain_otherdomain'), 268), (('return', 'domain_otherdomain'), 268), (('health_care', 'domain_otherdomain'), 268), (('investigation', 'into'), 268), (('radio', 'domain_otherdomain'), 268), (('domain_bloomberg.com', 'subsource_acrossthecurve'), 268), (('poll', 'domain_otherdomain'), 267), (('domain_otherdomain', 'subsource_trengriffin'), 267), (('make', 'america'), 267), (('trump', 'sign'), 266), (('can', 'you'), 266), (('domain_otherdomain', 'subsource_RadioFreeTom'), 266), (('raise', 'million'), 266), (('north_korea', 'domain_otherdomain'), 266), (('subsource_OwenJones84', 'subsource_OwenJones84'), 266), (('domain_npr.org', 'subsource_othersubsource'), 266), (('most', 'important'), 266), (('trump', 'executive_order'), 266), (('subsource_thegarance', 'trump'), 266), (('north', 'carolina'), 265), (('domain_otherdomain', 'subsource_mattyglesias'), 265), (('new', 'year'), 265), (('that', 'trump'), 264), (('off', 'domain_otherdomain'), 264), (('how', 'get'), 263), (('domain_otherdomain', 'subsource_adamfeuerstein'), 263), (('domain_otherdomain', 'subsource_TaylorLorenz'), 262), (('air', 'force'), 262), (('domain_bloomberg.com', 'subsource_pdacosta'), 262), (('product', 'hunt'), 262), (('subsource_econromesh', 'subsource_econromesh'), 262), (('star', 'war'), 261), (('high', 'domain_otherdomain'), 261), (('watch', 'live'), 261), (('domain_bloomberg.com', 'subsource_DRUDGE_REPORT'), 261), (('concern', 'about'), 261), (('sidewire', 'inc'), 261), (('that', 'make'), 261), (('how', 'can'), 260), (('domain_otherdomain', 'subsource_crampell'), 260), (('presidential', 'election'), 260), (('trump', 'aide'), 260), (('domain_usatoday.com', 'subsource_othersubsource'), 260), (('list', 'domain_otherdomain'), 260), (('challenge', 'domain_otherdomain'), 260), (('team', 'domain_otherdomain'), 259), (('right', 'wing'), 259), (('domain_wsj.com', 'subsource_DRUDGE_REPORT'), 259), (('tech', 'company'), 259), (('investigation', 'domain_otherdomain'), 259), (('domain_otherdomain', 'subsource_DonutShorts'), 259), (('research', 'domain_otherdomain'), 259), (('domain_otherdomain', 'subsource_AlderLaneeggs'), 258), (('issue', 'domain_otherdomain'), 258), (('brexit', 'talk'), 258), (('healthcare', 'bill'), 256), (('trump', 'make'), 256), (('png', 'domain_otherdomain'), 256), (('domain_otherdomain', 'subsource_3747S_14458E'), 256), (('app', 'store'), 255), (('middle', 'class'), 255), (('evidence', 'from'), 255), (('domain_politico.com', 'subsource_maggieNYT'), 255), (('idea', 'domain_otherdomain'), 254), (('act', 'domain_otherdomain'), 254), (('domain_otherdomain', 'subsource_chrismessina'), 254), (('trump', 'america'), 254), (('twitter', 'domain_otherdomain'), 253), (('domain_ft.com', 'subsource_kadhimshubber'), 253), (('than', 'you'), 253), (('investment', 'domain_otherdomain'), 252), (('domain_ft.com', 'subsource_Financial_Orbit'), 252), (('real', 'time'), 252), (('call', 'trump'), 252), (('domain_otherdomain', 'subsource_edwardnh'), 252), (('mental', 'health'), 252), (('podcasts', 'domain_otherdomain'), 252), (('ban', 'domain_otherdomain'), 251), (('domain_qz.com', 'subsource_othersubsource'), 251), (('civil', 'war'), 251), (('domain_bloomberg.com', 'subsource_TabbFORUM'), 250), (('worker', 'domain_otherdomain'), 250), (('will', 'have'), 250), (('utc', 'domain_nytimes.com'), 250), (('washington', 'domain_otherdomain'), 250), (('domain_otherdomain', 'subsource_realJohnVerley'), 250), (('line', 'domain_otherdomain'), 250), (('house', 'gop'), 249), (('domain_otherdomain', 