diff --git a/scripts/print_table.py b/scripts/print_table.py
new file mode 100644
index 0000000..9ce40a0
--- /dev/null
+++ b/scripts/print_table.py
@@ -0,0 +1,164 @@
+import pandas as pd
+from collections import OrderedDict
+import argparse
+import numpy as np
+import re
+
+
+def get_parser():
+    parser = argparse.ArgumentParser(
+        "Prints a table from metrics files\n"
+    )
+    parser.add_argument(
+        '--config', '-c', type=str, default='config.csv',
+        help='Path to the config csv with `name` and `path` columns. '
+             '`name` is a model name, and '
+             '`path` is a path to metrics file`'
+    )
+    parser.add_argument(
+        '--extension', '-e', type=str,
+        choices=['html', 'latex', 'csv'],
+        default='csv',
+        help='Format of a table'
+    )
+    parser.add_argument(
+        '--output', '-o', type=str,
+        default='output.csv',
+        help='Path to the output table'
+    )
+    parser.add_argument(
+        '--precision', '-p', type=int,
+        default=4, help='Precision in final table'
+    )
+    return parser
+
+
+if __name__ == "__main__":
+    parser = get_parser()
+    config, unknown = parser.parse_known_args()
+    if len(unknown) != 0:
+        raise ValueError("Unknown argument " + unknown[0])
+
+    metrics = OrderedDict()
+    models = pd.read_csv(config.config)
+    for path, name in zip(models['path'], models['name']):
+        metrics[name] = pd.read_csv(path, header=None)
+        metrics[name] = {x[1][0]: x[1][1]
+                         for x in metrics[name].iterrows()}
+        metrics[name]['Model'] = name
+    metrics = pd.DataFrame(metrics).T
+    metrics = metrics.rename(columns={'valid': 'Valid',
+                                      'unique@1000': 'Unique@1k',
+                                      'unique@10000': 'Unique@10k'})
+    targets = ['Model', 'Valid', 'Unique@1k',
+               'Unique@10k', 'FCD/Test', 'FCD/TestSF',
+               'SNN/Test', 'SNN/TestSF', 'Frag/Test',
+               'Frag/TestSF', 'Scaf/Test', 'Scaf/TestSF',
+               'IntDiv', 'IntDiv2', 'Filters']
+    directions = [2, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1]
+    metrics = metrics[targets]
+
+    bf_pattern = {
+        'csv': '{}',
+        'html': '<b>{}</b>',
+        'latex': r'!bf1! {} !bf2!'
+    }[config.extension]
+    it_pattern = {
+        'csv': '{}',
+        'html': '<i>{}</i>',
+        'latex': r'!it1! {} !it2!'
+    }[config.extension]
+
+    arrow = {
+        'csv': [' (↓)', ' (↑)', ''],
+        'html': [' (↓)', ' (↑)', ''],
+        'latex': [r' ($\downarrow$)', r' ($\uparrow$)', '']
+    }[config.extension]
+
+    for col, d in zip(targets[1:], directions[1:]):
+        metrics[col] = metrics[col] \
+            .astype(float) \
+            .round(config.precision)
+        max_val = (2 * d - 1) * np.max(
+            [(2 * d - 1) * m for m, n in zip(metrics[col],
+                                             metrics['Model'])
+             if n != 'Train'])
+        metrics[col] = [str(x) if x != max_val or n == 'Train'
+                        else bf_pattern.format(x)
+                        for x, n in zip(metrics[col],
+                                        metrics['Model'])]
+    for col in targets[::-1]:
+        metrics[col] = [it_pattern.format(x)
+                        if n == 'Train' else x
+                        for x, n in zip(metrics[col],
+                                        metrics['Model'])]
+
+    metrics = metrics.round(config.precision)
+    if config.extension == 'csv':
+        metrics.to_csv(config.output, index=None)
+    elif config.extension == 'html':
+        html = metrics.to_html(index=None)
+        html = re.sub('&lt;', '<', html)
+        html = re.sub('&gt;', '>', html)
+        header, footer = html.split('</thead>')
+        header += '</thead>'
+        header = header.split('\n')
+        values = [x.strip()[4:-5]
+                  for x in header[3:-2]]
+        spans = ['rowspan' if '/' not in x else 'colspan'
+                 for x in values]
+        first_header = [x.split('/')[0] for x in values]
+        second_header = [x.split('/')[1] for x in values
+                         if '/' in x]
+        new_header = header[:3]
+        i = 0
+        total = 0
+        while i < len(first_header):
+            h = first_header[i]
+            new_header.append(
+                ' ' * 6 + '<th {}="2">{}{}</th>'.format(
+                    spans[i], h, arrow[directions[total]]
+                )
+            )
+            i += 1
+            total += 1
+            while i < len(first_header) - 1 and first_header[i] == h:
+                i += 1
+        new_header.extend(['    </tr>',
+                           '    <tr>'])
+        for h in second_header:
+            new_header.append(' ' * 6 + '<th>{}</th>'.format(h))
+        new_header.extend(header[-2:])
+        header = '\n'.join(new_header)
+        html = header + footer
+        html = ('<html>\n<head>\n<meta charset="utf-8">\n</head>\n' +
+                html +
+                '\n</html>')
+        with open(config.output, 'w', encoding='utf-8') as f:
+            f.write(html)
+    elif config.extension == 'latex':
+        latex = metrics.to_latex(index=None)
+        latex = latex.split('\n')
+        header1 = [r'\multirow{2}{*}{' + x.strip() + arrow[d] + '} '
+                   if '/' not in x
+                   else r'\multicolumn{2}{c}{' + x.split('/')[0].strip() +
+                        arrow[d] + '}'
+                   for x, d in zip(latex[2].strip()[:-2].split('&'),
+                                   directions)]
+        header1 = [x for i, x in enumerate(header1)
+                   if i == 0 or header1[i - 1] != header1[i]]
+        header2 = [x.split('/')[1].strip()
+                   if '/' in x
+                   else ''
+                   for x, d in zip(latex[2].strip()[:-2].split('&'),
+                                   directions)]
+
+        latex[2] = ' & '.join(header1) + r'\\'
+        latex.insert(3, ' & '.join(header2) + r'\\')
+        latex = '\n'.join(latex)
+        latex = re.sub(r'!bf1!', r'{\\bf', latex)
+        latex = re.sub(r'!bf2!', '}', latex)
+        latex = re.sub(r'!it1!', r'{\\it', latex)
+        latex = re.sub(r'!it2!', '}', latex)
+        with open(config.output, 'w') as f:
+            f.write(latex)