'subsource_Jesse_Livermore'), 249), (('fight', 'domain_otherdomain'), 249), (('domain_theintercept.com', 'subsource_ggreenwald'), 249), (('reform', 'domain_otherdomain'), 248), (('thing', 'you'), 248), (('technology', 'domain_otherdomain'), 248), (('new', 'book'), 248), (('domain_otherdomain', 'subsource_AyeshaASiddiqi'), 248), (('tech', 'startup'), 247), (('domain_otherdomain', 'subsource_DanielJHannan'), 247), (('reality', 'domain_otherdomain'), 247), (('fight', 'against'), 247), (('subsource_jonnajarian', 'instagram_post'), 247), (('subsource_whpresscorps', 'trump'), 247), (('domain_washingtonpost.com', 'subsource_maggieNYT'), 247), (('domain_otherdomain', 'subsource_blakehounshell'), 247), (('network', 'domain_otherdomain'), 246), (('blog', 'archive'), 246), (('domain_otherdomain', 'subsource_nachkari'), 246), (('president', 'obama'), 246), (('domain_otherdomain', 'subsource_kausmickey'), 246), (('2018', 'domain_otherdomain'), 246), (('domain_otherdomain', 'subsource_alaidi'), 246), (('help', 'you'), 246), (('21st', 'century'), 246), (('your', 'life'), 246), (('new', 'study'), 246), (('domain_otherdomain', 'subsource_EconBrothers'), 246), (('center', 'domain_otherdomain'), 245), (('police', 'say'), 245), (('subsource_abrams', 'subsource_abrams'), 245), (('domain_otherdomain', 'subsource_mmasnick'), 245), (('amazon', 'uk'), 245), (('domain_otherdomain', 'subsource_munilass'), 245), (('about', 'his'), 245), (('inauguration', 'domain_otherdomain'), 245), (('domain_otherdomain', 'subsource_CitizenCohn'), 244), (('domain_buzzfeed.com', 'subsource_AlbertoNardelli'), 244), (('trump', 'do'), 244), (('domain_otherdomain', 'subsource_Fahrenthold'), 243), (('how', 'many'), 243), (('domain_otherdomain', 'subsource_fmanjoo'), 243), (('border', 'tax'), 243), (('100', '000'), 242), (('subsource_MissTrade', 'screen'), 242), (('help', 'us'), 242), (('domain_nytimes.com', 'subsource_tripgabriel'), 242), (('subsource_pdacosta', 'subsource_Frances_Coppola'), 242), (('charge', 'domain_otherdomain'), 242), (('could', 'have'), 241), (('domain_otherdomain', 'subsource_PrometheusAM'), 241), (('domain_otherdomain', 'subsource_CGasparino'), 241), (('2017', 'domain_youtube.com'), 241), (('domain_otherdomain', 'subsource_limlouisa'), 241), (('american', 'domain_otherdomain'), 241), (('chart', 'domain_otherdomain'), 241), (('explain', 'why'), 241), (('domain_otherdomain', 'subsource_billmckibben'), 241), (('domain_otherdomain', 'subsource_notayesmansecon'), 241), (('us', 'military'), 240), (('from', 'us'), 240), (('statistic', 'domain_otherdomain'), 240), (('will', 'trump'), 240), (('source_Huffington_Post', 'how'), 240), (('end', 'domain_otherdomain'), 239), (('do', 'know'), 239), (('subsource_DanielJHannan', 'subsource_DanielJHannan'), 239), (('domain_businessinsider.com', 'subsource_WillauerProsky'), 239), (('domain_otherdomain', 'subsource_RichardvReeves'), 239), (('domain_otherdomain', 'subsource_CassSunstein'), 238), (('domain_otherdomain', 'subsource_Benioff'), 238), (('domain_otherdomain', 'subsource_mhewson_CMC'), 238), (('can', 'do'), 238), (('watch', 'domain_otherdomain'), 238), (('domain_otherdomain', 'subsource_kimseverson'), 238), (('subsource_othersubsource', 'watch'), 238), (('this', 'how'), 237), (('domain_otherdomain', 'subsource_SlaughterAM'), 237), (('subsource_aleks_kins', 'subsource_hedge_funds'), 237), (('you', 'know'), 237), (('subsource_JohnLothian', 'subsource_JohnLothian'), 237), (('domain_otherdomain', 'subsource_MacroPru'), 237), (('domain_otherdomain', 'subsource_OwenJones84'), 237), (('domain_otherdomain', 'subsource_SteveForbesCEO'), 237), (('trump', 'could'), 237), (('your', 'business'), 236), (('find', 'out'), 236), (('domain_otherdomain', 'subsource_marcoarment'), 236), (('100', 'million'), 236), (('subsource_othersubsource', 'source_Calculated_Risk'), 236), (('tell', 'you'), 236), (('european', 'commission'), 235), (('case', 'against'), 235), (('domain_otherdomain', 'subsource_willwilkinson'), 234), (('repeal', 'obamacare'), 234), (('trump', 'take'), 234), (('subsource_pdacosta', 'subsource_othersubsource'), 234), (('pdf', 'domain_bloomberg.com'), 234), (('new', 'zealand'), 234), (('student', 'loan'), 234), (('subsource_jonallendc', 'sidewire'), 233), (('domain_otherdomain', 'subsource_rrhoover'), 233), (('domain_otherdomain', 'subsource_SunFoundation'), 233), (('say', 'will'), 233), (('fact', 'domain_otherdomain'), 233), (('make', 'sense'), 233), (('domain_documentcloud.org', 'subsource_chrisgeidner'), 232), (('domain_otherdomain', 'subsource_BarbarianCap'), 232), (('domain_otherdomain', 'subsource_pourmecoffee'), 232), (('domain_otherdomain', 'subsource_MorrisseyHelena'), 232), (('domain_otherdomain', 'subsource_marshallk'), 232), (('domain_bloomberg.com', 'subsource_Wexboy_Value'), 232), (('presidency', 'domain_otherdomain'), 232), (('than', 'ever'), 232), (('too', 'domain_otherdomain'), 231), (('ceo', 'domain_otherdomain'), 231), (('what', 'happens'), 231), (('court', 'domain_otherdomain'), 231), (('subsource_sarahkendzior', 'subsource_sarahkendzior'), 231), (('domain_otherdomain', 'subsource_OpenSociety'), 231), (('trump', 'just'), 230), (('world', 'large'), 230), (('five', 'year'), 230), (('domain_otherdomain', 'subsource_CarlBialik'), 230), (('domain_cnbc.com', 'subsource_SteveKopack'), 230), (('comey', 'fire'), 230), (('subsource_othersubsource', 'subsource_MissTrade'), 230), (('step', 'up'), 230), (('rate', 'domain_otherdomain'), 229), (('domain_otherdomain', 'subsource_QTRResearch'), 229), (('launch', 'new'), 229), (('domain_otherdomain', 'subsource_NickatFP'), 229), (('uk', 'domain_otherdomain'), 229), (('uk', 'election'), 229), (('care', 'about'), 229), (('domain_politico.com', 'subsource_Hadas_Gold'), 229), (('oil', 'gas'), 229), (('health_care', 'plan'), 229), (('know', 'what'), 229), (('domain_otherdomain', 'subsource_alexisgoldstein'), 228), (('milo', 'yiannopoulos'), 228), (('debate', 'domain_otherdomain'), 228), (('pay', 'domain_otherdomain'), 228), (('cyber', 'attack'), 228), (('conflict', 'interest'), 228), (('say', 'she'), 227), (('your', 'own'), 227), (('instagram_photo', 'video'), 227), (('domain_otherdomain', 'subsource_guan'), 227), (('you', 'get'), 226), (('economics', 'domain_otherdomain'), 226), (('domain_otherdomain', 'subsource_jowyang'), 226), (('domain_otherdomain', 'subsource_grodaeu'), 226), (('domain_otherdomain', 'subsource_CharlesMBlow'), 226), (('hard', 'brexit'), 225), (('house', 'domain_otherdomain'), 225), (('instagram', 'video'), 225), (('like', 'domain_otherdomain'), 225), (('subsource_oddvestor', 'subsource_activiststocks'), 225), (('domain_otherdomain', 'subsource_EricBalchunas'), 224), (('domain_otherdomain', 'subsource_NickBaumann'), 224), (('what', 'will'), 224), (('domain_yahoo.com', 'subsource_OpenOutcrier'), 224), (('trump', 'attack'), 224), (('subsource_seanmdav', 'subsource_seanmdav'), 223), (('subsource_DRUDGE_REPORT', 'trump'), 223), (('account', 'domain_otherdomain'), 223), (('loss', 'domain_otherdomain'), 223), (('billion', 'dollar'), 223), (('do', 'want'), 223), (('us', 'trade'), 223), (('happen', 'domain_otherdomain'), 223), (('balance', 'sheet'), 223), (('domain_otherdomain', 'subsource_odtorson'), 223), (('subsource_GlennKesslerWP', 'subsource_GlennKesslerWP'), 223), (('domain_otherdomain', 'subsource_markmackinnon'), 223), (('subsource_jonnajarian', 'subsource_othersubsource'), 223), (('health', 'plan'), 223), (('domain_theverge.com', 'subsource_Techmeme'), 222), (('budget', 'domain_otherdomain'), 222), (('with', 'china'), 222), (('subsource_EWErickson', 'instagram_post'), 222), (('domain_reuters.com', 'subsource_AmyResnick'), 222), (('refugee', 'domain_otherdomain'), 222), (('do', 'with'), 222), (('do', 'have'), 222), (('domain_otherdomain', 'subsource_lydiadepillis'), 222), (('subsource_othersubsource', 'subsource_pdacosta'), 222), (('domain_wsj.com', 'subsource_maggieNYT'), 222), (('chemical', 'attack'), 222), (('domain_otherdomain', 'subsource_GuyAdami'), 221), (('domain_thehill.com', 'subsource_DRUDGE_REPORT'), 221), (('repeal', 'bill'), 221), (('subsource_othersubsource', 'subsource_davemcclure'), 221), (('domain_otherdomain', 'subsource_MarkYusko'), 221), (('firm', 'domain_otherdomain'), 221), (('can', 'learn'), 221), (('thing', 'that'), 221), (('instagram_post', 'erick'), 221), (('homeland', 'security'), 221), (('domain_otherdomain', 'subsource_JamilSmith'), 220), (('pro', 'trump'), 220), (('trump', 'fire'), 220), (('domain_otherdomain', 'subsource_MugatuCapital'), 220), (('domain_otherdomain', 'subsource_lhamtil'), 220), (('trump', 'immigration_ban'), 220), (('domain_otherdomain', 'subsource_lopezlinette'), 220), (('war', 'with'), 220), (('revolution', 'domain_otherdomain'), 220), (('domain_bloomberg.com', 'subsource_WillauerProsky'), 219), (('director', 'domain_otherdomain'), 219), (('domain_huffingtonpost.com', 'subsource_OccupyWallStNYC'), 219), (('subsource_othersubsource', 'source_FT_Alphaville'), 219), (('domain_bloomberg.com', 'subsource_aussietorres'), 219), (('domain_forbes.com', 'subsource_aleks_kins'), 219), (('domain_dailymail.co.uk', 'subsource_othersubsource'), 219), (('domain_nationalreview.com', 'subsource_JonahNRO'), 219), (('twitter', 'account'), 219), (('domain_otherdomain', 'subsource_adwooldridge'), 218), (('ceo', 'say'), 218), (('domain_otherdomain', 'subsource_RPSeawright'), 218), (('source', 'say'), 218), (('domain_otherdomain', 'subsource_joseiswriting'), 218), (('south', 'africa'), 218), (('subsource_instapundit', 'subsource_instapundit'), 218), (('community', 'domain_otherdomain'), 217), (('domain_buzzfeed.com', 'subsource_JasonLeopold'), 217), (('domain_otherdomain', 'subsource_djrothkopf'), 217), (('what', 'they'), 217), (('domain_washingtonpost.com', 'subsource_daveweigel'), 217), (('domain_otherdomain', 'subsource_matthewstoller'), 217), (('pull', 'out'), 217), (('trump', 'putin'), 217), (('domain_otherdomain', 'subsource_SallieKrawcheck'), 216), (('regulation', 'domain_otherdomain'), 216), (('domain_otherdomain', 'subsource_radleybalko'), 216), (('india', 'domain_otherdomain'), 216), (('photo', 'domain_otherdomain'), 216), (('white_house', 'press'), 216), (('say', 'have'), 216), (('hit', 'record'), 216), (('domain_money.cnn.com', 'subsource_othersubsource'), 216), (('domain_otherdomain', 'subsource_adam_tooze'), 216), (('press', 'secretary'), 216), (('subsource_maggieNYT', 'trump'), 215), (('come', 'domain_otherdomain'), 215), (('domain_otherdomain', 'subsource_WallStJesus'), 215), (('survey', 'domain_otherdomain'), 215), (('domain_huffingtonpost.com', 'subsource_othersubsource'), 215), (('lawsuit', 'against'), 215), (('ann', 'reid'), 215), (('share', 'domain_otherdomain'), 215), (('syrian', 'refugee'), 215), (('domain_otherdomain', 'subsource_jilliancyork'), 215), (('domain_otherdomain', 'subsource_katiecouric'), 215), (('kick', 'off'), 215), (('joy', 'ann'), 215), (('what', 'make'), 214), (('may', 'domain_otherdomain'), 214), (('domain_youtube.com', 'subsource_palafo'), 214), (('house', 'price'), 214), (('roll', 'back'), 214), (('big', 'bank'), 214), (('domain_wsj.com', 'subsource_rationalwalk'), 214), (('domain_otherdomain', 'subsource_BaldwinRE'), 214), (('with', 'your'), 214), (('success', 'domain_otherdomain'), 213), (('come', 'out'), 213), (('domain_nytimes.com', 'subsource_DRUDGE_REPORT'), 213), (('debt', 'domain_otherdomain'), 213), (('trump', 'should'), 213), (('domain_apnews.com', 'subsource_dabeard'), 213), (('source_Abnormal_Returns', 'why'), 213), (('pdf', 'domain_nytimes.com'), 213), (('domain_otherdomain', 'subsource_FGoria'), 213), (('domain_bloomberg.com', 'subsource_AmyResnick'), 213), (('subsource_othersubsource', 'source_AVC'), 212), (('let', 'you'), 212), (('subsource_WillauerProsky', 'subsource_othersubsource'), 212), (('yet', 'domain_otherdomain'), 212), (('domain_otherdomain', 'subsource_ritholtz'), 212), (('special', 'counsel'), 212), (('appeal', 'court'), 212), (('get', 'out'), 212), (('confirmation', 'hearing'), 212), (('who', 'will'), 212), (('roger', 'ailes'), 212), (('bomb', 'threat'), 211), (('subsource_valuewalk', '2017'), 211), (('us', 'job'), 211), (('democracy', 'domain_otherdomain'), 211), (('man', 'domain_otherdomain'), 211), (('hate', 'crime'), 211), (('expert', 'say'), 211), (('domain_thehill.com', 'subsource_nycjim'), 211), (('conspiracy', 'theory'), 211), (('home', 'capital'), 211), (('claim', 'that'), 211), (('away', 'domain_otherdomain'), 211), (('domain_medium.com', 'subsource_MissTrade'), 210), (('travis', 'kalanick'), 210), (('roll', 'out'), 210), (('official', 'video'), 210), (('that', 'can'), 210), (('chemical', 'weapon'), 210), (('domain_otherdomain', 'subsource_EddyElfenbein'), 210), (('trump', 'speech'), 209), (('domain_reuters.com', 'subsource_sdonnan'), 209), (('global', 'warming'), 209), (('online', 'domain_otherdomain'), 209), (('trump', 'tax_plan'), 209), (('job', 'report'), 209), (('world', 'war'), 209), (('pick', 'up'), 209), (('rand', 'paul'), 208), (('domain_otherdomain', 'subsource_cschweitz'), 208), (('domain_techcrunch.com', 'subsource_othersubsource'), 208), (('electric', 'car'), 208), (('federal', 'judge'), 208), (('subsource_leimer', 'subsource_JimMarous'), 208), (('nikkei_asian_review', 'domain_otherdomain'), 207), (('domain_otherdomain', 'subsource_adamboultonSKY'), 207), (('dakota', 'access'), 207), (('age', 'trump'), 207), (('domain_thetimes.co.uk', 'subsource_othersubsource'), 207), (('domain_cnbc.com', 'subsource_jyarow'), 207), (('security', 'domain_otherdomain'), 207), (('domain_otherdomain', 'subsource_RichLowry'), 207), (('have', 'become'), 206), (('subsource_Livesquawk', 'subsource_AntBarton89'), 206), (('ipo', 'domain_otherdomain'), 206), (('domain_otherdomain', 'subsource_LaurenLaCapra'), 206), (('domain_otherdomain', 'subsource_McFaul'), 206), (('music', 'video'), 206), (('gop', 'health_care_bill'), 206), (('domain_foxnews.com', 'subsource_BretBaier'), 206), (('domain_imf.org', 'subsource_othersubsource'), 206), (('action', 'domain_otherdomain'), 205), (('subsource_TheEconomist', 'subsource_TheEconomist'), 205), (('domain_otherdomain', 'subsource_m_clem'), 205), (('trump', 'rally'), 205), (('national', 'park'), 205), (('fraud', 'domain_otherdomain'), 205), (('domain_otherdomain', 'subsource_yanisvaroufakis'), 204), (('domain_otherdomain', 'subsource_tomkeene'), 204), (('trump', 'ban'), 204), (('mind', 'domain_otherdomain'), 204), (('subsource_SpirosMargaris', 'artificial_intelligence'), 204), (('domain_reuters.com', 'subsource_Stalingrad_Poor'), 204), (('paris', 'agreement'), 204), (('low', 'domain_otherdomain'), 204), (('domain_washingtonpost.com', 'subsource_abuaardvark'), 204), (('domain_bloomberg.com', 'subsource_OpenOutcrier'), 203), (('tech', 'domain_otherdomain'), 203), (('shashi', 'tharoor'), 203), (('domain_otherdomain', 'subsource_MarketWatch'), 203), (('london', 'attack'), 203), (('trump', 'claim'), 203), (('unite_state', 'domain_otherdomain'), 203), (('make', 'us'), 203), (('domain_otherdomain', 'subsource_ggreenwald'), 203), (('hunt', 'domain_otherdomain'), 203), (('domain_yahoo.com', 'subsource_MylesUdland'), 202), (('will', 'take'), 202), (('trump', 'tower'), 202), (('here', 'domain_otherdomain'), 202), (('flight', 'domain_otherdomain'), 202), (('domain_otherdomain', 'subsource_behaviorgap'), 202), (('jeff', 'bezos'), 202), (('domain_latimes.com', 'subsource_DRUDGE_REPORT'), 202), (('domain_thehill.com', 'subsource_othersubsource'), 202), (('domain_otherdomain', 'subsource_amywebb'), 202), (('year', 'later'), 201), (('die', 'domain_otherdomain'), 201), (('video', 'show'), 201), (('turn', 'into'), 201), (('domain_otherdomain', 'subsource_wadhwa'), 200), (('domain_otherdomain', 'subsource_RobertMackey'), 200), (('what', 'can'), 200), (('subsource_othersubsource', 'subsource_ValaAfshar'), 200), (('susan', 'rice'), 200), (('subsource_aleks_kins', 'subsource_SpirosMargaris'), 200), (('domain_youtube.com', 'subsource_daveweigel'), 200), (('rally', 'domain_otherdomain'), 200), (('domain_www.cnn.com', 'subsource_maggieNYT'), 200), (('subsource_othersubsource', 'subsource_WillauerProsky'), 200), (('relationship', 'with'), 200), (('2017', 'domain_bloomberg.com'), 200)]

In [14]:

    

############################################################
# dump pickle file of embeddings
############################################################

mydict = {}
myreverse = {}

for i in range(vocabulary_size):
    k = reverse_dictionary[i]
    e = final_embeddings[i]
    mydict[k]=i
    myreverse[i]=k

pickle_list = [mydict, myreverse, final_embeddings]

output = open('embeddings.pkl', 'wb')
pickle.dump(pickle_list, output)
output.close()


    

In [14]:

    

############################################################
# dump files for embeddings projector
############################################################

# visualize embeddings
# put files somewhere accessible and run via 
# http://projector.tensorflow.org/
    
file1 = 'streeteye_test1.tsv'
file2 = 'streeteye_test2.tsv'

print("creating file %s" % file1)
with open(file1, 'w') as outfile:
    for i in range(vocabulary_size):
        e = final_embeddings[i]
        outfile.write("\t".join(["%f" % val for val in e]))
        outfile.write("\n")

print("creating file %s" % file2)
with open(file2, 'w') as outfile:
    #outfile.write("word\n")
    for i in range(vocabulary_size):
        w = reverse_dictionary[i]
        outfile.write(w)
        outfile.write("\n")


    

    




creating file streeteye_test1.tsv
creating file streeteye_test2.tsv

In [ ]: