diff --git a/examples/automated_deep_compression/ADC.py b/examples/automated_deep_compression/ADC.py
index 72620e68a92d2e1626e6774be69d327378325194..8003341c4f29bb3fe42bf9d91cb32b9af2a21e13 100755
--- a/examples/automated_deep_compression/ADC.py
+++ b/examples/automated_deep_compression/ADC.py
@@ -19,6 +19,7 @@ import copy
 import logging
 import numpy as np
 import torch
+import json
 import gym
 from gym import spaces
 import distiller
@@ -27,15 +28,15 @@ from collections import OrderedDict, namedtuple
 from types import SimpleNamespace
 from distiller import normalize_module_name
 
-from base_parameters import TaskParameters
+from rl_coach.base_parameters import TaskParameters
 
 # When we import the graph_manager from the ADC_DDPG preset, we implicitly instruct
 # Coach to create and use our CNNEnvironment environment.
 # So Distiller calls Coach, which creates the environment, trains the agent, and ends.
 from examples.automated_deep_compression.presets.ADC_DDPG import graph_manager, agent_params
 # Coach imports
-from schedules import ConstantSchedule, PieceWiseSchedule, ExponentialSchedule
-from core_types import EnvironmentSteps
+from rl_coach.schedules import ConstantSchedule, PieceWiseSchedule, ExponentialSchedule
+from rl_coach.core_types import EnvironmentSteps
 
 
 msglogger = logging.getLogger()
@@ -45,6 +46,26 @@ ALMOST_ONE = 0.9999
 USE_COACH = True
 PERFORM_THINNING = True
 
+#reward = -1 * (1-top1/100) * math.log(total_macs/self.dense_model_macs)
+#
+#reward = -1 * (1-top1/100) + math.log(total_macs/self.dense_model_macs)
+#reward = 4*top1/100 - math.log(total_macs)
+#reward = reward * total_macs/213201664
+#reward = reward - 5 * total_macs/213201664
+#reward = -1 * vloss * math.sqrt(math.log(total_macs))
+#reward = top1 / math.log(total_macs)
+#alpha = 0.9
+#reward = -1 * ( (1-alpha)*(top1/100) + 10*alpha*(total_macs/self.dense_model_macs) )
+
+#alpha = 0.99
+#reward = -1 * ( (1-alpha)*(top1/100) + alpha*(total_macs/self.dense_model_macs) )
+
+#reward = vloss * math.log(total_macs)
+#reward = -1 * vloss * (total_macs / self.dense_model_macs)
+#reward = top1 * (self.dense_model_macs / total_macs)
+#reward = -1 * math.log(total_macs)
+#reward =  -1 * vloss
+
 
 def do_adc(model, dataset, arch, data_loader, validate_fn, save_checkpoint_fn):
     np.random.seed()
@@ -63,7 +84,10 @@ def coach_adc(model, dataset, arch, data_loader, validate_fn, save_checkpoint_fn
 
     # Create a dictionary of parameters that Coach will handover to CNNEnvironment
     # Once it creates it.
-    if False:
+    if True:
+        exploration_noise = 0.5
+        #exploration_noise = 0.25
+        exploitation_decay = 0.996
         graph_manager.env_params.additional_simulator_parameters = {
             'model': model,
             'dataset': dataset,
@@ -71,14 +95,21 @@ def coach_adc(model, dataset, arch, data_loader, validate_fn, save_checkpoint_fn
             'data_loader': data_loader,
             'validate_fn': validate_fn,
             'save_checkpoint_fn': save_checkpoint_fn,
-            'exploration_noise': 0.5,
-            'exploitation_decay': 0.996,
-            'action_range': (0.10, 0.95),
+            #'action_range': (0.10, 0.95),
+            'action_range': (0.70, 0.95),
             'onehot_encoding': False,
             'normalize_obs': True,
-            'desired_reduction': None
+            'desired_reduction': None,
+            'reward_fn': lambda top1, top5, vloss, total_macs: -1 * (1-top5/100) * math.log(total_macs)
+            #'reward_fn': lambda top1, total_macs: -1 * (1-top1/100) * math.log(total_macs)
+            #'reward_fn': lambda top1, total_macs: -1 * max(1-top1/100, 0.25) * math.log(total_macs)
+            #'reward_fn': lambda top1, total_macs: -1 * (1-top1/100) * math.log(total_macs/100000)
+            #'reward_fn': lambda top1, total_macs:  top1/100 * total_macs/self.dense_model_macs
         }
     else:
+        exploration_noise = 0.5
+        #exploration_noise = 0.25
+        exploitation_decay = 0.996
         graph_manager.env_params.additional_simulator_parameters = {
             'model': model,
             'dataset': dataset,
@@ -86,18 +117,19 @@ def coach_adc(model, dataset, arch, data_loader, validate_fn, save_checkpoint_fn
             'data_loader': data_loader,
             'validate_fn': validate_fn,
             'save_checkpoint_fn': save_checkpoint_fn,
-            'exploration_noise': 0.5,
-            'exploitation_decay': 0.996,
             'action_range': (0.10, 0.95),
-            'onehot_encoding': True,
+            'onehot_encoding': False,
             'normalize_obs': True,
-            'desired_reduction': 2.0e8  #  1.5e8
+            'desired_reduction': 1.5e8,
+            'reward_fn': lambda top1, total_macs: top1/100
+            #'reward_fn': lambda top1, total_macs: min(top1/100, 0.75)
         }
 
+    #msglogger.debug('Experiment configuarion:\n' + json.dumps(graph_manager.env_params.additional_simulator_parameters, indent=2))
     steps_per_episode = 13
-    agent_params.exploration.noise_percentage_schedule = PieceWiseSchedule([(ConstantSchedule(0.5),
+    agent_params.exploration.noise_percentage_schedule = PieceWiseSchedule([(ConstantSchedule(exploration_noise),
                                                                              EnvironmentSteps(100*steps_per_episode)),
-                                                                            (ExponentialSchedule(0.5, 0, 0.996),
+                                                                            (ExponentialSchedule(exploration_noise, 0, exploitation_decay),
                                                                              EnvironmentSteps(300*steps_per_episode))])
     graph_manager.create_graph(task_parameters)
     graph_manager.improve()
@@ -107,8 +139,8 @@ class CNNEnvironment(gym.Env):
     metadata = {'render.modes': ['human']}
 
     def __init__(self, model, dataset, arch, data_loader, validate_fn, save_checkpoint_fn,
-                 exploration_noise, exploitation_decay, action_range,
-                 onehot_encoding, normalize_obs, desired_reduction):
+                 action_range, onehot_encoding, normalize_obs, desired_reduction,
+                 reward_fn):
         self.pylogger = distiller.data_loggers.PythonLogger(msglogger)
         self.tflogger = distiller.data_loggers.TensorBoardLogger(msglogger.logdir)
 
@@ -121,6 +153,7 @@ class CNNEnvironment(gym.Env):
         self.onehot_encoding = onehot_encoding
         self.normalize_obs = normalize_obs
         self.max_reward = -1000
+        self.reward_fn = reward_fn
 
         self.conv_layers, self.dense_model_macs, self.dense_model_size = collect_conv_details(model, dataset)
         self.reset(init_only=True)
@@ -131,8 +164,6 @@ class CNNEnvironment(gym.Env):
         self.debug_stats = {'episode': 0}
         self.action_low = action_range[0]
         self.action_high = action_range[1]
-        self.exploitation_decay = exploitation_decay
-        self.exploration_noise = exploration_noise
         # Gym
         # spaces documentation: https://gym.openai.com/docs/
         self.action_space = spaces.Box(self.action_low, self.action_high, shape=(1,))
@@ -289,7 +320,7 @@ class CNNEnvironment(gym.Env):
                 self.max_reward = reward
                 self.save_checkpoint(is_best=True)
                 msglogger.info("Best reward={}  episode={}  top1={}".format(reward, self.debug_stats['episode'], top1))
-
+            self.save_checkpoint(is_best=False)
         else:
             observation = self._get_obs(next_layer_macs)
             if True:
@@ -428,30 +459,8 @@ class CNNEnvironment(gym.Env):
         msglogger.info("Total compute left: %.2f%%" % (total_macs/self.dense_model_macs*100))
 
         top1, top5, vloss = self.validate_fn(model=self.model, epoch=self.debug_stats['episode'])
-        #reward = -1 * (1 - top1/100)
-        if self.desired_reduction is not None:
-            reward = top1/100
-        else:
-            reward = -1 * (1-top1/100) * math.log(total_macs)
-        #reward = -1 * (1-top1/100) * math.log(total_macs/self.dense_model_macs)
-        #
-        #reward = -1 * (1-top1/100) + math.log(total_macs/self.dense_model_macs)
-        #reward = 4*top1/100 - math.log(total_macs)
-        #reward = reward * total_macs/213201664
-        #reward = reward - 5 * total_macs/213201664
-        #reward = -1 * vloss * math.sqrt(math.log(total_macs))
-        #reward = top1 / math.log(total_macs)
-        #alpha = 0.9
-        #reward = -1 * ( (1-alpha)*(top1/100) + 10*alpha*(total_macs/self.dense_model_macs) )
-
-        #alpha = 0.99
-        #reward = -1 * ( (1-alpha)*(top1/100) + alpha*(total_macs/self.dense_model_macs) )
-
-        #reward = vloss * math.log(total_macs)
-        #reward = -1 * vloss * (total_macs / self.dense_model_macs)
-        #reward = top1 * (self.dense_model_macs / total_macs)
-        #reward = -1 * math.log(total_macs)
-        #reward =  -1 * vloss
+        reward = self.reward_fn(top1, top5, vloss, total_macs)
+
         stats = ('Peformance/Validation/',
                  OrderedDict([('Loss', vloss),
                               ('Top1', top1),
@@ -509,19 +518,22 @@ def collect_conv_details(model, dataset):
             conv.name = name
             conv.id = id
             conv_layers[len(conv_layers)] = conv
-
     return conv_layers, total_macs, total_nnz
 
 
+from examples.automated_deep_compression.adc_controlled_envs import *
 def random_adc(model, dataset, arch, data_loader, validate_fn, save_checkpoint_fn):
     """Random ADC agent"""
     action_range = (0.0, 1.0)
     env = CNNEnvironment(model, dataset, arch, data_loader,
-                         validate_fn, save_checkpoint_fn, action_range)
-
-    best = [-1000, None]
-    env.action_space = RandomADCActionSpace(action_range[0], action_range[1])
-    for ep in range(100):
+                         validate_fn, save_checkpoint_fn, action_range,
+                         onehot_encoding=False, normalize_obs=False, desired_reduction=None,
+                         reward_fn=lambda top1, total_macs: top1/100)
+
+    best_episode = [-1000, None]
+    update_rate = 5
+    env.action_space = RandomADCActionSpace(action_range[0], action_range[1], std=0.35)
+    for ep in range(1000):
         observation = env.reset()
         action_config = []
         for t in range(100):
@@ -531,14 +543,17 @@ def random_adc(model, dataset, arch, data_loader, validate_fn, save_checkpoint_f
             action = env.action_space.sample()
             action_config.append(action)
             observation, reward, done, info = env.step(action)
-            if reward > best[0]:
-                best[0] = reward
-                best[1] = action_config
-                msglogger.info("^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^")
-                msglogger.info("New solution found: episode={} reward={} config={}".format(ep, reward, action_config))
             if done:
                 msglogger.info("Episode finished after {} timesteps".format(t+1))
+                msglogger.info("^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^")
+                msglogger.info("New solution found: episode={} reward={} config={}".format(ep, reward, action_config))
                 break
+        if reward > best_episode[0]:
+            best_episode[0] = reward
+            best_episode[1] = action_config
+        if ep % update_rate == 0:
+            env.action_space.set_cfg(means=best_episode[1], std=0.4)
+            best_episode = [-1000, None]
 
 
 import os
diff --git a/examples/automated_deep_compression/adc_controlled_envs.py b/examples/automated_deep_compression/adc_controlled_envs.py
index 6d74a3cd5e53e50343b1afdc1177155898c2d83a..d4e02df7014be1fbee6f2431573328332a471699 100755
--- a/examples/automated_deep_compression/adc_controlled_envs.py
+++ b/examples/automated_deep_compression/adc_controlled_envs.py
@@ -1,15 +1,36 @@
 """This file contains a couple of environments used for debugging ADC reproduction.
 """
 import random
+import numpy as np
+from scipy.stats import truncnorm
 
 
 class RandomADCActionSpace(object):
-    def __init__(self, low, high):
-        self.low = low
-        self.high = high
+    def __init__(self, low, high, std):
+        self.clip_low = low
+        self.clip_high = high
+        self.layer = 0
+        self.num_layers = 13
+        #self.means = [high-low] * self.num_layers
+        self.means = [0.9, 0.9, 0.9, 0.9, 0.9, 0.8, 0.8, 0.7, 0.7, 0.6, 0.6, 0.5, 0.5]
+        self.std = std
 
     def sample(self):
-        return random.uniform(self.low, self.high)
+        return random.uniform(self.clip_low, self.clip_high)
+        action_values_mean = self.means[self.layer]
+        action_values_std = self.std
+        normalized_low = (self.clip_low - action_values_mean) / action_values_std
+        normalized_high = (self.clip_high - action_values_mean) / action_values_std
+        distribution = truncnorm(normalized_low, normalized_high, loc=action_values_mean, scale=action_values_std)
+        action = distribution.rvs(1)
+        # action = np.random.normal(self.means[self.layer], self.std)
+        # action = min(self.clip_high, max(action, self.clip_low))
+        self.layer = (self.layer + 1) % self.num_layers
+        return action
+
+    def set_cfg(self, means, std):
+        self.means = [0.01*m for m in self.means] + [0.99*m for m in means]
+        self.std = std
 
 
 class PredictableADCActionSpace(object):
diff --git a/examples/automated_deep_compression/comparing_model_spaces.ipynb b/examples/automated_deep_compression/comparing_model_spaces.ipynb
index 2c223f24cd1792087adbc54ded14b4064b2eac07..42b20efacd9483d0d7d90d172eca306b97c3396e 100644
--- a/examples/automated_deep_compression/comparing_model_spaces.ipynb
+++ b/examples/automated_deep_compression/comparing_model_spaces.ipynb
@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "code",
-   "execution_count": 60,
+   "execution_count": 103,
    "metadata": {},
    "outputs": [
     {
@@ -164,7 +164,7 @@
        "11  90.87  "
       ]
      },
-     "execution_count": 60,
+     "execution_count": 103,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -174,14 +174,14 @@
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
     "import seaborn as sns\n",
-    "\n",
+    "#master___2018.07.24-2*\n",
     "df = pd.read_csv(\"../classifier_compression/logs/master___2018.07.24-235532/arch_space.csv\")\n",
     "df"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 70,
+   "execution_count": 72,
    "metadata": {},
    "outputs": [
     {
@@ -217,270 +217,1045 @@
        "      <th>0</th>\n",
        "      <td>0</td>\n",
        "      <td>BEST_adc_episode_003_checkpoint.pth.tar</td>\n",
-       "      <td>4770986</td>\n",
-       "      <td>86797580</td>\n",
-       "      <td>11.60</td>\n",
+       "      <td>1880678</td>\n",
+       "      <td>40981916</td>\n",
+       "      <td>10.00</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
        "      <td>1</td>\n",
        "      <td>BEST_adc_episode_001_checkpoint.pth.tar</td>\n",
-       "      <td>4804162</td>\n",
-       "      <td>90519442</td>\n",
-       "      <td>10.07</td>\n",
+       "      <td>3243400</td>\n",
+       "      <td>75744988</td>\n",
+       "      <td>10.01</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
        "      <td>2</td>\n",
-       "      <td>BEST_adc_episode_142_checkpoint.pth.tar</td>\n",
-       "      <td>13038094</td>\n",
-       "      <td>279552826</td>\n",
-       "      <td>90.77</td>\n",
+       "      <td>BEST_adc_episode_006_checkpoint.pth.tar</td>\n",
+       "      <td>4893669</td>\n",
+       "      <td>130690152</td>\n",
+       "      <td>26.48</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
        "      <td>3</td>\n",
-       "      <td>BEST_adc_episode_113_checkpoint.pth.tar</td>\n",
-       "      <td>9877163</td>\n",
-       "      <td>234560606</td>\n",
-       "      <td>90.48</td>\n",
+       "      <td>BEST_adc_episode_114_checkpoint.pth.tar</td>\n",
+       "      <td>8249014</td>\n",
+       "      <td>161158222</td>\n",
+       "      <td>78.21</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
        "      <td>4</td>\n",
-       "      <td>BEST_adc_episode_016_checkpoint.pth.tar</td>\n",
-       "      <td>8534957</td>\n",
-       "      <td>188821328</td>\n",
-       "      <td>84.02</td>\n",
+       "      <td>BEST_adc_episode_008_checkpoint.pth.tar</td>\n",
+       "      <td>5780922</td>\n",
+       "      <td>158955054</td>\n",
+       "      <td>71.42</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>5</th>\n",
        "      <td>5</td>\n",
-       "      <td>BEST_adc_episode_002_checkpoint.pth.tar</td>\n",
-       "      <td>1643593</td>\n",
-       "      <td>83911054</td>\n",
+       "      <td>BEST_adc_episode_108_checkpoint.pth.tar</td>\n",
+       "      <td>7041071</td>\n",
+       "      <td>176417822</td>\n",
+       "      <td>80.34</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>6</th>\n",
+       "      <td>6</td>\n",
+       "      <td>BEST_adc_episode_044_checkpoint.pth.tar</td>\n",
+       "      <td>9410752</td>\n",
+       "      <td>227796580</td>\n",
+       "      <td>86.64</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>7</th>\n",
+       "      <td>7</td>\n",
+       "      <td>BEST_adc_episode_077_checkpoint.pth.tar</td>\n",
+       "      <td>6231760</td>\n",
+       "      <td>178635022</td>\n",
+       "      <td>81.33</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "   Unnamed: 0                                     File      NNZ       MACs  \\\n",
+       "0           0  BEST_adc_episode_003_checkpoint.pth.tar  1880678   40981916   \n",
+       "1           1  BEST_adc_episode_001_checkpoint.pth.tar  3243400   75744988   \n",
+       "2           2  BEST_adc_episode_006_checkpoint.pth.tar  4893669  130690152   \n",
+       "3           3  BEST_adc_episode_114_checkpoint.pth.tar  8249014  161158222   \n",
+       "4           4  BEST_adc_episode_008_checkpoint.pth.tar  5780922  158955054   \n",
+       "5           5  BEST_adc_episode_108_checkpoint.pth.tar  7041071  176417822   \n",
+       "6           6  BEST_adc_episode_044_checkpoint.pth.tar  9410752  227796580   \n",
+       "7           7  BEST_adc_episode_077_checkpoint.pth.tar  6231760  178635022   \n",
+       "\n",
+       "    Top1  \n",
+       "0  10.00  \n",
+       "1  10.01  \n",
+       "2  26.48  \n",
+       "3  78.21  \n",
+       "4  71.42  \n",
+       "5  80.34  \n",
+       "6  86.64  \n",
+       "7  81.33  "
+      ]
+     },
+     "execution_count": 72,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df2 = pd.read_csv(\"../classifier_compression/logs/master___2018.07.24-232342/arch_space.csv\")\n",
+    "df2\n",
+    "df3 = pd.read_csv(\"../classifier_compression/logs/master___2018.07.24-225916/arch_space.csv\")\n",
+    "df3\n",
+    "df4 = pd.read_csv(\"../classifier_compression/logs/master___2018.07.25-205658/arch_space.csv\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 76,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<matplotlib.figure.Figure at 0x7ff847d9e160>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "def add_results(df, plt, cmap):\n",
+    "    # create data\n",
+    "    x = df['MACs'].tolist()\n",
+    "    y = df['Top1'].tolist()\n",
+    "    z = df['NNZ'].tolist()\n",
+    "    z = [n/30000 for n in z]\n",
+    "    plt.scatter(x, y, s=z, c=x, cmap=cmap, alpha=0.4, edgecolors=\"black\", linewidth=2)\n",
+    "\n",
+    "# Change color with c and alpha. I map the color to the X axis value.\n",
+    "plt.figure(figsize=(20,10))\n",
+    "add_results(df, plt, cmap=\"Blues\")\n",
+    "add_results(df2, plt, cmap=\"Reds\")\n",
+    "add_results(df3, plt, cmap=\"Greens\")\n",
+    "\n",
+    "# Add titles (main and on axis)\n",
+    "plt.xlabel(\"Compute (MACs)\")\n",
+    "plt.ylabel(\"Accuracy (Top1)\")\n",
+    "plt.title(\"Network Space\")\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 77,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<matplotlib.figure.Figure at 0x7ff8479e5128>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "# Change color with c and alpha. I map the color to the X axis value.\n",
+    "plt.figure(figsize=(20,10))\n",
+    "\n",
+    "add_results(df4, plt, cmap=\"Oranges\")\n",
+    "\n",
+    "# Add titles (main and on axis)\n",
+    "plt.xlabel(\"Compute (MACs)\")\n",
+    "plt.ylabel(\"Accuracy (Top1)\")\n",
+    "plt.title(\"Network Space\")\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 78,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style>\n",
+       "    .dataframe thead tr:only-child th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: left;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>Unnamed: 0</th>\n",
+       "      <th>File</th>\n",
+       "      <th>NNZ</th>\n",
+       "      <th>MACs</th>\n",
+       "      <th>Top1</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>0</td>\n",
+       "      <td>BEST_adc_episode_003_checkpoint.pth.tar</td>\n",
+       "      <td>1880678</td>\n",
+       "      <td>40981916</td>\n",
        "      <td>10.00</td>\n",
        "    </tr>\n",
        "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>1</td>\n",
+       "      <td>BEST_adc_episode_001_checkpoint.pth.tar</td>\n",
+       "      <td>3243400</td>\n",
+       "      <td>75744988</td>\n",
+       "      <td>10.01</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>2</td>\n",
+       "      <td>BEST_adc_episode_006_checkpoint.pth.tar</td>\n",
+       "      <td>4893669</td>\n",
+       "      <td>130690152</td>\n",
+       "      <td>26.48</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>3</td>\n",
+       "      <td>BEST_adc_episode_114_checkpoint.pth.tar</td>\n",
+       "      <td>8249014</td>\n",
+       "      <td>161158222</td>\n",
+       "      <td>78.21</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>4</td>\n",
+       "      <td>BEST_adc_episode_008_checkpoint.pth.tar</td>\n",
+       "      <td>5780922</td>\n",
+       "      <td>158955054</td>\n",
+       "      <td>71.42</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>5</th>\n",
+       "      <td>5</td>\n",
+       "      <td>BEST_adc_episode_108_checkpoint.pth.tar</td>\n",
+       "      <td>7041071</td>\n",
+       "      <td>176417822</td>\n",
+       "      <td>80.34</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
        "      <th>6</th>\n",
        "      <td>6</td>\n",
-       "      <td>BEST_adc_episode_125_checkpoint.pth.tar</td>\n",
-       "      <td>12157454</td>\n",
-       "      <td>265993040</td>\n",
-       "      <td>90.58</td>\n",
+       "      <td>BEST_adc_episode_044_checkpoint.pth.tar</td>\n",
+       "      <td>9410752</td>\n",
+       "      <td>227796580</td>\n",
+       "      <td>86.64</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>7</th>\n",
        "      <td>7</td>\n",
-       "      <td>BEST_adc_episode_130_checkpoint.pth.tar</td>\n",
-       "      <td>12513507</td>\n",
-       "      <td>265619958</td>\n",
-       "      <td>90.59</td>\n",
+       "      <td>BEST_adc_episode_077_checkpoint.pth.tar</td>\n",
+       "      <td>6231760</td>\n",
+       "      <td>178635022</td>\n",
+       "      <td>81.33</td>\n",
        "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
+      "text/plain": [
+       "   Unnamed: 0                                     File      NNZ       MACs  \\\n",
+       "0           0  BEST_adc_episode_003_checkpoint.pth.tar  1880678   40981916   \n",
+       "1           1  BEST_adc_episode_001_checkpoint.pth.tar  3243400   75744988   \n",
+       "2           2  BEST_adc_episode_006_checkpoint.pth.tar  4893669  130690152   \n",
+       "3           3  BEST_adc_episode_114_checkpoint.pth.tar  8249014  161158222   \n",
+       "4           4  BEST_adc_episode_008_checkpoint.pth.tar  5780922  158955054   \n",
+       "5           5  BEST_adc_episode_108_checkpoint.pth.tar  7041071  176417822   \n",
+       "6           6  BEST_adc_episode_044_checkpoint.pth.tar  9410752  227796580   \n",
+       "7           7  BEST_adc_episode_077_checkpoint.pth.tar  6231760  178635022   \n",
+       "\n",
+       "    Top1  \n",
+       "0  10.00  \n",
+       "1  10.01  \n",
+       "2  26.48  \n",
+       "3  78.21  \n",
+       "4  71.42  \n",
+       "5  80.34  \n",
+       "6  86.64  \n",
+       "7  81.33  "
+      ]
+     },
+     "execution_count": 78,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "df4"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 92,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Relative import of code from distiller, w/o installing the package\n",
+    "import os\n",
+    "import sys\n",
+    "import torch\n",
+    "module_path = os.path.abspath(os.path.join('..', '..'))\n",
+    "if module_path not in sys.path:\n",
+    "    sys.path.append(module_path)\n",
+    "import distiller\n",
+    "import models\n",
+    "import apputils"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 101,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dataset = 'cifar10'\n",
+    "dummy_input = torch.randn(1, 3, 32, 32)\n",
+    "arch = 'vgg16_cifar'\n",
+    "checkpoint_file = \"../classifier_compression/logs/master___2018.07.25-205658/BEST_adc_episode_044_checkpoint.pth.tar\" \n",
+    "\n",
+    "model = models.create_model(pretrained=False, dataset=dataset, arch=arch)\n",
+    "apputils.load_checkpoint(model, checkpoint_file);"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 99,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "t = distiller.weights_sparsity_summary(model)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 100,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<div>\n",
+       "<style>\n",
+       "    .dataframe thead tr:only-child th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: left;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>Name</th>\n",
+       "      <th>Shape</th>\n",
+       "      <th>NNZ (dense)</th>\n",
+       "      <th>NNZ (sparse)</th>\n",
+       "      <th>Cols (%)</th>\n",
+       "      <th>Rows (%)</th>\n",
+       "      <th>Ch (%)</th>\n",
+       "      <th>2D (%)</th>\n",
+       "      <th>3D (%)</th>\n",
+       "      <th>Fine (%)</th>\n",
+       "      <th>Std</th>\n",
+       "      <th>Mean</th>\n",
+       "      <th>Abs-Mean</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
        "    <tr>\n",
-       "      <th>8</th>\n",
-       "      <td>8</td>\n",
-       "      <td>BEST_adc_episode_107_checkpoint.pth.tar</td>\n",
-       "      <td>9963371</td>\n",
-       "      <td>233031242</td>\n",
-       "      <td>89.59</td>\n",
+       "      <th>0</th>\n",
+       "      <td>features.module.0.weight</td>\n",
+       "      <td>(61, 3, 3, 3)</td>\n",
+       "      <td>1647</td>\n",
+       "      <td>1647</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.26</td>\n",
+       "      <td>-2.40e-03</td>\n",
+       "      <td>1.90e-01</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>9</th>\n",
-       "      <td>9</td>\n",
-       "      <td>BEST_adc_episode_047_checkpoint.pth.tar</td>\n",
-       "      <td>10195618</td>\n",
-       "      <td>238916830</td>\n",
-       "      <td>89.24</td>\n",
+       "      <th>1</th>\n",
+       "      <td>features.module.2.weight</td>\n",
+       "      <td>(61, 61, 3, 3)</td>\n",
+       "      <td>33489</td>\n",
+       "      <td>33489</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.08</td>\n",
+       "      <td>-1.19e-02</td>\n",
+       "      <td>5.79e-02</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>10</th>\n",
-       "      <td>10</td>\n",
-       "      <td>BEST_adc_episode_025_checkpoint.pth.tar</td>\n",
-       "      <td>9163282</td>\n",
-       "      <td>208964092</td>\n",
-       "      <td>88.96</td>\n",
+       "      <th>2</th>\n",
+       "      <td>features.module.5.weight</td>\n",
+       "      <td>(116, 61, 3, 3)</td>\n",
+       "      <td>63684</td>\n",
+       "      <td>63684</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.07</td>\n",
+       "      <td>-8.42e-03</td>\n",
+       "      <td>5.25e-02</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>11</th>\n",
-       "      <td>11</td>\n",
-       "      <td>BEST_adc_episode_006_checkpoint.pth.tar</td>\n",
-       "      <td>7850262</td>\n",
-       "      <td>180695568</td>\n",
-       "      <td>81.23</td>\n",
+       "      <th>3</th>\n",
+       "      <td>features.module.7.weight</td>\n",
+       "      <td>(112, 116, 3, 3)</td>\n",
+       "      <td>116928</td>\n",
+       "      <td>116928</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.06</td>\n",
+       "      <td>-1.06e-02</td>\n",
+       "      <td>4.48e-02</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>12</th>\n",
-       "      <td>12</td>\n",
-       "      <td>BEST_adc_episode_132_checkpoint.pth.tar</td>\n",
-       "      <td>12676828</td>\n",
-       "      <td>264931726</td>\n",
-       "      <td>90.73</td>\n",
+       "      <th>4</th>\n",
+       "      <td>features.module.10.weight</td>\n",
+       "      <td>(184, 112, 3, 3)</td>\n",
+       "      <td>185472</td>\n",
+       "      <td>185472</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.05</td>\n",
+       "      <td>-6.52e-03</td>\n",
+       "      <td>3.98e-02</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>13</th>\n",
-       "      <td>13</td>\n",
-       "      <td>BEST_adc_episode_004_checkpoint.pth.tar</td>\n",
-       "      <td>8719624</td>\n",
-       "      <td>158544946</td>\n",
-       "      <td>41.62</td>\n",
+       "      <th>5</th>\n",
+       "      <td>features.module.12.weight</td>\n",
+       "      <td>(177, 184, 3, 3)</td>\n",
+       "      <td>293112</td>\n",
+       "      <td>293112</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.04</td>\n",
+       "      <td>-5.51e-03</td>\n",
+       "      <td>3.21e-02</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>14</th>\n",
-       "      <td>14</td>\n",
-       "      <td>BEST_adc_episode_024_checkpoint.pth.tar</td>\n",
-       "      <td>9783647</td>\n",
-       "      <td>217139246</td>\n",
-       "      <td>88.01</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>15</th>\n",
-       "      <td>15</td>\n",
-       "      <td>BEST_adc_episode_149_checkpoint.pth.tar</td>\n",
-       "      <td>13085489</td>\n",
-       "      <td>280045100</td>\n",
-       "      <td>90.88</td>\n",
-       "    </tr>\n",
-       "    <tr>\n",
-       "      <th>16</th>\n",
-       "      <td>16</td>\n",
-       "      <td>BEST_adc_episode_112_checkpoint.pth.tar</td>\n",
-       "      <td>10391004</td>\n",
-       "      <td>241402680</td>\n",
-       "      <td>90.10</td>\n",
+       "      <th>6</th>\n",
+       "      <td>features.module.14.weight</td>\n",
+       "      <td>(209, 177, 3, 3)</td>\n",
+       "      <td>332937</td>\n",
+       "      <td>332937</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.04</td>\n",
+       "      <td>-6.98e-03</td>\n",
+       "      <td>2.94e-02</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>17</th>\n",
-       "      <td>17</td>\n",
-       "      <td>BEST_adc_episode_005_checkpoint.pth.tar</td>\n",
-       "      <td>9241390</td>\n",
-       "      <td>153550990</td>\n",
-       "      <td>51.40</td>\n",
+       "      <th>7</th>\n",
+       "      <td>features.module.17.weight</td>\n",
+       "      <td>(487, 209, 3, 3)</td>\n",
+       "      <td>916047</td>\n",
+       "      <td>916047</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.02</td>\n",
+       "      <td>-2.65e-03</td>\n",
+       "      <td>1.72e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>8</th>\n",
+       "      <td>features.module.19.weight</td>\n",
+       "      <td>(487, 487, 3, 3)</td>\n",
+       "      <td>2134521</td>\n",
+       "      <td>2134521</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>-1.06e-03</td>\n",
+       "      <td>1.09e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>9</th>\n",
+       "      <td>features.module.21.weight</td>\n",
+       "      <td>(447, 487, 3, 3)</td>\n",
+       "      <td>1959201</td>\n",
+       "      <td>1959201</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>-9.06e-04</td>\n",
+       "      <td>9.68e-03</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>10</th>\n",
+       "      <td>features.module.24.weight</td>\n",
+       "      <td>(298, 447, 3, 3)</td>\n",
+       "      <td>1198854</td>\n",
+       "      <td>1198854</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>-7.74e-05</td>\n",
+       "      <td>8.77e-03</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11</th>\n",
+       "      <td>features.module.26.weight</td>\n",
+       "      <td>(431, 298, 3, 3)</td>\n",
+       "      <td>1155942</td>\n",
+       "      <td>1155942</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>2.33e-05</td>\n",
+       "      <td>8.45e-03</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>18</th>\n",
-       "      <td>18</td>\n",
-       "      <td>BEST_adc_episode_140_checkpoint.pth.tar</td>\n",
-       "      <td>13104277</td>\n",
-       "      <td>279860848</td>\n",
-       "      <td>90.76</td>\n",
+       "      <th>12</th>\n",
+       "      <td>features.module.28.weight</td>\n",
+       "      <td>(262, 431, 3, 3)</td>\n",
+       "      <td>1016298</td>\n",
+       "      <td>1016298</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>-2.21e-04</td>\n",
+       "      <td>8.53e-03</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>19</th>\n",
-       "      <td>19</td>\n",
-       "      <td>BEST_adc_episode_143_checkpoint.pth.tar</td>\n",
-       "      <td>13071701</td>\n",
-       "      <td>280396028</td>\n",
-       "      <td>90.85</td>\n",
+       "      <th>13</th>\n",
+       "      <td>classifier.weight</td>\n",
+       "      <td>(10, 262)</td>\n",
+       "      <td>2620</td>\n",
+       "      <td>2620</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.09</td>\n",
+       "      <td>-4.34e-07</td>\n",
+       "      <td>6.56e-02</td>\n",
        "    </tr>\n",
        "    <tr>\n",
-       "      <th>20</th>\n",
-       "      <td>20</td>\n",
-       "      <td>BEST_adc_episode_017_checkpoint.pth.tar</td>\n",
-       "      <td>9965772</td>\n",
-       "      <td>207411696</td>\n",
-       "      <td>85.35</td>\n",
+       "      <th>14</th>\n",
+       "      <td>Total sparsity:</td>\n",
+       "      <td>-</td>\n",
+       "      <td>9410752</td>\n",
+       "      <td>9410752</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.00</td>\n",
+       "      <td>0.00e+00</td>\n",
+       "      <td>0.00e+00</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "</div>"
       ],
       "text/plain": [
-       "    Unnamed: 0                                     File       NNZ       MACs  \\\n",
-       "0            0  BEST_adc_episode_003_checkpoint.pth.tar   4770986   86797580   \n",
-       "1            1  BEST_adc_episode_001_checkpoint.pth.tar   4804162   90519442   \n",
-       "2            2  BEST_adc_episode_142_checkpoint.pth.tar  13038094  279552826   \n",
-       "3            3  BEST_adc_episode_113_checkpoint.pth.tar   9877163  234560606   \n",
-       "4            4  BEST_adc_episode_016_checkpoint.pth.tar   8534957  188821328   \n",
-       "5            5  BEST_adc_episode_002_checkpoint.pth.tar   1643593   83911054   \n",
-       "6            6  BEST_adc_episode_125_checkpoint.pth.tar  12157454  265993040   \n",
-       "7            7  BEST_adc_episode_130_checkpoint.pth.tar  12513507  265619958   \n",
-       "8            8  BEST_adc_episode_107_checkpoint.pth.tar   9963371  233031242   \n",
-       "9            9  BEST_adc_episode_047_checkpoint.pth.tar  10195618  238916830   \n",
-       "10          10  BEST_adc_episode_025_checkpoint.pth.tar   9163282  208964092   \n",
-       "11          11  BEST_adc_episode_006_checkpoint.pth.tar   7850262  180695568   \n",
-       "12          12  BEST_adc_episode_132_checkpoint.pth.tar  12676828  264931726   \n",
-       "13          13  BEST_adc_episode_004_checkpoint.pth.tar   8719624  158544946   \n",
-       "14          14  BEST_adc_episode_024_checkpoint.pth.tar   9783647  217139246   \n",
-       "15          15  BEST_adc_episode_149_checkpoint.pth.tar  13085489  280045100   \n",
-       "16          16  BEST_adc_episode_112_checkpoint.pth.tar  10391004  241402680   \n",
-       "17          17  BEST_adc_episode_005_checkpoint.pth.tar   9241390  153550990   \n",
-       "18          18  BEST_adc_episode_140_checkpoint.pth.tar  13104277  279860848   \n",
-       "19          19  BEST_adc_episode_143_checkpoint.pth.tar  13071701  280396028   \n",
-       "20          20  BEST_adc_episode_017_checkpoint.pth.tar   9965772  207411696   \n",
+       "                         Name             Shape NNZ (dense) NNZ (sparse)  \\\n",
+       "0    features.module.0.weight     (61, 3, 3, 3)        1647         1647   \n",
+       "1    features.module.2.weight    (61, 61, 3, 3)       33489        33489   \n",
+       "2    features.module.5.weight   (116, 61, 3, 3)       63684        63684   \n",
+       "3    features.module.7.weight  (112, 116, 3, 3)      116928       116928   \n",
+       "4   features.module.10.weight  (184, 112, 3, 3)      185472       185472   \n",
+       "5   features.module.12.weight  (177, 184, 3, 3)      293112       293112   \n",
+       "6   features.module.14.weight  (209, 177, 3, 3)      332937       332937   \n",
+       "7   features.module.17.weight  (487, 209, 3, 3)      916047       916047   \n",
+       "8   features.module.19.weight  (487, 487, 3, 3)     2134521      2134521   \n",
+       "9   features.module.21.weight  (447, 487, 3, 3)     1959201      1959201   \n",
+       "10  features.module.24.weight  (298, 447, 3, 3)     1198854      1198854   \n",
+       "11  features.module.26.weight  (431, 298, 3, 3)     1155942      1155942   \n",
+       "12  features.module.28.weight  (262, 431, 3, 3)     1016298      1016298   \n",
+       "13          classifier.weight         (10, 262)        2620         2620   \n",
+       "14            Total sparsity:                 -     9410752      9410752   \n",
        "\n",
-       "     Top1  \n",
-       "0   11.60  \n",
-       "1   10.07  \n",
-       "2   90.77  \n",
-       "3   90.48  \n",
-       "4   84.02  \n",
-       "5   10.00  \n",
-       "6   90.58  \n",
-       "7   90.59  \n",
-       "8   89.59  \n",
-       "9   89.24  \n",
-       "10  88.96  \n",
-       "11  81.23  \n",
-       "12  90.73  \n",
-       "13  41.62  \n",
-       "14  88.01  \n",
-       "15  90.88  \n",
-       "16  90.10  \n",
-       "17  51.40  \n",
-       "18  90.76  \n",
-       "19  90.85  \n",
-       "20  85.35  "
+       "   Cols (%) Rows (%)  Ch (%)  2D (%)  3D (%)  Fine (%)   Std      Mean  \\\n",
+       "0         0        0     0.0     0.0     0.0       0.0  0.26 -2.40e-03   \n",
+       "1         0        0     0.0     0.0     0.0       0.0  0.08 -1.19e-02   \n",
+       "2         0        0     0.0     0.0     0.0       0.0  0.07 -8.42e-03   \n",
+       "3         0        0     0.0     0.0     0.0       0.0  0.06 -1.06e-02   \n",
+       "4         0        0     0.0     0.0     0.0       0.0  0.05 -6.52e-03   \n",
+       "5         0        0     0.0     0.0     0.0       0.0  0.04 -5.51e-03   \n",
+       "6         0        0     0.0     0.0     0.0       0.0  0.04 -6.98e-03   \n",
+       "7         0        0     0.0     0.0     0.0       0.0  0.02 -2.65e-03   \n",
+       "8         0        0     0.0     0.0     0.0       0.0  0.01 -1.06e-03   \n",
+       "9         0        0     0.0     0.0     0.0       0.0  0.01 -9.06e-04   \n",
+       "10        0        0     0.0     0.0     0.0       0.0  0.01 -7.74e-05   \n",
+       "11        0        0     0.0     0.0     0.0       0.0  0.01  2.33e-05   \n",
+       "12        0        0     0.0     0.0     0.0       0.0  0.01 -2.21e-04   \n",
+       "13        0        0     0.0     0.0     0.0       0.0  0.09 -4.34e-07   \n",
+       "14        0        0     0.0     0.0     0.0       0.0  0.00  0.00e+00   \n",
+       "\n",
+       "    Abs-Mean  \n",
+       "0   1.90e-01  \n",
+       "1   5.79e-02  \n",
+       "2   5.25e-02  \n",
+       "3   4.48e-02  \n",
+       "4   3.98e-02  \n",
+       "5   3.21e-02  \n",
+       "6   2.94e-02  \n",
+       "7   1.72e-02  \n",
+       "8   1.09e-02  \n",
+       "9   9.68e-03  \n",
+       "10  8.77e-03  \n",
+       "11  8.45e-03  \n",
+       "12  8.53e-03  \n",
+       "13  6.56e-02  \n",
+       "14  0.00e+00  "
       ]
      },
-     "execution_count": 70,
+     "execution_count": 100,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "df2 = pd.read_csv(\"../classifier_compression/logs/master___2018.07.24-232342/arch_space.csv\")\n",
-    "df2\n",
-    "df3 = pd.read_csv(\"../classifier_compression/logs/master___2018.07.24-225916/arch_space.csv\")\n",
-    "df3"
+    "t"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 71,
+   "execution_count": 102,
    "metadata": {},
    "outputs": [
     {
      "data": {
-      "image/png": "iVBORw0KGgoAAAANSUhEUgAABI8AAAJcCAYAAABwj4S5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAIABJREFUeJzs3VmQXNed5/fvuUve3DNr34ECUCgABYICSZAUuIiSKKlJtUQp3B67PRPj6A63x+EHtx0xS/jJbTtmHuyxPRP2hB/smHF7xp6JdrdtuSU1RVFUixSphSRIQASqABRq37Mq9/XmXY4fsoiFqCIAihRF8v+JQAQq8557zl3IYP14zv8orTVCCCGEEEIIIYQQQuzF+LgHIIQQQgghhBBCCCF+e0l4JIQQQgghhBBCCCH2JeGREEIIIYQQQgghhNiXhEdCCCGEEEIIIYQQYl8SHgkhhBBCCCGEEEKIfUl4JIQQQgghhBBCCCH2JeGREEIIIcSHRCn1E6XUH33c4xBCCCGE+DBJeCSEEEKI32pKqUWlVE4plbjpsz9SSv3kLtv/qVLqH35kA/yAlFJPKKV+ppQqK6UKSqnXlFIPf9zjEkIIIYR4LwmPhBBCCPFJYAL/6cc9iP2ojrv+7yqlVBr4HvA/Ad3ACPBfAe5HM0IhhBBCiA9OwiMhhBBCfBL8Y+DvKaWye32plDqulHpxdwbPFaXUv7P7+d8B/hbwD5RSNaXUd5VSf6iU+u5NbWeVUn9+088rSqnTu39/TCn1xu7soDeUUo/ddNxPlFL/SCn1GtAADr9nTENKqV8ppf7+HkOeBNBa/xutdaC1bmqtf6i1/tVu2z/YnYn0z3b7vqyUevqmc/+hUmpGKVVVSs0rpf6j9/T9LaXUeaVURSk1p5R6ZvfzjFLqnyulNpRSa0qpf6iUMu/mAQghhBDis0vCIyGEEEJ8ErwJ/AT4e+/9Ync524vAvwb6gd8H/mel1JTW+n8B/k/gv9VaJ7XW3wReBp5UShlKqWEgApzdPddhIAn8SinVDXwf+B+BHuB/AL6vlOq5qfu/DfwdIAUs3TSmQ7v9/DOt9T/e43quAoFS6n9XSj2rlOra45hHgTmgF/gT4P/ZHRNADvgGkAb+EPgnSqkHd/t+BPiXwN8HssAXgMXddn8K+MAE8ADwNUBqNAkhhBDifUl4JIQQQohPiv8C+E+UUn3v+fwbwKLW+n/TWvta67eB/xv4G3udRGs9D1SB03SClReAdaXUceAp4Kda6xD4XWBWa/2vds/7b4DLwDdvOt2faq0v7X7v7X42Bfw18Ce74dVeY6gATwAa+F+BbaXUXyqlBm46LAf8U621p7X+M+DK7pjQWn9faz2nO14Gfgg8udvuPwD+hdb6Ra11qLVe01pf3j3314H/TGtd11rngH9CJ2wTQgghhNiX9XEPQAghhBDibmitLyqlvgf858DMTV8dBB5VSpVu+swC/tX7nO5l4It0ZuC8DJToBEdnd38GGOam2US7lujUJ3rXyh7n/lvANeAv3qd/tNYzwB9AZ9kd8H8A/xT493YPWdNa6/f0Pbx7/LN0ZiNN0vmfgXHgnd3jxoC/2qPLg4ANbCil3v3M2OcahBBCCCGuk5lHQgghhPgk+RPgP+T2AOdlrXX2pj9JrfV/vPu9vu0sN8KjJ3f//jKd8OgpboRH63QCl5sdANZu+nmvc/+XwA7wr++2npDW+jKdJWX33fTxiLop5dnte10p5dCZWfXfAQNa6yydsOjdY1eAI3t0s0KnIHfvTfcprbU+eTdjFEIIIcRnl4RHQgghhPjE0FpfA/4M+OObPv4eMKmU+ttKKXv3z8NKqRO732/xnmLWdAKiLwExrfUq8FPgGTq1jd7ePeavds/7N5VSllLq36WzJO17dximR2fJXAL4l3vtwrZb4PvvKqVGd38eozPj6Bc3HdYP/PHu9fwN4MTumCKAA2wD/u4spK/d1O6fA3+olHp6t67TiFLquNZ6g87ytv9eKZXe/e6IUuqpO1yPEEIIIT7jJDwSQgghxCfNf00nmAFAa12lE578Pp3ZQpvAf0MnYIFOmDKllCoppb6z2+YqUKMTGr1bg2geeE1rHex+lqdTT+nvAnngHwDf0Frv3GmAWus28G8BA8C/2CNAqtIpiP1LpVSdTmh0cbevd/0SOEpnFtM/Av5trXV+93r/GPi/gCLwN4G/vKnv19ktog2U6QRl786g+vfphE/Tu23/Ahi60/UIIYQQ4rNN3bqUXgghhBBCfNyUUn8A/JHW+omPeyxCCCGEEDLzSAghhBBCCCGEEELsS8IjIYQQQgghhBBCCLEvWbYmhBBCCCGEEEIIIfYlM4+EEEIIIYQQQgghxL6sj3sAd6O3t1ePj49/3MMQQgghhBBCCCGE+NQ4d+7cjta6707HfSLCo/Hxcd58882PexhCCCGEEEIIIYQQnxpKqaW7OU6WrQkhhBBCCCGEEEKIfUl4JIQQQgghhBBCCCH2JeGREEIIIYQQQgghhNiXhEdCCCGEEEIIIYQQYl8SHgkhhBBCCCGEEEKIfUl4JIQQQgghhBBCCCH2JeGREEIIIYQQQgghhNiXhEdCCCGEEEIIIYQQYl8SHgkhhBBCCCGEEEKIfUl4JIQQQgghhBBCCCH2JeGREEIIIYQQQgghhNiXhEdCCCGEEEIIIYQQYl8SHgkhhBBCCCGEEEKIfUl4JIQQQgghhBBCCCH2JeGREEIIIYQQQgghhNiXhEdCCCGEEEIIIYQQYl8SHgkhhBBCCCGEEEKIfUl4JIQQQgghhBBCCCH2JeGREEIIIYQQQgghhNiXhEdCCCGEEEIIIYQQYl8SHgkhhBBCCCGEEEKIfVkf9wCEEEIIIYQQQgghPiy+7+N5HlprALTWOI6DZd0agbTbbSqVCmEYYpommUzmtmNEh9wVIYQQQgghhBDiMyoMQ2q1GmEYYlkWiUQCpdSH3k+lUmF5eZnt7e3rgY3jOPT09DAwMMDY2Bimad7WTmt9V+NZX1/n6tWrLC8vc3lmhkK+QKNRJ1SKZCJFX18Px44dwzRNdnZ2uHz5MgsLC9TqNTQQj8fpymQZHh7mxIkTPPLII0xNTWHb9od+Lz6JJDwSQgghhBBCCCE+Bs1mk2azCXTCi2g0+hvp13Vdrly5wvz8PDs7OzRaDUKt8T0fUxlks1kmJiY4ffo0qVTq1+pre3ubN954g6WFBeo7edxqDa/ZBK0xLAsPqPltSs0mSilM06RUKhGGIbZt09/fT29vL+Pj4wwODjI6OsrExASRSASAzc1NXn31VVYXF7n81nlyK6s0ag08L8DzA0JloC2Lsu9SrJZotZt4QRufEBwDLAMM1elbGSRjcQbfHOD5l37Imfsf4JlnnuH06dMYxme76o+ER0IIIYQQQgghxG/I1tYWMzMzrK2tUSpXaPsBAI5t0ZXNMDo6ytTUFD09Pfuew3VdarUaALFYjHg8vudx9Xqda9eukcvlKBaLuK7L2toam1tbtPHxbE21Vadar1GulCmVSjQbTZQXEtUWSSvO5NGjnD17ls9//vOcPHnyemhzJ2EYcu7cOc698Qb5hUWaO3mGunsYTWdIDQxTrVaYnZtjYX6ela1tiq02hcCnHoYkzQgxNA4h10xFIhGjt7+PoQMHGD9xgq7hQe6//3583+ftc+dYfucSy1euYmMRDx1GhofJZHuIxuLkq0XOX71Ebn2BklfHVyHETYxYBDvjEEnGMCImoR+ivYBmACuNHLnZAkvbq1xbmueZp7/GM888Qzqdvsen/emh3l0D+NvszJkz+s033/y4hyGEEEIIIYQQn0qlUonNzU1c1wXAcRwGBwfJZrMf88g+ParVKq+88grX5hfZLNTJV1p4IUSjneCn1awTsRQ96RiD3QmOT07w5JNPEovFACgUCkxPT7OyskK5WCTYfVZGJEIqk2FkZISpqSkGBgao1Wr84he/YPbaHLlCjUrDpVyps7RwjUqjRJ0a0XQMO2aiIgbNoE2lWcUNfAId4NVdvEoTXfeIBRa9qS6OHTzKmQcevKuZOGEY8uMf/5hL595i+8os4/0DTB44iGPbhGHI3Nw15ucXubqwSrXRxgug1qhRd6u4YQDJNKPdI0QBv9mA0EUpj77+XiKJGMneHtrxKNVmg1izjWq2wDcJtcnA8BhOtHPP1rY3ubhwmYurV6hrF19pSNmY2RiqK4Zpmzi2jRN1iNg2bbdNs1KHVkgEAzswydoJjo0d4bmvfJ1vf/vbZDKZj/ZF+Q1TSp3TWp+503Ey80gIIYQQQgghPoPCMGRxcZHp6WkWV5YoN6v4YWcWjGWaZKMpxg+MMzU1xcGDBz9Vy3Zc18XzPAzD2HfWzodpeXmZF198kWtrBXJll8HRQ9x3/CDxZOp6PZ8wDKlXy2ytLXF+bontUoONjQ2eeuop5ubmuDozQ2Nzi9ZOHu22SUSjKKVotFrkLZPNnm5mfvUrrFgMz/NYy9fZzNfo6h0iNTBKoTKNmU1iJAP6esbYya+zVtikFTQwYxaRVJRsby/ReBTLtlGGQSNXpnhtg3Klxbn5d1grbbKwusTvfPmrPPvssyQSiT2v9+c//zkX3zxH4eo1zp44Se9uCBmGIdPTl5hbWGZ2YY3QjKEiUWyvTr9tY1vdNJSmojWtwGNo6DCmUlQKW7SbVQo7ZQYNG387z2p+m1qzSTSRYDDZSzSeZGTk4PX7uZHPcXlllourV2hYPtowiaQdzJ44YdRAA1pD2/MBMA0TJxYlEonQqNfxa220EVBTLgu5FV746UtEo1F+7/d+7zNZVPuzd8VCCCGEEEII8RlXLBZ54YUXWM2tk6vlKXt1eof6iEQdAOpNl+XNqywVN7g0O8NAVx9PP/00fX19OI6z5zm11pTLZVqtFkEQEIlESKfTex7v+z6rq6s0Gg08z8O2bRKJBKOjo3sWTf51aa1ZW1vj8uXLbG1tUS2X0X6AMg2iiQR9fX0cPXqUw4cPf+jBwPLyMt//q+d5Zz5HNNPPw08+iB25/Z4YhkEq00Uq08XooUmuXjzHq29d4eWXX+ZQdzdmqcLBvn4OnThJJpG4HuZprak2GyxubvLWK6+yVKuxqW0m7nuYM0/8Dk40xsVzP2OnlqNptRk+doKN1XkaYYvAUWjTwbcVsUQUwzaxb3pe8f4MkVSM0tUNIq6i4rU4P3eJWrOO53l861vfui18W1tb48Jbb7FzdZbHTp6iJ31jps61a9dYWl7n2tImKpbFUja6WcZo1klaDql4Gq01q7UCpXqF9fw6BwfG6eofo17eoVHNk9suoMI2yTAgYhoUCmUWWpovPHryenDUaDVY3Fpmen2OdlShQ4WZcTC7YphJB1Nr2oHXSY8MA88PaLWaJMwkyjRIpJI0jAZBvY3r+VT8BgubK7x98QIHDhzg7NmzH+o78kkg4ZEQQgghhBBCfIbkcjm+9/3vcXljnpblMXrsIPcdGMK6aVepMAzJbeWYPv8Ob719Abul+fGLL3Lq/vsZGhnhxIkTHD9+nHg8Trvd5sqVK0xPT7O1XaDtB4QaLFORiNocnZhgamqKwcFByuUyMzMzXL58mVx5h5bXJtABpjKJ2Q4DXX0cP36cqakpksnkh3K9Ozs7/OQnP2FreRl3K4dXrmB4bSKWjR8GlFDspJMsXLxIanCQxx9/nImJiQ+l70ajwY9+9BIXF7bJDh7iyPH776pdNBbn8LFT/OAv3kStzJHuSvH7X/86meTtxauVUqTjCca6uthwYiwvb9KdSOG1mtRrFUqFbdY3Fim4BQ6cuo+1tXny1W0aXg2SNvF4EtO2cZt1qrUapmkS3V0qB2DFIqSPDFC5ukE6Gic0DBZzq/z456+QTCb5xje+cctuaD/72c/YuTbP0eHRW4KjQqHA6toacyub2PEsATah18Ro1IibNql4+vr1DCe6aJS3aNYrVBoVMokMiUxvJ6DMVQgqZRKxKD3KodB0cdIWa9vrHBwcA2B+fZnFnTVc7eHrACsdRUdNjGTk3ZuGbVq0PR9tKJQy8f2QttfuhJ1KEYvHqAUBQRDgqZAmbc7NXODgyBinTp360N7PTwoJj4QQQgghhBDiM6JSqfD8889zaX0WuzvGww8/jGndOtNnY2OD5eVlmuUyul5lpCtKbmWbzXyeYHUVf/IYxbl53nz9dQKtAUWh1maz1MQNDaLxJIZh4nttWvVtFreqXHhnmlqlSDQWpaZbbFXzRFIxUtkUpuXg+j7bhRzzc6vM51Y499ZbPHD6NI8++uivtW38xYsXee2VV6jPL2JWqxwbGWbs4DjpePzGLBXXZaNQYHZ1ldzyKj8sFlm+/36++MUv/tpL9V577TXm1/LYie67Do4AwiBg5sIv6Q01oRFhKNXF9uYWmYm9dz4LAp/Z2VmKFZcHJj9HwWuzvrLItB3BNE1y9U16Dx6gXC1QrZeoNsqopI1pG0Tj7y490zRbDcxalYgTwTBuvBeRZBSnL41bcskmEignZPraZQ6OjHH48GFOnjwJdN6drdU1wlqdyalTt4xxYWGeta0iynJwA0U8HqFW3sb2PFKZ3luONQ2DvliaNbdBsVYkk+iEUPFUltxKGycI8Zptal7IgVQ3W40GW/kcgz0DuO02xUaZ7WoBPwKGYRIaYGWit7xLShmYhiIMNCqi8MMQz/NwIhFQCmUYRKNR2qFLq96m0qyRiSZZ2lhlZmaGhx9++K6f56eBhEdCCCGEEEII8QnXbreZnZ1ldnaWWq1Gu93Gsizi8TiHDh3ixIkTxONxXnnlFS7MX6JlBxw+egKv3ca0OrNMtNbMz8+zurCAt72NAwxnu+gbGkZNHuPSpWtEqyHdtk22VuON119nPdBsWEmOP3CWwycfobt/+JbApdWss7pwlRde/gHN5jYq5XP/mdOc+sJDpLK371xVzpdYm1/m7fVpam6dWq3Gl7/85Q8U4ly8eJGfvvQSlekZjvb387mpKaw9lsTFHYcjQ0McGRri2vo6b0/PML279O4rX/nKBw6vyuUyV2evsV5ocObJJ+6p7fL8ZRobq8Rdl5Of+zzri1dY31jn4PhBbMu+7fi1tXW2CxW0GaGrb4AuDfWFq2wsXiOv23gZk2RPFwuzlyjVC5jxCC3dJpnsun4O23bwPY9WO6Ber5NK3fp84oMZitsr+J5PPJPAwObc9HnGRw5w4sQJDMNgbm6O2laO8cGhW55ZuVyiWK5SqjbAThN1HNxGBeU2iEcTGOr255t1Ymw2yjSbNTzfw7Zs2m4Tx45gYqLDELcdkLIdsiqk0mywXdzB9dps10uEBgRKY0QsAtvAcG6PP0zDIAxD3t1ILAxCwjDE2H1P7IiN23JRtiIINa72WFhbYmFhQcIjIYQQQgghhBCfDLVajfPnz3P58hVy5Tq5skvT9Qm0xlSKiG1waW6VF1/8Ee22y4V3LrDRzHPg2CF+8dKrWIZFV28XI4cP0PBarMzP4edyHOrrp+89u0odPXqQt9+cZqjZ5HAqzUS6i+25ZUaHuwnDkHRX720hjxON0241SfR3k8vl6BoaoGX5KHvvMCjTkyXTk6W4XeDSL8+jL2kcx+HJJ5+8p/uSy+V47ZVXqEzP8PChwxwZGrqrdhPDw3SnUrz0qwtctSwGBwc5derUnRvu4fLly2wW6/QNjeFE774odxD4rC1dg9wGk2NHiMfjxBJpqo02ua0tRkZGbzle65DNzQ1KtRZ9I4dQKFBwbHScuXM/ZccJGR4/Sa1aptlu4Ic+gamIODGUeu/zitGoVWi1WiQTKZRxIzgzbQsrHcNtt+lK9dGolyiWSqzlNlhaWuLQoUPkcjla1Sp940duOe/OTp5qvYVh2jQ9j0wyS724gem5xBJ77+hnKIO4FaHse7TaTWzLxm+3sFAYhokOPbwgwA9CetMpStUCxVqZMAyptGqEBmhTgakwnL3raClloMOAMAwxTYsw1ATBjfAIpbAsC21D0ApwA49SrUKhULheq+uzQsIjIYQQQgghhPgE2t7e5vnnn2duPc9msUUi28foifvJdPVi2jaB77OxPM+br/yA7dwKTb9IXVfIDPfQiAWEQZt23WWnXmR+fp7yToGuqMmj9x0ns8cuWtFohEx3iuWNHdydAlEzyVce/RIX1xfJL88zk/gFp848Sb1Wxmu30VpT2tlicXGGopfn1Je/TKmYY7NQZWbmMmceemjfWT1dfd2cOvsg7/zsLZwLEQ4dOsTo6Oiex76X1pqXX36Z2vwCk/0Ddx0cvas7leLsseO8OjvLLzO/4NChQx+ovs3GxgalqsuBqbsb97t2ttbxiwWSlk0m0VmmlspkqeXXKZcrjIzcenytVqNSa6CVRSJxY7ZQNOIQMSw8t4infXS9Stt3wTbwgzZx5/ZnbBgmyjDxbq7/cxM74eA3m6BB2waRuMNqboO1tTUOHTpEqVSiXa+Tec/9qtWq1Bud8IhAQRigfB/btPecdXT9GkybahDgem1SgO95EAZYVoTQ9VFKEQYhUdNE+x71RoNQadx2m9DQKAM0oOx9irArhVKA1ihAownDALgRCpmmgWcq/NCn5bqouKLpNqlUKvT09Ow79k8bCY+EEEIIIYQQv5VarRZLS0s0m02CIMBxHAYGBujr6/u4h/ax29nZ4S//8rtML+fR0S5OP/lFEqlbZwoVchvMX/kVQQoSqT6aBQ8zahI70E1kKEU06qADTX2nzNrsMm61RLQZp5ArkTm09xbsw8N9vD67QrWoeOr0cZLpNA/EjvPjX73BbKvJzuYaAT6hDtDAxsIcpXae3qkjKMtgcOwQC1cvUqrUKJaKdHd173uNmZ4sY5PjrC3luHTp0l2HRysrK+RWVrBrNe6fOnnX9/Rmo729jG5tsrO2xqVLl3j00Ufvqb3WmkKhQK3VJpXpunODm1SKO1Cv0pu+0S4aS5D3Amq12m3H12o1Wm2fWPz2ZxZzHIKqix94+C0f13PB7GxTb+yzq51pWQShj+97t4VHZtTG1XV8t000GQPfI1/Ms7OzA3R20dOhxnzPDLRms0mz1QYjhmWb+H67EwKZ7x9JGEoBIVqHAGhC0BrTNPG1BmUShhodhEQMC7fdxt+9LA0ow+j87deom6UMA5Qi1CFBEKBUZwlbEAQf+JyfRBIeCSGEEEIIIT4yQRDQarXwPI9IJEI0Gr1j/Zrt7W0uXbrE7Ow18tUWrh+gNZimIhOzGRseZGpqiomJiY9kW/ffds1mk+eff56ZlQJWepATD5697Z4Wtje5eO5VNmorpIZ7GRsfpfnGT6lWapTKFdq1KulUCssyO0Wvu23wIhRrbeaW17Ask7Gxwdv6jjoRKl4T07VJd/fiBwFX1pbYLOXZrG6yFW6THRrAdhw812W7uUld1XDCCgtX3yGRSJNIpqk0Sqyvb7xveAQwdHCEpcvzzM3PU61WSaX2Lhh9s8uXL+NubnFiZHTPGkd36/jYAX506RJXrlzhkUceuafaR77v47bbgIFtR+6p31qlDG6TVPbGrBbTtgnCEM/3bju+1Wrh+QGRePS27xw7An5A221hRmz80EdFFAb73xfT7IQ7e4UjCtVJZXRnFzZtBjTdFo1GAwDbtlGmgRcEODe9k2EYEuoQxbuBELBbbP39BDrcnR3UOZdhGKCMTogE8G6opHfPq/SNrEgDIWACod6/k9u+2ntM10MxDZZpfqaWrIGER0IIIYQQQoiPwNbWFtPT08xem6Xte4S7NXhiTozJyUmmpqbo7r49OHjzzTf5xetvsFVxyVdd0r2DJHu6MEyDdqvFlY1lVvLzXFte5/DoRZ599lni8buvJ/NpMD09zcJGAT+S5r4HPn9bcOT7HpfefI1cY43MaD+x3gxzMxeo57eImAFJrXFCiPoeyWiESrWGXa/jpG1UVHF1NUc4G5DJpEinb53NUq1W8ANNJB7H9ducn7/KajNPrUujdEhsKMvhRx/CNEw2rsySHOklkR3E6opSLhdpVhokW2mCdpudnR1arkv0PbNbbhaJOvQO95OrF7h8+fJdFSne2trCK5cZO3zkjse+n950mqjW1EslyuUy2ezetXn2opRCAaF+n9BiH4HvQRAQubkw9m7QslesobXezT9u/zbmRDE0eG0XM3K3Yce74c4eY2v7GIaBGbH2/L6rq4tIPE65VqO/68bMKcuyME2Dtu4UpFbKBsMg9P33HUkr8AijMRzb2T2PA5ZF0HZRhkIHGq01Sin8MMC0LAw6YZOlTHzlozVob59ZQjpEq059pVBrTFPdFkgHYYjSnbtvWzamMkjE4mTeUxPs007CIyGEEEIIIcSHZmdnh1deeYXVzXUKrSKFVhnLsTEtk8ALCIsBq+VNLrxzgYNjB3nqqaeuzyb5+c9/zs/feItruTrDh47x8CNHiSdunWkyeeohttaXWbxykersCu32/8e3v/1tYrHYx3G5v3FhGDIzM8NWqcnkQ4/sufQot7ZEpVlAJWzsbIytxVkixW0SQYAVcxjpHaRWLWKjSMdihG4bKxrFNi3qtoebsVks5OlbXOGB+4/fcu5Wy8UPQpxIlAvzV1lqblN1fE5MHGO1tE0tYtJuNYnFk7RqdbygTXagn2g2Raq3m+LaJpVWEV3zqDbiNBr19w2PAHqH+tm4uEixWLzj/Wk0GtQrFcwgJPUhvBNdqSSVeoN8Pn9P4ZFlWSQSCSKWotmoEYvffc2kzlIrRXjTbJm262Jbna3jb+/LxjQUwR5BTDwSxdYGQcvF7urGNq1OIMP+S660DlEGtxTLfpffcIkYJk4iRqNYxQpC4rHY9QC3v7+faDpNrli4JTyKxxMkYlHaNR/f11iRNNq08P36vuMIwpCm74GZIuZ0rtt2YmDZtNotUqZFEHjo0CNU4AOxSBTbtolGHJptF2Wa6MAnbPnXQ6Zb++h8pgyFDgOM3ZDrZqEfQNAJkBzDoifbTW/v7cXhP+0kPBJCCCGEEEJ8KFZXV3n+Bz9gqbRKnRajhw9w4sgDxJM3ZgZVSxVWri0xu7jE9tUChUKBr3/96+TzeV4/d55ruQYnH/4CvYMje/ZhWhbDBw7TOzDM+Z//hMvLOV544QW+9a1vfeAt1T9JFhcX2dwpQSRBV+/AnsesLcylnZhBAAAgAElEQVRS9cpEe9PkV5ewC9sMZ7uxtSana3h+gGlF8AOfZrOF1hoDSDkRMlEHo1ezmFvn0uwCRw6PkU7emH1UrzcwQ0Wl3aTdKFBxPCaOHca2bSzDBB0S7i53Cv0ATYhhdX7tNAyT7tFhCqvrNCsVcpub+N77zzwBsJ0IQRjguu4dj22322jfJ3rXs2zeXzQSoeR7tNvte27b19dHMrZKubhzT+FRLJ6k5kSptxqkdusYNRs1HNskuceyvWQygWNblJq3BzG2ZRExHNyGixOLEbGjnaLPJoRBsGf4GAQBpjKwrFvvYRiEtMsNkrE00XScwkqOpGfSO9JDb28vABMTE7w92M/yhYscPzh+fdlgOp0mGY+SL+VRRNEalB0hQOMFHrZ5+/Mqug0CO0IilsLa/T4SjaEiDu3AR5s2tmXge22qnouKREjE4sScGGknSbVdx/B8AkAFmrDpY8Zv7kcT6ABldMZoGJ2d1dRNoZAOQ4LARzc8ElYEKzSYGDvE0aNH3+cJfjp9tqIyIYQQQgghxEcil8vx/A9+wGx+gfhQmqeee5rJzx2/JTgCSGXTTJ05xReeexqdsbi8Ncf3v/99XnvtNRZ36hy9/+F9g6ObRZwopz//RfJNWFhZZ21t7aO6tN8q8/Pz5CotRg5O7Pl9ubBDsZTDw6XRqmCV8gxluuhOpsnEk0S1Rb1WJ+JE8QKN22rd0l4pg6GuNKlMjIbX4sLla7d8v5OvEHUVvg7IG3UOHj5wvfaLfs86puszV25auqWUomtkEM8IaLpN6o39Z568S4edujd3M9PDNE1QBkEQ3vHYuxEEIRjGB6qtdfDgQfoycTZXFu+pXSrTBbE4pXoV6CxLKxd3SMUj9Oyx1DOVShF3bNxmA8+/NeQylEHMSWKFJioAx4yi/RDLtPH2COM0Gt9rY5nqtpo+rXwVW5kksikCz8cMwWu6jPQNXS9m3t/fz/CBA9jZDJeXblx3f38fqUQUpUMc26DZqBNLpAmdGI3W7e+AFwZst6qE0ThdqZsLjhsYpo2KOrQCn1jUxrYUa8VtjHiC3kwP/dke+uJZzNDo/DEVeCFBuYm+aTbX9XfEMNBhiGWaRN5Tn6rttlEhBE2PhBljoLufod4BJicnb39wn3ISHgkhhBBCCCF+LVprXnrpJeaLy3SN9XLq86fv+Mu2HbF58AsPY3dHubBwiZ/94g0CM8rQgcN33W8kGmX08CTbVZfp6elf9zI+EZrNJq4XkkjvvYSqViniBk2UY6FaTZKWRc/uLmzdqQwxbBr1BkoZaK0JdLhbwPjmZVKKkYFuAh2wUyjR2p1143k+lVId1fAxoiZOIko8cWNpWBAEoAyM3R20TNvGUCa++55AwzCx41E85VHIF+54za1Gk4hp3bbz114SiQROMkEz8K+P+9dRrNUwY7F7WrL2romJCQZ70rj1Evncxl236x0cgXSG7UoRz/eolAqY2iebSuxZJ8y2I/T29pKMRSgXdq5/HuqQXLlAdvgAI90HqOcKRO0oETOC4Wu8Zuv6Lmbv8tptTEMRsSNY1o2FSoHr0VgvknQSZEf6KKzksH2Drmw3w32DHDhw4Pqxjz/+OL0Th1nY2iRX7DzfaDRGf18v/b0ZvFaFttvAjiQgGqPpubS9G89Ka81arYjvREkkM6Ti6RvjCDy038aOxwhMA9OxCaMWJbeBaUfoSmeJOlH6sz30J7uJahPDBwINbkBQau720dk5jd1/T5mmgR2xseybrtn3abtt2sUGqUicGDaPnX6Ys2fPEoncWxH0TwMJj4QQQgghhBC/luXlZTYLOfxIyNTDp+66nWEYfO6xB1kprrOS26Grd+Cel54NHzxCse4zN79wfcenm4VhyMLCAj/84Q/5zne+w5//+Z/zne98hx/96EcsLy+jP0BB44+T5+0WH98nnPM9j0CHNOpVzGad7uSNX7yjdoSUEyMSmJRLlc61h2BHIijLxPVu7ORlRywcy8RvttgqlQBYWFijx45jKSiHDbr7bgQZzXaLltYY0RhOtBMoJbqyOGaU+tattYo8z8WORQjNkFq9hufdvoPYzTaW1uiOZ6/Pbnk/hmHQ29uLlUyS2x33B9VwXapui0gyuWdocye2bXPmzENMjGS5evEcbbd150Z0lq11D46i01muri6ws7lKXzbO+Pj49V3H3mt0dISudJRyIUer1QlIVne2aDsOo5NTHBybJEaUsNomm+gmaHqYGLRqN2b9hDrEbTVwbJPYTUXoQz+gPLdF3I6SGegmaPsEFZd2ucGZk6d58MEHb5kV1t/fz5lHHqF/6hivX55mfWcbgMOHjzDU10XMNrB0m3qjSizVTRhPUa6XCMOAIAxZruWpKo2KpxnqHr5+Xq1DKvkNnIhJNBYjO9iHF42w4TaId2eJ+D6teg2Ag4NjHOgaJGHGiIQWljLQrk9QaeEV6rTbHphmZxdHw8C2rVtqb4VBSKPWwC81sXxF3DP50iNPcPLYFFNTU3f1HD9tpOaREEIIIYQQ4tdy6dIl8s0iYycO3nMRWTtik+hKsbFUpu0277lvJxoj1dVD3W1SLBavF+5tt9tcvHixU1y6mGOnWcL124Q6xFAGUcvhncuXGOzpZ2pqipMnT94y0+K3VSQSwVQKf59dqpRSuM0GvlsnErdIx2/dLW2gq5fGlku+WgfTJ51wiMVi1Ot13HqDhOOgjM6spLhtoz2frWIRv+pSzVU4nhkiTFW4EjZI3rQksVirEsbiZLr7rr8D2aFBcvML1Eo5/KaLFev8ct6o1UjEIxhkcf02jUZj352rqqUKbqXJwKFeJib2Xqr3XkeOHGHp4kWurq1xoL//rtrs5dr6GnZvL4cOH/7A78apU6dYWFigUp/lV6+/wqmHn8CJ3nl3wMPH7mdrdYGZ11/l9PAAw4OH6O/fu8YVQDqd4cDoCM2Wx8bqAun+QZa2N+HQESZOPEDEidJq1gnyi9R3NsjEMhSrRXTCpGXWcWJxmvUatmUSi0WvF+b2Gi6VhRxOaJLJZohnU2xfW8Uoe5w+dh/3HZvixIkTt43noYceotls8rYyeOvyVTZ2djg+Ps7RoxO02m3emVlA+wFtO4kZT9IOPJYK6zSVInBiqGSWAwMHsHfrLoWBRzm/AX4L329wYGwE09LM57Yw0wnMUDPQnWZzbZFkuovunn6Ojh2m5bWZ2Zoj8AK0MnHrHl7bw2hEMLpjOKkYtmUSj8Wv1zry2h71UgWv1MRuaTJ2mqcf/yKPP3yWr371q5+J2mp7+e3/t6MQQgghhBDit1a9XmdxeYmqX+fBQ2Mf6Bxdgz20zQW2N9YIw/CeAygr4hAEjetFjavVKs8//zzzG0ts1XcwYjZjxw6Q7s52dn3zA0o7BZbnllld2GR5e435+Xl+53d+53r49NsqkUgQi1hUijt09w3e9r0dcQg9n6DRINE9cNvm7vGIw2hPP+2NFfLNGnmrQiKRwnEcWm2XZrtNPBrFc30cw6Ttt1meX0PFuzke6+Px41NsbW/jNDya9Qp2pJe271Fu1Al7B8h0913vy4rYZAYGqC1XqK5t0zUxiu+3cZs1upIRonaKUGu899muffnKAv2pHo4fP37XAc7Ro0f55cAAO6trLOW2OPg+oct+qs0ml9fXSZ48ycmTJ++5/buUUnzta1/DdV0uza5w7tUfcfjE/fQPHdj3PQ+DgFJ+i1azgTM6QMFt4mTvvC38oUOHqVSr5OeW+Om5nxE/forDh4/TvVtY/XOPfAH1hoGtLObXLuNYBs12E9fzaVYqRJNR4k6UVCqF32zT3K7g7lRJOnFS2TSxdJKN6QXMasDE+BG+8PDjPP3003uGKUopnnzySbLZLL9Mp9iZX+Slt87Rm0rhpJOMjPQwt7RGtZrHxaTpNvAI8cKQdCTDZN8Y0UgMrQOatQq18jZG2IagxdDQAC4eqZ4+Hjh9CleHJC2bufMXqFar1PMN1rfWiMcTZCNRBiIZ3EaTpt8mNEK0B2HQRnkhfs3H7krQ9hoEOqBVb+FVmqhmgBNajPUO8o0vP8vXvvJVHn/88dvqQH2WSHgkhBBCCCGE+MCq1Sot3yWZTWM7H6wOSDKTxIxatNtNvLZ7fdnT3Qp8D2O3wG+9Xue73/0uM+vXaFoeJ594gO7+ntvaZHu7OHjsMDsb21x5+xLNhRbe9zyee+65PbdD/21x9OhRzr8zzdXleQ5MTN0WQHT1DmDj4Jbr6OG9l+Rl4kn6onGoNrB9i621bUKlCcOAutck4bbZWN0h3KxjhJouz2HiSDdf+9wDdKdSdGWzON42XqtBuVRgu1nHT6ZJdvUQeU9dop6xEUqbmxS38hiORdv2STgW6VSKoOFBCPvN41iYuUZ9u8zE6PF7CnAikQhnH3uMH5fLvHH5CtlEkkwiceeGuzzf57XpS0SGhzl26j6Ghobuuu1eYrEYzz33HKm//mtmrs4xf+0dFq9eon/4AKlMF9HYjR3VqqUCuY1l4jZ88ZEpTHWCsNXiwtw8m4UCJ8cPkU3uvXNbq92m7Ths+00iYwNUvWZnJ7JKkVS6i1S6izNPfIVr0+dJRlJcmb9AY6tIw9/GjFuEcRsjncJfLmNog1jEIWMnsO0IjUqNynqeZGjz0NRpvvyFL91V2Hrq1CkOHDjAW2+9xezVq9TzefxqHdP3iLWb5Le2ydca1G2TtrZwIg62VszMv0PEUCg/wFIapT2cqE2mK0PN1hw9dZKBI4c4+9hjTE5OcuHCBQbHD7I6v8DS7Cw7G1vkWgU8PyQSj9JNL0ajiqs92nRmH+m2D01NPd8k0CFKg6EUMcOmL93HA1OneO6bz/HEE0/cUtPps0rCIyGEEEIIIcQH1m63CQlvKTR7rxzHwbYs6vUKvte+p/Ao8H2qpQKjAzHi8TgvvPACVzfm8GKah588i/0+W7Yrpegb7ifdneGtl3/JldVrvPjii3zzm9/8wNfyURsZGWF4oJeF3CL5rXX6hm6tAxSNJ+jpG2R12aRcKDHWd/usG9/3iSiYGhwk1tXFTqVC1XMptmvUGz7l9RL1zSqHWhHShk1vdoDnHnmEvky2s5zNiZFKpbCMkPn1RdxEmuTQKAOj47f1FU0lGT5+jNbbb7Hxq6tkxjKkJ8cYHBhk8co8ZiR227sT+AHzl2bJr2wx1X+Er37lq6TT6dvO/X5OnDjB8vIyV5pNXrpwnsdPTDHQ1XXHdtVmk9emL1GPxxk6Nsljjz12T/3uJxqN8uyzz3LkyFXefvtt1ja3KVQ22NxewvWDzjG2STIW4b4DWUaHBzhz5gzj4+NcvnyZ1157jfLiEj965wJZx6EnnSEZi6GUot5qUahUyNdrxAb6OfPt50ikUlSqNTYLVS69+QpaWSRSGWw7gg41juUwkhkm5SvqzSINv0W77OPmy/i7dcBcx8aJRUlE42SdFCO9h3j49EN8+ctf5nOf+9xdzxDMZDJ86Utf4uzZs6yurrKzs0OlUsF1XVZXV5mfn2d5bZNq3SVfrFAulaj6HrYOiJqaeDRCd7aPobERRg8d4uDRCY4dO8b9999PKpUC4MyZMzzwwAOsrKyQy+XY2Nhgc3OTWq2GUopMJkMikWB7e5u5uTm2cjm28jmKpSKtZgsdhkQjUQb6+nnggQf40pe+xIMPPkhfX98dru6zQ8IjIYQQQgghxAdm2zYGimD3F+APYmBgkIgJ5fwOpnlvv6JsrS2RsGFsZIhSqcTi+hJl3eDRJ5543+DoZk7U4YEnH+YXL77G/PIC6+vrDA8P37nhx0ApxdTUFHOrW8xfvkBXbz/We7YXHxmfZOaNn7CTK+Af9m8JZzSaarlAzDbpznbR39/Pgb4+CrUa1UaTYqnIzNY1Uq7FZCSOYdtkstnrtZP8IEBrTbPaomZrVDaNth1812VrbYlMdx8RJ4qhDILAp1YtUa5u46Qc+poJKAc05gusVnzcVoP48CDJ3Zk09WqN9flVtpbXSUcS3D90jK88/TTj4+Mf6F49/fTTBEHAvGXz1zPTjHf3cHx0dM+ZO03XZW5jg+nVFSIjwwxOTvK7v/u7H/ostMnJSY4ePcr6+jpra2vs7OzQbHZqfSUSCXp7exkdHWVw8MaSxOPHjzM6Osr58+e5euUK9UKBzVqdoFYBwIxEsEYGGchmOTo5yenTp+nq6qJcLjM9Pc3s7CzlSo16yyMImhiO4tCpA6TPnmB8fJzu7m4WFxe5cuUK8/Pz1Jr1ThHzEGKxKP19/Zw8eZIHH3yQycnJD7zTWDQaZWJi4rbaVWEYksvluHbtGvPz8xSLRVqtFlprHMehp6eH3t5eRkZG6Ovro7+/f88ljKZpMj4+flfvS7lcZmVlhVKphFKKnp4e+vr66Orquudls58VEh4JIYQQQgghPrBEIoFjRqiVNwn8ANPaexew95OMJYhoaLhNGrUK0fjdLTEKw5CVuSsMpSJMTU0xPT3Ndr3IwROHiNzjErpoPMbYxAF25ne4dOnSb214BJ0wYWZmhtrVVX71+ivc/8gXbgmQDh49SaKrn9pGkatX5jh2fALTMtFoKsU82muRTEav7yCmlKInlaI7mcSrtDicHaDb7GfAjjC9s0Mhv8OfvfoqhgKtFCWliSa6qcYMznzhUQDy+QKVZoPtlWsEoUZrjaEU0YhJX9JmYugIceskjUKVUrHEwpVFWs0mm2WT1ysevuejQuhP9nBf/1EOjo7x6KOP3hKi3CvLsnjmmWc4PzjIm6+/zsbaGosXzpOwLLpTKRzbxvMDyvUapWYLu6eb5H0nmbzvPh5//PGPbPmiUoqRkRFGRkbuuk0ymeSJJ57g0UcfZWtri+3tber1OlprYrEYfX19DAwM3DLmTCbD2bNnOXv2LNVqlUKhgOd5WJZFNpslk8lcr1d0//33A51/phqNBkEQYBgGsVjsIy8kbxgGg4ODDA4O8sQTT3ykfb0rk8nsW6Rd7E3CIyGEEEIIIcQHlk6nGRkaYbW2xcbSGqNH7r02yPriKmM9w4Qxk+m3fs6DT36VeCL1vm201sy8/UtMr8bQ2AA9PT0srSxRDRqcHr/zlu57GT18gNdm5pmbn+Ox+mMk7qFOzm+Sbdt8/etfx3X/X2aWtnnrtR9x+Pjn6BkYRimFE43ywGNf4e3vbtPMVZlVcwwO9xMGLvgumUSE4eHhW2YktVpt5udXaRUbTKWHePaph/jJhfN0Dw8xeOQI6UwGHYZYts14Nkt6bo7F+gaDPf30DvXjtlw2NzfJ5/N4nkeoQyzTIh6PMzg4SHd3N8roBBW5tU2K6zukzV4ePPogEcfBNAyiEYfJyUmmpqbo6bm9TtUHYRgGDz74IIcPH+bixYtcvXqVVqXCTr2O9gOUE8Hs6yWbiHPo8GFOnjx5T6HOb5pt24yOjjI6em/veCqVur7E6/0YhnF9JpgQN5PwSAghhBBCCPFrOXnyJP8/e/cdHHd6oHf+eTsCjZxB5EhEhiExw+FwNCMOKU5crbxa+3at2tWtfaXaO9fa5z2X93xln311Z1fZ5T+89p3l0p3OJ9try1qVNmgkSpMDh5oAajhDIhA5B4LIqfN7fxBDTWADDRCN+P1UoYju/r1vP42qmUI9eH/v2zPSp6HugU2XR5FIRCN9Q2qsrFd2epaGJmd17a2XVHfiEeUdKbnvSU4rSwvqbruuwOy46oszdenSJY2OjmrOv6j8ksK4b1f7PG9ykrIKczW3vKiRkRHV1dVtaZ6dkJqaqq997WvyXr6snqEJ9d+4qu6bXhWV1ygjO1d5hSXKqqzT3Ie/UMAfUMfYmHzpySo+kqP8ylK53G4FAkEtL69qfPyO5mcXleNIUmNSns4fO3Z382W3RycfPaNvfOMbMsYoGo3K6by7suyjjz7Sy2++qs7WmzrxpRalZaarvKJc5RXl6+YOBUMa6OhVy7FTerT5tL785S8rFArJ4/HI4/Ek7Bj0zMxMPf7443rsscc0MzOjmZkZBYNBOZ1OZWZmKjc391CfpAVshPIIAAAAwAOpqqpSfnquJsfvqPdml6qbj8Y9tv2DG0q2XpUUFuvXfu3X9PLLL8vTM6C+j66q64ZXxeXVSsvMlnE4FPCvanJkQAt3JpSXlqSa0hw988zTys/P18DAgELRsFJTNrex8uf5UnwKLS4rEAg80Dw7IS0tTV/72tfU2dmp9vZ2jd2e1u3xW5oabFckYhUILCkpJ1u+hVlVJ2fKuHwKrjrU/nGfIjYqh4y8DqeyHckq8+arKr9AxysqFAiH9c7NNiXV1X5mY+RPiiPp7m1OU1NTCt8I66O3W1Xfcky5R9bfXHh5cUk3f3FdmcanxrJanT9/Xl6vd0dPt3M4HMrNzVVubu6OvSdwEFAeAQAAAHggTqdTTz75pJYuL6u7o19WUs0GBVI0GlV7603NjtzR0ZwqPfHEE/J4PHr22WdV1tamtrY2jU/NaHr0lqYHo7KSnMYoI8Wt6oocHT1aq5MnT97btyQajd7dZ8f5YJvdGoeRtXdXRO0HHo9Hx48f17FjxzQyMqKuri4tLy8rGAzKPlSl69evyzs3r/xoVBWFhZpeWNDi6qqC4bDcTqeSPB5VFOSr5kiRvC6X+m/f1ntd3XJVV6rh1Kl7e+F8njFG58+fl7VWrg6Xuj+4qT6fR8XVZSooOXLvljhrraYnpjTaO6SFqTmVZBaqsfyonnnmGXm93p38UQF4AJRHAAAAAB5YeXm5nvryeemN19XfOaTbIxMqq63QkfLiz2yiHQoENdI/rOGeQbmCRrXZlXrm6afvHYntcDh07NgxNTc3a3R0VL29vfc28PV6vSosLNTRo0e/UDx41/bNCQaCD/Q5QsGQXA7nvis2jDEqLS1VaWnpZ55//vnndfnyZU10dqpjZFQ5SUl6qLpaZbm5cq9thLzs96t7bEzdY6Pyu9zy1B9V0+nTG25e7HQ6dfHiReXl5enjjz/W5OyUJjpH1P1hh1wel4zDoXAwJJ8rSQWpOaorK1VdXZ0effTRLZ/YBWB3UB4BAAAA2Bb19fVKTk7Wm2++qdsLdzT0UZ9uXe9QakaaXG6nwqGIFmfnleZKUVFyngrz83XhwgXl5+d/YS5jzKY2Bs7JyVGqJ0Wjo5OqPV6/pb1zotGopsZuqya15N5JZPudz+fTV7/6VbUWFamzs1OLk5N6d3RU73R2yCEja6OyDqdc2dly19So4MgRnThxQg0NDXHNb4zRyZMndezYMfX399+9fW58TOFoRLJWTodTWZlZamxsVF1d3b4r5QDcRXkEAAAAYNuUl5frG9/4hvr6+tTe3q7hsREFQ0FFg1YO49CRzGxVllWoqalJpaWl9/bTeVAlJSUqzM7XyMCEZibvKKdw/f137uf2yIS81qmigiMPdET8XuN2u3X27Fk9/PDD6unpUUdHh6anpxUJhyVJHq9XxcXFamxsVHFx8ZaKN6fTqZqaGtXU1CgSiSgQCMhaK6/Xm/Cj3gEkHv8VAwAAANhWTqdTtbW1qq2t1cLCgpaWlhQOh+XxeJSampqQo8AdDocaGho0cHtYQ90Dmy6PrLUa6h5Qri9bjY2N255vL3C5XKqvr1d9fb0kKRwOy+FwbFuB9wmn0ymfz7etcwLYXZRHAAAAAB6ItVYrKyv3Tijzer3y+Xwyxig9PV3p6Q92Alq86uvr1Xrtmjpu96i/o1eVDdVxj+3+qFPhxYDyC8tUW1ubwJR7ByuCAMSL/1sAAAAA2JLV1VV1dnaqo6NDs/OzCtu7J5S5jFM5WTlqaGhQXV3djh3FnpycrIsXLij4s4C62/sUCYdV3Xx03duwotGobn3YrjuDE6rLrdKlr1yS2+3ekbwAsF9QHgEAAADYlFAopKtXr+rWrVuaXp3VtH9eEaeVJ+nuCVqB1YAGF8Y0cHtY73/wvhrqG3T27Fk5nc4NZn5wlZWVuvjURZnXX9NA74jGB0dVXFWm4qpSeZN+tVmzf2VVI33DGu0bUrL1qD63Sl+5+BUVFxcnPCMA7DeURwAAAADitrq6qsuXL6truEfjK3eUU5SrppqTyinMvbfCx1qrO+O3NdQ1oJsT3ZpZmdfMzIyefvrpHTltq66uTikpKbpy5Yompid1p3dSAx098iR55XS7FAmFFVwNKCspQ9UpJTqSX6gvfelLB2qTbADYTsZau9sZNtTS0mJbW1t3OwYAAABwqIXDYf34xz/WzYEOLThWderJR5Savv7m1wuz8/rlWx8ox5GuY1WNeu6553ZkBZJ0t8QaHR1Ve3u7+vr7FYwEFbVROYxDXpdX1VVVampq0pEjR3YkDwDsNcaYa9balo2uY+URAAAAgLi8++67ujXco3mzokcvnpM3eeO9jNKzMnTm4jm9+/IVdQ50qeBagR555JEdSCsZY1RSUqKSkhIFg0GtrKwoHA7L7XYrOTlZHo9nR3IAwH63vWcyAgAAADiQAoGAOm91anxpSqeeeDiu4ugTySnJOnnutMaWb6u9vV3hcDiBSe/P4/EoMzNTubm5ysjIoDgCgE2gPAIAAACwoa6uLk0vzymzIEtpmembHp+Vly1fZqruLM2ot7c3AQkBAIlCeQQAAABgQ+3t7ZpenVVpbcWW5yitLdf06qza29u3LxgAIOHY8wgAAABY4/f71d3drbm5OQWDQbndbqWmpqq2tlZpaWm7HW/XBINBzc7Oyq+Q8orytzxPYWmRbr77kaampmStvXc6GwBgb6M8AgAAwKE3NTWlmzdvqru7W/OBBfkjAVkblZFDHqdb773/niorKtXU1KTS0tLdjrvjgsGgwjYit8cth2PrNy84XU453U5FohGFQiH2HQKAfYLyCAAAAIdaW1ub3nz7Tc0G5rQQXFTukTyVF1fI5XYrHA5r5va0BoeHNdVxRz39PTp14pTOnj17qFbNGGNkJFlrH3guayUjc6h+fgCw31EeAQAA4NC6ceOG3rzypkaXxlV6tEyn6x+WLzXlMyKU8WYAACAASURBVNdU1FYo4A9oqHdQ3R93KfxhROFwWE888cQupd55Xq9XTodToUBQkXBETpdzS/MEA0HZSFQul0tut3ubUwIAEoUNswEAAHAojYyM6O133tbo0riazjSrueXYF4qjT3iTvKptOqpHzp/RZOC2rt+8rps3b+5w4t3jcrlUWFCoVJdP44OjW55ntH9YGZ5UFRUVbWM6AECiUR4BAADgULp+/bqmVqdV2VSlsuryuMbkFubpxKMnNbkypevXrysajSY45d7R1NSknORMDXUPbGm8tVbD3YPKSc5SU1PT9oYDACQU5REAAAAOnbm5OQ2NDMlv/apuqNnU2KLyYrlTPZqen9bg4GCCEu49VVVVyk/PVWBhVVNjk5sePz44JuuPqCArT2VlZQlICABIFMojAAAAHDrt7e1aCC6oqKJYbs/m996pqK3UXHBB7e3tCUi3NzmdTh07dkwlqYW6cfW6Fmbm4x47OzWjjg9uqDS1UCdOnGCzbADYZyiPAAAAcOhMTU1pObSq4vLiLY0vLi/WSnhFU1NT25xsbzt58qSO1zerKDlPH7z2riaGxtY9gS0ajWq0f1i/fP19laYU6qFjJ7llDQD2IU5bAwAAwKETDAYVtVF5vN4tjfckeRSJRhUMBmWtPTQraYwxOn/+vIwxcnXeVNd7bbp1vUNltRU6Ul4kT9Ldn2dgNaCx/mEN9w7JGTKqTCvWQ8dO6vHHH9/lTwAA2ArKIwAAABw6TqdTRlI0GtnS+EgkIocxcjgch6Y4+oTT6dRTTz2lgoIC3bhxQ5MztzXVOabej7oUtXd/nk6HU+meVJUlF6iwqEDHjx9XQ0PDLicHAGwV5REAAAAOneTkZLmcbi3MLSgzJ2vT4xfnFuVyuOTz+RKQbu8zxqi5uVlNTU0aHh5We3u7JicnFQwGJUler1eFhYVqampSUVHRoSvYAOCgoTwCAADAoVNVVaXOvk4Ndg+qrLp80+MHuweU7klTVVVVAtLtH8YYlZWV3Ts97TDdwgcAhwkbZgMAAODQqa6uVlZKlpZmFjU3PbupscFAUGODo8rwpKuxsTFBCfcniiMAOJgojwAAAHCgzczMaGRkRENDQ5qYmFAgEJDL5VJ9fb2ykzJ1/RfXFQqG4prLWqvrv/hQqc4UlZeVKz09PcHpAQDYfdy2BgAAgAMnGAyqq6vr7l48U9MKRayisnI6jFK9HtXW1qimpkblQ+XqHu/RL169qjNfflTe5Ninr4XDYX149ZdanJxXRXa5HnvssR38RAAA7B7KIwAAABwot27d0ttXrmh60a+ZlaAiDo9S0tLlcDoVDAQ0OD6jsbmPdbO9U4X5uSrNKtHo7Jhe+/ErKq4oUUVtpdKzfrWiaGV5RYPdgxrqGZQ36lZ5Vpmee/Y5ZWZm7uKnBABg51AeAQAAYNctLCyos7NTk5OTCgQCcjqdSklJUU1NjSoqKuRwxLfbwvXr1/X21V9oYHpVaTn5qmusU96Rks+MX1la1FBvl/oGejTvH1d5fpbqy+o0Pjmu+aF5Xel5S+5kj9xut8LhsIKrfqW503TEk6/CvEJduHBBWVmbP6ENAID9ylhrdzvDhlpaWmxra+tuxwAAAMA2m5qa0rVr19Q/0K+l0JJWwiuK2qiMjFxOt9LcqcpJz1FDQ4NOnjwpp9MZc66uri699Mpr6pte0dETD6u0+ui67+1fWVbrldfki66quapE586dU1dXl2513ZI/4FdEUTnkkMflVnVVtZqamlRYWLjdPwIAAHaNMeaatbZlo+tYeQQAAIBd0d/fr1deeUW3V6a0Gl1RUXmx6srr5U3yKmqjmpueU39Xv6Zvz2hqfkpjY2N6+umn5fF4vjBXOBzW1atXNTCzoprjpzcsjiQpyZeih5+4qHdf+5l6h8d0dGpKjz/+uM6ePavl5WWFQiG5XC75fD653e5E/AgAANgXKI8AAACw40ZGRvTzl36u8eVxHaksUuNDX5I36bObVWfnZquqrkq3x2/r2tVrujV4S/q59Nxzz31hBVJfX5/uLCzLm5al8pr6uHN4k5JVd+yUun/5jtrb29XU1CSn08kpagAAfEp8N48DAAAA2yQUCumll1/SxMqEyurL9dDZh75QHH1a/pF8PfH0E1rSsnqGevThhx9+4Zq2tjbNLIdUXlO36Tz5xaUKOzwav31H4+Pjmx4PAMBBR3kEAACAHdXd3a25lXmlZKWo6aGmuMakpKao5fEWzQRm1dHRoWg0eu+1lZUVjU9MajUiFZaUbzqPw+FQUUWV5lZDGhgY2PR4AAAOOsojAAAA7Kj29nYtBhdUVV8tY0zc4/IK8+RLT9bMwoz6+/vvPe/3+xWKWiX5UuR0bW1XhtS0DIUiVn6/f0vjAQA4yCiPAAAAsGNmZmY0OTUp67IqKiva9PjK2kothhbV3d1977l7q5A2UUR9gTGyn54LAADcQ3kEAACAHbO8vKxQNKyM7Aw5HJv/VTQzJ0uhaFgrKyv3nvN6vXIao6DfL2vtlnIFVlflchh5vbH3XgIA4LCiPAIAAMCOiUQisorK4XBufPF9OJwOWVlFIpF7z6WmpiorI02OaFDTk1vb8HpsqE9pXqcKCwu3NB4AgIOM8ggAAAA7xuPxyGEcCvgDWxof8AfkMEYej+fec8YYNTQ0KMfn0VBv16bnnL0zpeDSgvIy01RZWbmlXAAAHGSURwAAANgxOTk58nl8mp+e1/Li0qbHjw6MyufyKT8//zPP19fXKzvFq+mJEc3emYp7vmg0qp6268pJ8ai+vl5O59ZWRAEAcJBRHgEAAGDHeL1e1dbUKtWTqv7ugU2NDQaCGhkYUZo7TY2NjZ95zefz6dixZpVkJun61Tc0Pzu94XzRaFQ3W68qOH9HxTlpam5u3lQeAAAOC8ojAAAA7KjGxkalu9M02D2oxfnFuMd1fNShZEeSysvKlZGR8YXXz5w5o+aj1SrwGX3wxkvq7bihgH/1C9dZazU1PqoP3nxZi5PDqspL07PPPiufz/dAnwsAgIPKlcjJjTF/V9J/J8lKuiHp9yQdkfR9STmSrkn6HWttMJE5AAAAsHfk5+ertrpW/i6/rr56VY9deExpGWnrjum43qGRnhEVpxbp9OnT973G4XDo0qVL8nrf1I32Tt3pa1Nf+8fKKy5TWkamHA6nggG/JkYGpeCqsn1uFRVl69lnn1VeXl4iPioAAAeC2epxphtObEyxpCuSGq21q8aYH0j6qaTnJP3IWvt9Y8y/k/SRtfbb683V0tJiW1tbE5ITAAAAOy8cDuvFF19U70iv5iMLqjhaoYraCqWkpty7JhqNamJ0Qn2dfVqYWtCRlEJdvHBRNTU1G84/MjKitrY29Q8MaG4lpGAkKmslh8Mo1etSYU6mGhoaVF9fr6SkpER+VAAA9ixjzDVrbctG1yV05dHa/MnGmJAkn6RxSU9J+utrr39P0j+RtG55BAAAgIPF5XLphRde0Guvvabu3m5Ndk+qt61HmbmZ8iZ5FY1azc/OywajSnenqyyzVBcvXFR5eXlc85eUlKikpERLS0vq7+/X6uqqIpGIvF6vcnNzVVpaKmNMgj8lAAAHQ8LKI2vtqDHmX0oakrQq6SXdvU1tzlobXrtsRFLx/cYbY74l6VuSVFZWlqiYAAAA2CUul0uXLl3Sydsn1dbWpu6ebq2urCqyHJFklOPIVl5RnpqamnT06FF5PJ5Nv0dqaqqOHTu2/eEBADhEElYeGWOyJP26pEpJc5L+VNIz8Y631n5H0neku7etJSIjAAAAdl9+fr7y8/N19uxZTU9PKxAIyOl0KiUlRbm5ubsdDwCAQy+Rt61dlNRvrZ2SJGPMjySdk5RpjHGtrT4qkTSawAwAAADYJ5KSklRcfN9F6QAAYBc5Ejj3kKRHjTE+c/eG8guS2iW9Luk31675pqS/SGAGAAAAAAAAPICElUfW2vck/VDSLyXdWHuv70j6I0l/aIzpkZQj6buJygAAAAAAAIAHk9DT1qy1/1jSP/7c032SHknk+wIAAAAAAGB7JPK2NQAAAAAAAOxzlEcAAAAAAACIifIIAAAAAAAAMVEeAQAAAAAAICbKIwAAAAAAAMREeQQAAAAAAICYKI8AAAAAAAAQE+URAAAAAAAAYqI8AgAAAAAAQEyURwAAAAAAAIiJ8ggAAAAAAAAxUR4BAAAAAAAgJsojAAAAAAAAxER5BAAAAAAAgJgojwAAAAAAABAT5REAAAAAAABiojwCAAAAAABATJRHAAAAAAAAiInyCAAAAAAAADFRHgEAAAAAACAmyiMAAAAAAADERHkEAAAAAACAmCiPAAAAAAAAEBPlEQAAAAAAAGKiPAIAAAAAAEBMlEcAAAAAAACIifIIAAAAAAAAMVEeAQAAAAAAICbKIwAAAAAAAMREeQQAAAAAAICYKI8AAAAAAAAQE+URAAAAAAAAYqI8AgAAAAAAQEyURwAAAAAAAIiJ8ggAAAAAAAAxuXY7AAAAAIAHF41GNTAwoMHBQfn9fkWjUXk8HuXk5Ki+vl4+n2+3IwIA9inKIwAAAGAf8/v9amtrU0dHh5anbssuzkqhsGSjktOlXl+Krn3wgSqrq9Xc3KzCwsLdjgwA2GcojwAAAIB9amZmRpcvX9biyJA0d0cZbpfqSouVmZoqh8PIHwxqYHxSAz3t6pkcU++tTj36+Jd04sSJ3Y4OANhHKI8AAACAfWhmZkZ/8ed/rsBgr/Kc0sMPHVNRbs4XrqspLtLyql83BwZ1Y7BH70YiCofDOn369C6kBgDsR5RHAAAAwD4TCAT0s5/9TIGhPpWnePXUQ8fldDpjXp+SnKQzDXXKzUjXGx+3q/UXDmVmZqq6unoHUwMA9itOWwMAAAD2mc7OTi2MDinXEdX5DYqjT6suOqIzdTWy48NqbW2VtTbBSQEABwHlEQAAALCPWGvV3t4uzU7rVG2NXHEWR59oLC+Vz0Y0NzmhsbGxBKUEABwklEcAAADAPjI8PKyF25NKdViV5uduerzD4VB9Wans3LTa2toSkBAAcNBQHgEAAAD7yMjIiOzinGpLimWM2dIc9WUlsgtzGhkZ2eZ0AICDiPIIAAAA2Ef8fr8UDistOXnLc/iSvHJYq1AwqHA4vI3pAAAHEeURAAAAsI9YayVr5XBsbdXRJxwOI1mraDS6TckAAAcV5REAAACwj3g8Hsnp0moguOU5wpGIwlErh8t1dz4AANZBeQQAAADsI3l5eVJKqvrGJrY8R+/YuIwv9e5cAABsgPIIAAAA2Edqamrkzc7V1PKK7swvbGmO9oEhmawcNTY2bnM6AMBBRHkEAAAA7CMul0t1dXVSRrY+7u3f9PixO9OaXgkoKSdP1dXVCUgIADhoKI8AAACAfaa5uVmu3AL1Tc/pek9f3OPmlpb06ocfyxQUq6mpSU6nM4EpAQAHBeURAAAAsM+kp6frqYsXZUoq1No3pF+0dSoYCq07ZmhySn959QMFsgtU3tikU6dO7VBaAMB+59rtAAAAAAA2r6qqSue/cklvvv6a2sZH1Pnqm6otLlJ9WYkyUnxyGKNAKKy+8Ql1DA5rPhSRKShWRUOTLl68KIeDvyMDAOJDeQQAAADsU0ePHlV6erpaW1s1OjiozvkZdb53TTYckqxkHA7JlyqTVaDUvHw1Nzfr+PHjFEcAgE2hPAIAAAD2scLCQr3wwguanZ1Ve3u7BgcH5ff7FYlE5PV6lZNz91S18vJySiMAwJZQHgEAAAAHQFZWls6dO6dz587tdhQAwAHDnx4AAAAAAAAQE+URAAAAAAAAYqI8AgAAAAAAQEyURwAAAAAAAIiJ8ggAAAAAAAAxUR4BAAAAAAAgJsojAAAAAAAAxER5BAAAAAAAgJgojwAAAAAAABAT5REAAAAAAABiojwCAAAAAABATJRHAAAAAAAAiInyCAAAAAAAADFRHgEAAAAAACAmyiMAAAAAAADERHkEAAAAAACAmCiPAAAAAAAAEBPlEQAAAAAAAGKiPAIAAAAAAEBMlEcAAAAAAACIifIIAAAAAAAAMVEeAQAAAAAAICbKIwAAAAAAAMREeQQAAAAAAICYKI8AAAAAAAAQE+URAAAAAAAAYqI8AgAAAAAAQEyURwAAAAAAAIiJ8ggAAAAAAAAxUR4BAAAAAAAgJsojAAAAAAAAxER5BAAAAAAAgJgojwAAAAAAABAT5REAAAAAAABiojwCAAAAAABATJRHAAAAAAAAiInyCAAAAAAAADFRHgEAAAAAACAmyiMAAAAAAADERHkEAAAAAACAmCiPAAAAAAAAEBPlEQAAAAAAAGKiPAIAAAAAAEBMlEcAAAAAAACIifIIAAAAAAAAMVEeAQAAAAAAICbKIwAAAAAAAMREeQQAAAAAAICYKI8AAAAAAAAQE+URAAAAAAAAYqI8AgAAAAAAQEyURwAAAAAAAIiJ8ggAAAAAAAAxJbQ8MsZkGmN+aIzpNMZ0GGPOGmOyjTEvG2O61/7NSmQGAAAAAAAAbF2iVx79saSfWWvrJZ2Q1CHpf5b0qrW2VtKra48BAAAAAACwByWsPDLGZEh6QtJ3JclaG7TWzkn6dUnfW7vse5K+lqgMAAAAAAAAeDCJXHlUKWlK0r83xnxojPl/jDEpkgqsteNr10xIKrjfYGPMt4wxrcaY1qmpqQTGBAAAAAAAQCyJLI9ckk5J+ra19iFJy/rcLWrWWivJ3m+wtfY71toWa21LXl5eAmMCAAAAAAAglkSWRyOSRqy17609/qHulkmTxpgjkrT27+0EZgAAAAAAAMADSFh5ZK2dkDRsjKlbe+qCpHZJfynpm2vPfVPSXyQqAwAAAAAAAB6MK8Hz/4GkPzHGeCT1Sfo93S2sfmCM+ZuSBiX9tQRnAAAAAAAAwBYltDyy1l6X1HKfly4k8n0BAAAAAACwPRK55xEAAAAAAAD2OcojAAAAAAAAxER5BAAAAAAAgJgojwAAAAAAABAT5REAAAAAAABiojwCAAAAAABATJRHAAAAAAAAiInyCAAAAAAAADFRHgEAAAAAACAmyiMAAAAAAADE5NroAmPMSUlfklQkaVXSTUmvWmvnE5wNAAAAAAAAuyzmyiNjzO8YY1ol/W+SsiQNSlqQdFHSG8aY7xpjSnYmJgAAAAAAAHbDeiuPsiU9aa1dvt+LxpgWSQ2SRhIRDAAAAAAAALsvZnlkrf3j9QZaa1u3Pw4AAAAAAAD2ki1tmG2M+V+2OwgAAAAAAAD2nq2etvb725oCAAAAAAAAe1LM29aMMTOxXpKUlpg4AAAAAAAA2EvW2zB7WdIZSZOfe95I6k9YIgAAAAAAAOwZ69229p8klVprI5/7Ckv6wQ7lAwAAAAAAwC5a77S1f7DOa/9TYuIAAAAAAABgL1nvtrV7jDFflfS4JCvpirX2xwlNBQAAAAAAgD1hw9PWjDH/RtLfkdQtqUfS3zbG/OtEBwMAAAAAAMDui2fl0UVJjdZaK0nGmP9X0s2EpgIAAAAAAMCesOHKI909Wa3kU4+PSOpNTBwAAAAAAADsJfGsPEqS1GGMeXft8RlJ7xtjfiRJ1trfSFQ4AAAAAAAA7K54yqN/mvAUAAAAAAAA2JM2LI+sta8aY3Iltaw91WqtvZPYWAAAAAAAANgL4jlt7euSfinpdyT9rqRWY8xfSXQwAAAAAAAA7L54blv7XyU9bK2dlCRjTIGklyT9WSKDAQAAAAAAYPfFc9qa45PiaM3tOMcBAAAAAABgn4tn5dFLxpifSPova49/S9LPExcJAAAAAAAAe0U85dHfk/RXJT2+9vh7kn6YsEQAAAAAAADYM2KWR8aYl6y1l6y1VtIP1r4AAAAAAABwiKy3d1HejqUAAAAAAADAnrTebWsZxpjfiPWitfZHCcgDAACwJdFoVEtLSwoGg3I6nUpKSlJycvJuxwIAANj31i2PJL0gydznNSuJ8ggAAOy6hYUFtbe3q7OzU6v+VUUVlZGRwzhUdKRIjY2NqqyslNPp3O2oAAAA+9J65dGgtfZv7FgSAACATfD7/XrrrbfU19en1ciq/JFVeX0eeb1ehSMRrSyvanFoQUOjg8pIzdSZM2d09OjR3Y4NAACw76xXHt1vxREAAMCuW1pa0k9+8hON3RlTwK6qtLJUtQ21ysnNvndNKBTSYN+Quju6NTo7oldfW9TS0pJOnTq1i8kBAAD2n/XKo9/daLAxxqydxgYAALAjAoGALl++rNE7I/KkuXXh4rNKSU35wnVut1s1ddWqqatWd2ePPnzvQ737/rvyer1qamraheQAAAD703qnrf0bY8wfGGPKPv2kMcZjjHnKGPM9Sd9MbDwAAIDP+uCDDzQ2NSp3qktPPfvUfYujz6utr9HD51o0H5zTlXeuaGFhYQeSAgAAHAzrlUfPSIpI+i/GmDFjTLsxpk9St6TflvSvrLX/3w5kBAAAkCQFg0HdunVLy+FlnfvyY/J6PXGPraypVGlViVZCK+ro6EhgSgAAgIMl5m1r1lq/pH8r6d8aY9ySciWtWmvndiocAADAp3V1dWk5uKy8wlxlZGVsevzRhlqN9I2os7NTp0+flsu13h38AAAAkNZfeXSPtTZkrR2nOAIAALvp1q1bWg2vqLa+Zkvjc/JylJ6drsWVBQ0ODm5zOgAAgIMprvIIAABgL1hYWFAkGlH+kYItz1FQVKBQNKzFxcVtTAYAAHBwUR4BAIB9IxQKySoqj8e95TncbresrILB4DYmAwAAOLg2LI/WTlzL2okwAAAA67m7R5FRKBTa8hzhcFhGRm731gsoAACAwySelUcFkj4wxvzAGPOMMcYkOhQAAMD9pKamymVcmp6a2fIc01PTcjqcSklJ2cZkAAAAB9eG5ZG19h9KqpX0XUn/raRuY8w/M8ZUJzgbAADAZ9TW1irJlayezp4tjV+YW9DUxB2leFJUUVGxveEAAAAOqHhPW7OSJta+wpKyJP3QGPMvEpgNAADgM+rq6pTsTtb48LiWl5Y3Pb67s1vJzmQdPXpUHo8nAQkBAAAOnnj2PPo7xphrkv6FpHckHbPW/veSTkv6eoLzAQAA3OPz+VRdVa0kZ7Leffs9RSKRuMeOj06o71a/kpzJamxsTGBKAACAgyWelUfZkn7DWvu0tfZPrbUhSbLWRiW9kNB0AAAAn3PmzBnlZeRrfmpRb796Ja5T08aGx/TOa+8o1ZWuUw+dUk5Ozg4kBQAAOBjiKY8uS7q3K6UxJt0Yc0aSrLUdiQoGAAAOpmg0+kDjU1NT9eyzzyo/LV9zE/P66Z9d1s3rbVpdWf3MddZajY9O6O1Xr+jKa1eV6khTc0OzHnnkkQd6fwAAgMPG3N3OaJ0LjPlQ0qm1fY9kjHFIarXWntqBfJKklpYW29raulNvBwAAttHi4qLa29vV09OjlZUVRaNRud1uFRYWqrGxUeXl5drKYa4LCwt6+eWXNTE1IX94VUEbVHZelrxer6LRiBbmF+VfCijZ5VOyO1mnT53W6dOnE/AJAQAA9idjzDVrbctG17nimct+qmGy1kaNMfGMAwAAh5jf79dbb72l/r4+KeSXDfqlaETGGAUlDS/MaHigX2mZWTp79qwqKys3NX96erq+/vWva2xsTG1tberr71NoNqSgliQZuY1X+bmFamhoUH19vZKTkxPyOQEAAA66eEqgPmPM35b07bXH/4OkvsRFAgAA+93S0pJefPFFzd+ekAn5VV1eqsajtcrLyZbD4dDqql/d/QNq7+rW4sSwXvr5vM49/iU1Nzdv+r2KiopUVFSklZUVzc3NKRAIyOl0KikpSXl5eVta1QQAAIBfiac8+n1J/1rSP5RkJb0q6VuJDAUAAPavYDCoy5cva35yTDmpyXr6yxeV4vN95prk5CQdb6zXsYY6fdzeqfeuf6x3rryt5ORkVVdXb+l9fT6ffJ97HwAAADy4Dcsja+1tSb+1A1kAAMABcOPGDc1Mjisz2aPnL3xZXq835rXGGJ1oapAxRu99dEPvvPOOKioq5HQ6dzAxAAAA1rNheWSMSZL0NyU1SUr65Hlr7d9IYC4AALAPRaNRdXR0SIEVnTv35LrF0acdb6xXz8CgphcX1NfXp9ra2gQnBQAAQLwccVzzHyUVSnpa0puSSiQtJjIUAADYnwYGBrS8MKeM1BQVFRZsamxjbY0UXFVbW1uC0gEAAGAr4imPaqy1/0jSsrX2e5Kel3QmsbEAAMB+NDo6KoWCqquu2vTYmspyOaIRTU5OKBQKJSAdAAAAtiKe8uiT397mjDHNkjIk5ScuEgAA2K8CgYAUjSo1ZfMbV7tcLiUleSVr784DAACAPSGe09a+Y4zJ0t3T1v5SUqqkf5TQVAAAYF9yOBySubv30VZEo1HJuTYPAAAA9oR1yyNjjEPSgrV2VtJbkja/Bh0AABwaycnJksOpmbn5TY9dWVmVPxCU8abFvdE2AAAAEm/dP+tZa6OS/v4OZQEAAPtcVVWV5E7Srd4+RSKRTY3t7OmV3B5VVFTI6XQmKCEAAAA2K5414a8YY/6eMabUGJP9yVfCkwEAgH2noKBAufn5CkSsegeG4h4XiUTU0dMreZLV1NSUwIQAAADYrHj2PPpv1v79W596zopb2AAAwH00NTXpzckJXb32S+VmZyk7K3Pd6621euPqu1oJRpRVnKvi4uIdSgoAAIB4bLjyyFpbeZ8viiMAAHBfdXV1qqo9qpDTq798+RX1Dw3LWnvfa5eWl/XSm2+rb3RC7oxsXbhwYYfTAgAAYCMbrjwyxvzu/Z631v6H7Y8DAAD2O2OMnnrqKVlr1d/TpVfeeVepSV411NYoLydbTqdTK6ur6u4f0NDYuOROkjcrT88884xycnJ2Oz4AAAA+J57b1h7+1PdJki5I+qUkyiMAAHBfTqdTX/nKV3TzyBHdvHlTC7Mz+qD9lvTJJtrGSG6PHOk5qq6u0enTp5WReEySEAAAIABJREFUkbG7oQEAAHBfG5ZH1to/+PRjY0ympO8nLBEAADgQjDE6duyYmpubNTIyou7ubq2srCgSicjr9aqwsFB1dXVKTk7e7agAAABYRzwrjz5vWVLldgcBAAAHkzFGpaWlKi0t3e0oAAAA2IJ49jz6se6eribd3WC7UdIPEhkKAAAAAAAAe0M8K4/+5ae+D0satNaOJCgPAAAAAAAA9pB4yqMhSePWWr8kGWOSjTEV1tqBhCYDAAAAAADArnPEcc2fSop+6nFk7TkAAAAAAAAccPGURy5rbfCTB2vfexIXCQAAAAAAAHtFPOXRlDHmq588MMb8uqQ7iYsEAAAAAACAvSKePY9+X9KfGGP+z7XHI5J+N3GRAAAAAAAAsFdsWB5Za3slPWqMSV17vJTwVAAAAAAAANgTNrxtzRjzz4wxmdbaJWvtkjEmyxjzf+xEOAAAAAAAAOyuePY8etZaO/fJA2vtrKTnEhcJAAAAAAAAe0U85ZHTGOP95IExJlmSd53rAQAAAAAAcEDEs2H2n0h61Rjz79ce/56k/5C4SAAAAAAAANgr4tkw+58bYz6SdHHtqf/dWvvzxMYCAAAAAADAXhDPyiNZa38m6WeSZIx53Bjzf1lr/1ZCkwEAAAAAAGDXxVUeGWMekvTbkv6apH5JP0pkKAAAAAAAAOwNMcsjY8xR3S2MflvSHUn/VZKx1p7foWwAAAAAAADYZeutPOqU9LakF6y1PZJkjPm7O5IKAAAAAAAAe4Jjndd+Q9K4pNeNMf+3MeaCJLMzsQAAAAAAALAXxCyPrLV/bq39LUn1kl6X9D9KyjfGfNsYc2mnAgIAAAAAAGD3rLfySJJkrV221v5na+2vSSqR9KGkP0p4MgAAAAAAAOy6DcujT7PWzlprv2OtvZCoQAAAAAAAANg71tswGwAA4FCLRqMKBAIKhUJyuVxKSkqSw7Gpv70BAADse5RHAAAAnzM3N6f29nbdunVL/oBfVlZGRh63R7W1tWpsbFRubu5uxwQAANgRlEcAAABrVlZW9Oabb2pgcECrkVWthlfk8rrkcrkUCUcUWgpq/uNZtbW3qehIkc6fP6/09PTdjg0AAJBQlEcAAACSFhYW9OKLL2pidlx+61dZZZlq6quVlZP1q2vmF9TT2avB3gEtDy9p7s/m9NxzzykvL28XkwMAACQWN+0DAIBDz+/366c//anGZkblzfTo+a8/p4fPtXymOJKk9Ix0nTrzkJ7/zeeVUZCuiYUJ/fSnP9XCwsIuJQcAAEg8yiMAAHDovffeexqfHpM3w6snLz2ppOSkda/3eDx6/MLjyirI0NTilK5cubJDSQEAAHYe5REAADjU/H6/urq7tBxe1tknHpXb7Y5rnMPh0NknzyqogAaHBjU/P5/gpAAAALuD8ggAABxqXV1dWgkuq6AoX2kZaZsa6/F6VFZZptXIqtrb2xOUEAAAYHdRHgEAgEOtu7tbq+EVVddVb2l8TX21VsMr6u7u3uZkAAAAewPlEQAAONSWl5cVthHl5OVsaXxWTpaiimp1dVXRaHSb0wEAAOw+yiMAAHCohcNhWVk5Xc4tz+F0OWVlFQ6HtzEZAADA3kB5BAAADjWPxyOHjIKB4JbGR6NRRUJhGZm4N9sGAADYTyiPAADAoZaZmSm3w6PxkfEtjR8fGZfL4VZmZqaMMducDgAAYPdRHgEAgEOtoaFByS6fem71bml8T2ePfK4UNTQ0bHMyAACAvYHyCAAAHGqVlZXKTM3U8tzyplcfzc3O6fb4lHwen+rq6hKUEAAAYHdRHgEAgEPN4XCoublZaZ40vff2+5qfm49r3Mryit5+5YrS3OlqqG+Q1+tNcFIAAIDdQXkEAAAOvRMnTqi28qiSbLJev/yGRoZGZa2Nef3k+G29+pNX5Qy6VF5UrkcffXQH0wIAAOwsV6LfwBjjlNQqadRa+4IxplLS9yXlSLom6XestVs73gQAAGAbOBwOXbx4UeZVo+6+Lr3/xvvypnhVfbRKBcWFcrvdCodDujN5Rz2dPVqeX1GqO1UVpZW6dOmSXK6E/0oFAACwa8x6f1Xbljcw5g8ltUhKXyuPfiDpR9ba7xtj/p2kj6y1315vjpaWFtva2prQnAAAANZa3bhxQ21tbZqem9ZqZEWhaEjWWhlj5DIu+Vw+ZaRmqqGhQQ899JCcTuduxwYAANgSY8w1a23LRtcl9M9kxpgSSc9L+qeS/tDcPb/2KUl/fe2S70n6J5LWLY8AAAB2gjFGx48f17FjxzQ8PKzOzk4tLCwoGAzK7XYrNTVVdXV1qqiokMPB3f8AAOBwSPQa638l6e9LSlt7nCNpzlobXns8Iqn4fgONMd+S9C1JKisrS3BMAACAXzHGqKysjN9BAAAAlMANs40xL0i6ba29tpXx1trvWGtbrLUteXl525wOAAAAAAAA8UjkyqNzkr5qjHlOUpKkdEl/LCnTGONaW31UImk0gRkAAAAAAADwABK28sha+w+stSXW2gpJvyXpNWvtNyS9Luk31y77pqS/SFQGAAAAAAAAPJjd2Onxj3R38+we3d0D6bu7kAEAAAAAAABxSPSG2ZIka+0bkt5Y+75P0iM78b4AAAAAAAB4MJwxCwAAAAAAgJgojwAAAAAAABAT5REAAAAAAABiojwCAAAAAABATJRHAAAAAAAAiInyCAAAAAAAADFRHgEAAAAAACAmyiMAAAAAAADERHkEAAAAAACAmCiPAAAAAAAAEBPlEQAAAAAAAGKiPAIAAAAAAEBMlEcAAAAAAACIifIIAAAAAAAAMVEeAQAAAAAAICbKIwAAAAAAAMREeQQAAAAAAICYKI8AAAAAAAAQE+URAAAAAAAAYqI8AgAAAAAAQEyURwAAAAAAAIiJ8ggAAAAAAAAxUR4BAAAAAAAgJsojAAAAAAAAxER5BAAAAAAAgJgojwAAAAAAABAT5REAAAAAAABiojwCAAAAAABATJRHAAAAAAAAiInyCAAAAAAAADFRHgEAAAAAACCm/7+9uw+y8rrvBP893Q3dNAhJgBAYBOgF0ULICEuWZEt27LzIUuQ468xk4lSSmUwlm8rWzOykNlMz2dRUZmqrtiq7qZ2a2kkmmdRMypmtlCdeJ9kolkZ25NiS7EiOACEQDUKAQLQQr0JG0EDT3Wf/6IuMUD/iRe6+vHw+VRSX5znPfX73Fqef2997znmERwAAAAA0Eh4BAAAA0Eh4BAAAAEAj4REAAAAAjYRHAAAAADQSHgEAAADQSHgEAAAAQCPhEQAAAACNhEcAAAAANBIeAQAAANBIeAQAAABAI+ERAAAAAI2ERwAAAAA0Eh4BAAAA0Eh4BAAAAEAj4REAAAAAjYRHAAAAADQSHgEAAADQSHgEAAAAQCPhEQAAAACNhEcAAAAANBIeAQAAANBIeAQAAABAI+ERAAAAAI2ERwAAAAA0Eh4BAAAA0Eh4BAAAAEAj4REAAAAAjYRHAAAAADTqancBAEy8kZGR7Ny5MwcPHszw8HC6u7uzePHizJ49u92lAQAAFznhEcBlbHh4OOvWrUt/f3+ODw4mqa09Jauffz7Xz5uXVatWZdGiRe0sEwAAuIgJjwAuU0NDQ3niiSeyZ/fupI5m1qxZuXHJkkyZMiWHDx/O1m3bsveN3Xliz558/P77s2LFinaXDAAAXISERwCXoVprnnzyyezZvTvTe6flhz/9qcyfP/9dbe67796sX78hq9esyd9+5zvp7e3NTTfd1KaKAQCAi5UFswEuQ6+//noGdr2Wnu6p+YnPPvKe4ChJurq68pGPrMp9996bjI7mu9/9bmqt4zwbAABwJRMeAVyG+vv7k1pzx4oVmTlz5vu2veOOFbnqqhl5+/D3smvXrkmqEAAAuFQIjwAuM8PDw9m5Y0c6Sklf37Kzti+l5La+vqTWbNu2bRIqBAAALiXCI4DLzLFjx1JrTW9vb6ZNm3ZOx8yZMztJMjg4OJGlAQAAlyDhEcBlpqNj7Ef76OjoOR9zqu2pYwEAAE7xWwLAZWbatGmZMmVKBgcHc+jQoXM65vXXdydJrrrqqoksDQAAuAQJjwAuMx0dHVl6661JKenv33TW9sPDw3l5y5akdKSvr28SKgQAAC4lwiOAy9Dy5cuT0pFNmzdnYGCgsV2tNU8/80yGTp7M3Ouvz5w5cyaxSgAA4FIgPAK4DM2aNSsr77wzo0me+NrX88IL63L8+PF3tdm3b1+e+NrXsnXb9nRNmZoHHnigPcUCAAAXta52FwDAxLjnnnsyMjKSlzZsyPNr1mTN2rWZN+/6TJkyJW+/fSRvvvlmUjoytbsnn3noIaOOAACAcQmPAC5TpZR8/OMfz6JFi7Jx48bs3LEju9/Y29qZ9Eyfkb6+vixfvjwzZsxob7EAAMBFS3gEcJlbuHBhFi5cmCNHjuTQoUMZGRlJd3d35s6dm87OznaXBwAAXOSERwBXiBkzZhhhBAAAnDcLZgMAAADQSHgEAAAAQCPhEQAAAACNhEcAAAAANBIeAQAAANBIeAQAAABAI+ERAAAAAI2ERwAAAAA06mp3AQBcPE6ePJlXXnklu3fvzokTJ9LZ2ZmZM2fm1ltvzZw5c9pdHgAA0AbCIwBy4sSJrF69Oi+//HIGjw9laHg0o6kpSTo7OvLi+vWZP29e7rrrrtxwww3tLhcAAJhEwiOAK9zRo0fz2GOPZc/+gzl2cjTXzb0+y2+6JdOnT8/wyEj27d2T7a9sybbXXs+ePXvzwAP3Z8WKFe0uGwAAmCTCI4Ar2NDQUB5//PHs3ncw3dNn5lOf/FSuuXbWu9rcsGhxVq66K5teWp+X1q/LM9/+drq7u7N06dI2VQ0AAEwmC2YDXMHWrVuXN/YdSPf0mfnRhx55T3B0SldXV+648yP5yEfvzZETI3nmmWcyNDQ0ydUCAADtIDwCuEKNjIxk8+bNOT48mns+dn+6u7vPesyy227P7Ouuz+DxobzyyiuTUCUAANBuwiOAK9Srr76aw0cHM/PqWblu7vXnfNytfctzfHgk/f39E1gdAABwsRAeAVyh9uzZk5MjNTfefPN5Hbdw0eLU0pmDb75p6hoAAFwBhEcAV6ihoaHUWtPTM+28juvo6MjUqd2ptQqPAADgCiA8ArhCdXV1pSQZHh4+72NHhodTUtLV5aadAABwuRMeAVyhrr766nR2lLyxe+C8jjt4YH+Gh4cybVrPOS2yDQAAXNqERwBXqKVLl6Znald273otg0ePnvNxWzZvSndXR5YtW5ZSygRWCAAAXAyERwBXqN7e3tx8002Z0tWRF9Y8f07HHDywP6/t2J7uzo4sX758gisEAAAuBsIjgCvYqlWrctW0qRnYuT3PP/e3GR0dbWx7YP++fOvJr2daV8ny5bdl5syZk1gpAADQLlY6BbiCzZo1Kw/+2I/la1/7el7dujlv7H49S5f15cabl6anpye11uzb80a2bN6U13ftTO/Uztx84+I88MAD7S4dAACYJKXW2u4azuruu++uq1evbncZAJetPXv25Jvf/GbefOt7OTE8mqHh0ZSOjtTR0XR2lHR3daRnalduX748H/vYx9LRYeAqAABc6kopa2qtd5+tnZFHAGTevHn5whe+kF27dmXjxo3ZvXt3hoeHU0pXZsyYkb6+vvT19aW3t7fdpQIAAJNMeARAkqSUkkWLFmXRokVJkpGRkXR0dLijGgAAXOGERwCMq7Ozs90lAAAAFwGLVgAAAADQSHgEAAAAQCPhEQAAAACNhEcAAAAANBIeAQAAANBowsKjUsoNpZRvllL6SykbSyn/vLV9Vinlr0spr7T+vnaiagAAAADgg5nIkUfDSX691ro8yX1J/kkpZXmS30jyjVrr0iTfaP0bAAAAgIvQhIVHtdY3aq1rW4/fTrIpyYIkP5nkj1vN/jjJ/zBRNQAAAADwwUzKmkellCVJViX5bpLra61vtHbtSXJ9wzG/UkpZXUpZvX///skoEwAAAIAzTHh4VEqZkeTPkvxarfXw6ftqrTVJHe+4Wusf1lrvrrXefd111010mQAAAACMY0LDo1LKlIwFR39Sa/3z1ua9pZT5rf3zk+ybyBoAAAAAuHATebe1kuS/JNlUa/13p+16NMk/aj3+R0n+cqJqAAAAAOCD6ZrA574/yS8k2VBKWdfa9ptJfjvJl0spv5RkZ5J/MIE1AAAAAPABTFh4VGv9dpLSsPtHJuq8AAAAAPzgTMrd1gAAAAC4NAmPAAAAAGgkPAIAAACgkfAIAAAAgEbCIwAAAAAaCY8AAAAAaCQ8AgAAAKCR8AgAAACARsIjAAAAABoJjwAAAABoJDwCAAAAoJHwCAAAAIBGwiMAAAAAGgmPAAAAAGgkPAIAAACgkfAIAAAAgEbCIwAAAAAaCY8AAAAAaCQ8AgAAAKCR8AgAAACARsIjAAAAABoJjwAAAABoJDwCAAAAoJHwCAAAAIBGwiMAAAAAGgmPAAAAAGgkPAIAAACgkfAIAAAAgEbCIwAAAAAaCY8AAAAAaCQ8AgAAAKCR8AgAAACARsIjAAAAABoJjwAAAABoJDwCAAAAoJHwCAAAAIBGXe0uALgynThxIq+++moGBwfT0dGRmTNnZvHixens7Gx3aQAAAJxGeARMqrfffjtr1qzJ1q1bMzwy/K59vdN609fXl1WrVmXKlCltqhAAAIDTCY+ASXPgwIE89thjOXb8WGqtWbhwQWbPmZNaa14fGMjBg29m7Qtrs2vXrjzyyCPp6elpd8kAAABXPOERMCmOHDmSxx9/PMeODWbBwgW5/4H7M3PmzO83uPee7NmzJ08/9XT2H9ifJ554Ip/73OfS0WFpNgAAgHbyWxkwKdavX5/BY4OZ/6H5+cxDn3l3cNQyb968fPYnPpvp03uzd++evPrqq22oFAAAgNMJj4AJNzw8nJdffjm11tz3sfvedzRRb29vVt65MrXW9Pf3T2KVAAAAjEd4BEy4gYGBnBg6kblzr8vs2bPP2n7p0qXp7OrK7t27c+zYsUmoEAAAgCbCI2DCnQqAZs2adU7tp0yZkhkzpr/rWAAAANpDeARMuM7OziRj09fO1ejo6LuOBQAAoD3cbQ2YcNdcc01KSgYGBjIyMnLWQOjQoUM5fPjtTJ0yNdOnTx+3zb59+7Jv374MDw9n6tSpWbhw4biLcAMAAPDBCI+AD+zw4cM5cOBARkZGMn369MybN+9di2LPnTs3s2fPzoED+7N9+/YsXbr0fZ9v48b+lFKybNmydHW9+8fUli1bsmHDhhw4cCA19Z3tJSWLFi3KnXfemfnz5/9gXyAAAMAVTHgEXLA33ngjL7zwQl7btSujo2NRTkdJZl51VW677basXLnynVFGK1asyFNPfSt/+51nc+2112bOnDnjPufWrVuzqX9TOkpHli9f/s72WmueeeaZ9G/qT6013T3dWbxkcbq7u3PkyJHs3LEzO3buyK5du/LJT34yfX19k/EWAAAAXPaER5Ng79692bJlS44ePZqenp7ccsstWbBgQUop7S4NLtjmzZvz1NNP58TJkaR05kMfWpgjR4/mtZ2v5rWB3dm7b38GBgby8MMPZ8qUKVm2bFl27dqVbdu35a8e/WqW3748t93W985Us3379qV/Y39eeWVrOkpH7r333nctsL169er0b+pPR0fJffd/fOyObKdNfzt+/HheXPdiXlq/IU899VR6e3uzaNGiSX9fAAAALjfCowk0Ojqav/mbv8m2bdvHJtfUJCV5+eUtWbDgQ3nwwQczderUNlcJ52/37t156umnc2xoJLevWJklS27Mc89+J4cOHshVV12V1Jp1L67Pztdey5QpU/Lwww+nlJIf/uEfTldX19jUs/Ubsv7F9enunppak6GhoZRS0tnRmfvuuy8f/vCH3znf4OBg1q1bl6TmRz/zYBYuXPiemnp6enLvffdm6tSpWbt6bZ577jnhEQAAwA+A8GgCPfvss9m6bXu6uqZk+e235/rrr8/Bgwez8aWXMvD67nzrW9/Kgw8+2O4y4bytXbs2J06OBUd3fHhl/uov/zxHjx7OtN6e3LBoUU4ODWXGjN6sX78hTzzxtdxzzz2ZPXt2Ojs78+lPfzq33357+vv7s3Xr1pwcGrsD2/Te6Vm2bFluu+229yx8vXnz5oyMjmTRksXjBkenW3nnymzq35RDh97M7t2786EPfWjC3gcAAIArgfBoghw7diz9/ZuSmjzy2c/muuuuS5IsXrw4t9xyS/7sK1/Jq6/uyFtvvZVrrrmmzdXCuXvrrbcy8PrrKR1duePDK7P1lZdz5MjbmT1ndh7+8UfeGU1356qP5O3Dh7Pl5S157rnn8sgjj7zzHHPnzs3cuXPzyU9+8p0RR1OnTm2cyrlt27bUWnPb8tvOWl9HR0eW9S3Li2vXZdu2bcIjAACAD6jj7E24EAMDAxkZHc2ChQvfCY5OmTlzZm6++ebUJDt27GhLfXCh9u/fn5HRmgULb8jUqVOzc8eOlFKzcuXKd03DvHbWrNxx58p0T+vOxo0bx32ujo6O9PT0pLu7+33XABscHExS37UG0vuZNXtWasZCXAAAAD4Y4dEEOXnyZJJkxowZ4+7vnT49STI8PDxpNcEPwsjISJJkypQpSZKTwydTkkyf/t7/61fNmJmOjo4MDQ19oHN2dIz9qBodHT2n9qfanToOAACAC+c3qwlyzTXXpCTZtWvXuL/w7nrttZRWO7iU9Pb2pqMkBw8eSJJcc821qUle27nzPW23vrIlJ4eGMmfOnA90zrF+UrJr165zaj+wayCllFx99dUf6LwAAAAIjybM/Pnzc801V2fw6NE88/TT74y8GB4eznPPPZeDBw5k2rSe3HjjjW2uFM7PggULMr23N28dejP79u1N323LU0dLXtqwPi+uW5ejR4/mrUOH8vRT38qWLS9ndGQ0DzzwwAc6Z19fX0op2bRxU2qt79v2+PHj2b5te0pKli1b9oHOCwAAgAWzJ0wpJT/0Qz+Ur371sbyyZUu2b9+ea6+9Nt/73vdycmgoHR0ln/jEJ9LZ2dnuUuG8dHZ2pq+vL8+vWZvvPPNUHnzokaxcdVdefGFN1q5ZnRfWrE6S7NmzJ3v37M1dq+5MX1/fBzrnTTfdlGeffTZvHnwzf/fdv8u99907brvh4eF848lvpI6MZsmSG99z1zYAAADOXznbt/gXg7vvvruuXr263WVckH379uX555/PwMDr72y7/vq5+ehHP5oFCxa0sTK4cCdPnsyjjz6aPXv3p3RNzdJb+9I9dWq2bduanTtezd49b6SrqzNLFt+Qv/dTP5UlS5Z84HMODAzk8ccfz8joSBYuWpgVd6x4pw+NjIxk+/bt2fDihhx681BmTJ+Rz3/+841rjgEAAJCUUtbUWu8+azvh0eQ4cuRIBgcH09PTYzQEl4Xjx4/nySefzMDA6xkeqRmpNTVjc2G7Okum9XTn05/+9A8kODpl586defLJJ3Ny+GRqrenu6U53d3cGBwczfHJ4bJ2jmVfn4Ycftp4YAADAWQiPgEmxd+/e9Pf3Z//+/RkZGUlvb2+WLl2apUuXvnNHth+kwcHBbN68OZs2bcqRI0fe2T5nzpzcfvvtueWWW9LVZUYuAADA2QiPgMtarTVHjx7N8PBwuru7M23atHaXBAAAcEk51/DI1/PAJamUYk0jAACASdDR7gIAAAAAuHgJjwAAAABoJDwCAAAAoJHwCAAAAIBGwiMAAAAAGgmPAAAAAGgkPAIAAACgkfAIAAAAgEbCIwAAAAAaCY8AAAAAaCQ8AgAAAKCR8AgAAACARsIjAAAAABoJjwAAAABoJDwCAAAAoJHwCAAAAIBGwiMAAAAAGgmPAAAAAGgkPAIAAACgkfAIAAAAgEbCIwAAAAAaCY8AAAAAaFRqre2u4axKKfuT7Gx3HVewOUkOtLsIuETpP3Dh9B+4cPoPXDj9hyvJ4lrrdWdrdEmER7RXKWV1rfXudtcBlyL9By6c/gMXTv+BC6f/wHuZtgYAAABAI+ERAAAAAI2ER5yLP2x3AXAJ03/gwuk/cOH0H7hw+g+cwZpHAAAAADQy8ggAAACARsIjAAAAABoJj0iSlFIeKqW8XErZWkr5jXH2/2IpZX8pZV3rzy+3o064GJVS/qiUsq+U8lLD/lJK+b9b/Wt9KeUjk10jXKzOof98qpTyvdOuP7812TXCxaqUckMp5ZullP5SysZSyj8fp41rEIzjHPuPaxC0dLW7ANqvlNKZ5PeS/FiSgSTPl1IerbX2n9H0T2ut/3TSC4SL3xeT/G6S/9qw/+EkS1t/7k3y+62/gbP3nyR5ptb62ckpBy4pw0l+vda6tpRyVZI1pZS/PuMznGsQjO9c+k/iGgRJjDxizD1JttZat9dah5L8tyQ/2eaa4JJRa306yZvv0+Qnk/zXOua5JNeUUuZPTnVwcTuH/gM0qLW+UWtd23r8dpJNSRac0cw1CMZxjv0HaBEekYz9kNx12r8HMv4Pzr/XGu78lVLKDZNTGlwWzrWPAeP7WCnlxVLKfy+l3N7uYuBiVEpZkmRVku+escs1CM7iffpP4hoESYRHnLu/SrKk1vrhJH+d5I/bXA8AV4a1SRbXWlcm+Q9J/r821wMXnVLKjCR/luTXaq2H210PXErO0n9cg6BFeESSvJ7k9JFEC1vb3lFrPVhrPdH6539Octck1QaXg7P2MWB8tdbDtdYjrcePJ5lSSpnT5rLgolFKmZKxX3z/pNb65+M0cQ2CBmfrP65B8H3CI5Lk+SRLSyk3llKmJvlCkkdPb3DG3PjPZWxOMHBuHk3yD1t3vLkvyfdqrW+0uyi4FJRS5pVSSuvxPRn77HKwvVXBxaHVN/5Lkk211n/X0Mw1CMZxLv3HNQi+z93WSK11uJTyT5N8LUlnkj+qtW4spfxvSVbXWh9N8j+XUj6XsbsSvJnkF9tdpWepAAAGQ0lEQVRWMFxkSilfSvKpJHNKKQNJ/k2SKUlSa/2DJI8n+fEkW5MMJvnH7akULj7n0H/+fpL/qZQynORYki/UWmubyoWLzf1JfiHJhlLKuta230yyKHENgrM4l/7jGgQtxf99AAAAAJqYtgYAAABAI+ERAAAAAI2ERwAAAAA0Eh4BAAAA0Eh4BAAAAHAJKaX8USllXynlpXNou6iU8s1SygullPWllB8/3/MJjwCAS0YpZV4p5b+VUraVUtaUUh4vpdzaplp+8wKP+0op5abW4x2llGfO2L/uzA+CpZR/X0p5vZTSccb2h0spq0sp/a0PhP/X+5z3jlLKFy+kZgDgovPFJA+dY9t/neTLtdZVSb6Q5D+e78mERwDAJaGUUpL8RZJv1VpvrrXeleR/TXJ9m0o67/ColHJ7ks5a6/bTNl9VSrmhtf+2cY7pSPL5JLuS/NBp21ck+d0kP19rXZ7k7iRbm85da92QZGEpZdH51g0AXFxqrU8nefP0baWUm0spT7S+YHumlNJ3qnmSma3HVyfZfb7nEx4BAJeKTyc5WWv9g1Mbaq0v1lqfKWN+p5TyUillQynlZ5KklPKpUspTpZS/LKVsL6X8dinl50opf9dqd3Or3RdLKX/QGsWzpZTy2db2Xyyl/O6p85VSvtp6zt9OMq01SuhPWvt+vvW860op/6mU0jnOa/i5JH95xrYvJ/mZ1uOfTfKlM/Z/KsnGJL/f2n/Kv0zyv9daN7fei5Fa6++3avnp1nvxYinl6dOO+auMfeMIAFx+/jDJP2t9wfYv8v0RRv82yc+XUgaSPJ7kn53vEwuPAIBLxYokaxr2/VSSO5OsTPKjSX6nlDK/tW9lkl9NcluSX0hya631niT/Oe/+8LQkyT1JHknyB6WUnqZCaq2/keRYrfXOWuvPtUYM/UyS+2utdyYZyVhQdKb7x3kNf9aqP0l+ImMBz+lOBUp/keSRUsqU1vb3ez9+K8lnaq0rk3zutO2rk3yi6XUBAJemUsqMJB9P8v+WUtYl+U9JTn0W+tkkX6y1Lkzy40n+nzOnwp9N1w+yWACANnkgyZdqrSNJ9pZSnkry0SSHkzxfa30jSUop25J8vXXMhoyNZjrly7XW0SSvlFK2J+nLufuRJHcleX5sdl2mJdk3Trv5Sfafse1gkkOllC8k2ZRk8NSOUsrUjH3I+19qrW+XUr6b5DNJvnqWer6T5IullC8n+fPTtu9L8qFzfVEAwCWjI8lbrS+xzvRLaa2PVGt9tvUF2ZyM/1ml8ckBAC4FGzMW0JyvE6c9Hj3t36N59xdp9YzjapLhvPvzUtNopJLkj1sjke6stS6rtf7bcdoda3iOP03ye3nvlLXPJLkmyYZSyo6MhWSnpq41vh+11l/N2OKYNyRZU0qZfVr9xxpeAwBwiaq1Hk7yainlp5OxtSJLKStbu1/L2Bddp9ZX7Ml7v8x6X8IjAOBS8TdJukspv3JqQynlw6WUTyR5JsnPlFI6SynXJflkkr87z+f/6VJKR2sdpJuSvJxkR5I7W9tvyNi0tlNOnjaF7BtJ/n4pZW6rrlmllMXjnGNTklvG2f4XSf7PJF87Y/vPJvnlWuuSWuuSJDcm+bFSSm+S30nym6fuNteq8Vdbj2+utX631vpbGftweEPr+W5NctZb+gIAF7dSypeSPJtkWSlloJTySxmbMv9LpZQXM/Yl00+2mv96kv+xtf1LSX6x1nrml2bvy7Q1AOCSUGutpZTPJ/n3pZR/leR4xsKdX0vy7SQfS/JixkYM/cta657T7jJyLl7LWOA0M8mv1lqPl1K+k+TVJP0ZC37Wntb+D5OsL6Wsba179K+TfL21hsDJJP8kyc4zzvFYxhbAfvKM1/Z2kv8jSVrT3tIKiB7K2HpNp9odLaV8O8lP1Fr/tJTya0m+1Gpb8/3pbL9TSlmasRFR32i9L8nYNL3HzuM9AQAuQrXWn23Y9dA4bfsztu7iBSvnGTYBAFx2SilfTPLVWutXJvg805J8M2MLa49M5LnGOXd3kqeSPFBrHZ7McwMAlzbT1gAAJkmt9ViSf5NkQRtOvyjJbwiOAIDzZeQRAAAAAI2MPAIAAACgkfAIAAAAgEbCIwAAAAAaCY8AAAAAaCQ8AgAAAKDR/w+tnEtFZCtpCgAAAABJRU5ErkJggg==\n",
+      "text/html": [
+       "<div>\n",
+       "<style>\n",
+       "    .dataframe thead tr:only-child th {\n",
+       "        text-align: right;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe thead th {\n",
+       "        text-align: left;\n",
+       "    }\n",
+       "\n",
+       "    .dataframe tbody tr th {\n",
+       "        vertical-align: top;\n",
+       "    }\n",
+       "</style>\n",
+       "<table border=\"1\" class=\"dataframe\">\n",
+       "  <thead>\n",
+       "    <tr style=\"text-align: right;\">\n",
+       "      <th></th>\n",
+       "      <th>Name</th>\n",
+       "      <th>Shape</th>\n",
+       "      <th>NNZ (dense)</th>\n",
+       "      <th>NNZ (sparse)</th>\n",
+       "      <th>Cols (%)</th>\n",
+       "      <th>Rows (%)</th>\n",
+       "      <th>Ch (%)</th>\n",
+       "      <th>2D (%)</th>\n",
+       "      <th>3D (%)</th>\n",
+       "      <th>Fine (%)</th>\n",
+       "      <th>Std</th>\n",
+       "      <th>Mean</th>\n",
+       "      <th>Abs-Mean</th>\n",
+       "    </tr>\n",
+       "  </thead>\n",
+       "  <tbody>\n",
+       "    <tr>\n",
+       "      <th>0</th>\n",
+       "      <td>features.module.0.weight</td>\n",
+       "      <td>(42, 3, 3, 3)</td>\n",
+       "      <td>1134</td>\n",
+       "      <td>1134</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.29</td>\n",
+       "      <td>-6.32e-03</td>\n",
+       "      <td>2.22e-01</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>1</th>\n",
+       "      <td>features.module.2.weight</td>\n",
+       "      <td>(42, 42, 3, 3)</td>\n",
+       "      <td>15876</td>\n",
+       "      <td>15876</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.10</td>\n",
+       "      <td>-1.46e-02</td>\n",
+       "      <td>7.41e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>2</th>\n",
+       "      <td>features.module.5.weight</td>\n",
+       "      <td>(104, 42, 3, 3)</td>\n",
+       "      <td>39312</td>\n",
+       "      <td>39312</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.08</td>\n",
+       "      <td>-9.57e-03</td>\n",
+       "      <td>6.07e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>3</th>\n",
+       "      <td>features.module.7.weight</td>\n",
+       "      <td>(93, 104, 3, 3)</td>\n",
+       "      <td>87048</td>\n",
+       "      <td>87048</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.06</td>\n",
+       "      <td>-1.11e-02</td>\n",
+       "      <td>4.74e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>4</th>\n",
+       "      <td>features.module.10.weight</td>\n",
+       "      <td>(244, 93, 3, 3)</td>\n",
+       "      <td>204228</td>\n",
+       "      <td>204228</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.05</td>\n",
+       "      <td>-6.46e-03</td>\n",
+       "      <td>3.85e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>5</th>\n",
+       "      <td>features.module.12.weight</td>\n",
+       "      <td>(161, 244, 3, 3)</td>\n",
+       "      <td>353556</td>\n",
+       "      <td>353556</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.04</td>\n",
+       "      <td>-5.12e-03</td>\n",
+       "      <td>3.07e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>6</th>\n",
+       "      <td>features.module.14.weight</td>\n",
+       "      <td>(237, 161, 3, 3)</td>\n",
+       "      <td>343413</td>\n",
+       "      <td>343413</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.04</td>\n",
+       "      <td>-6.94e-03</td>\n",
+       "      <td>2.89e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>7</th>\n",
+       "      <td>features.module.17.weight</td>\n",
+       "      <td>(341, 237, 3, 3)</td>\n",
+       "      <td>727353</td>\n",
+       "      <td>727353</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.02</td>\n",
+       "      <td>-2.80e-03</td>\n",
+       "      <td>1.83e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>8</th>\n",
+       "      <td>features.module.19.weight</td>\n",
+       "      <td>(311, 341, 3, 3)</td>\n",
+       "      <td>954459</td>\n",
+       "      <td>954459</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.02</td>\n",
+       "      <td>-1.36e-03</td>\n",
+       "      <td>1.24e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>9</th>\n",
+       "      <td>features.module.21.weight</td>\n",
+       "      <td>(369, 311, 3, 3)</td>\n",
+       "      <td>1032831</td>\n",
+       "      <td>1032831</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>-1.17e-03</td>\n",
+       "      <td>1.05e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>10</th>\n",
+       "      <td>features.module.24.weight</td>\n",
+       "      <td>(390, 369, 3, 3)</td>\n",
+       "      <td>1295190</td>\n",
+       "      <td>1295190</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>-1.11e-04</td>\n",
+       "      <td>8.75e-03</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>11</th>\n",
+       "      <td>features.module.26.weight</td>\n",
+       "      <td>(116, 390, 3, 3)</td>\n",
+       "      <td>407160</td>\n",
+       "      <td>407160</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>3.76e-05</td>\n",
+       "      <td>8.66e-03</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>12</th>\n",
+       "      <td>features.module.28.weight</td>\n",
+       "      <td>(303, 116, 3, 3)</td>\n",
+       "      <td>316332</td>\n",
+       "      <td>316332</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.01</td>\n",
+       "      <td>-1.54e-04</td>\n",
+       "      <td>8.87e-03</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>13</th>\n",
+       "      <td>classifier.weight</td>\n",
+       "      <td>(10, 303)</td>\n",
+       "      <td>3030</td>\n",
+       "      <td>3030</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.08</td>\n",
+       "      <td>-1.55e-05</td>\n",
+       "      <td>6.21e-02</td>\n",
+       "    </tr>\n",
+       "    <tr>\n",
+       "      <th>14</th>\n",
+       "      <td>Total sparsity:</td>\n",
+       "      <td>-</td>\n",
+       "      <td>5780922</td>\n",
+       "      <td>5780922</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.0</td>\n",
+       "      <td>0.00</td>\n",
+       "      <td>0.00e+00</td>\n",
+       "      <td>0.00e+00</td>\n",
+       "    </tr>\n",
+       "  </tbody>\n",
+       "</table>\n",
+       "</div>"
+      ],
       "text/plain": [
-       "<matplotlib.figure.Figure at 0x7ff847b415f8>"
+       "                         Name             Shape NNZ (dense) NNZ (sparse)  \\\n",
+       "0    features.module.0.weight     (42, 3, 3, 3)        1134         1134   \n",
+       "1    features.module.2.weight    (42, 42, 3, 3)       15876        15876   \n",
+       "2    features.module.5.weight   (104, 42, 3, 3)       39312        39312   \n",
+       "3    features.module.7.weight   (93, 104, 3, 3)       87048        87048   \n",
+       "4   features.module.10.weight   (244, 93, 3, 3)      204228       204228   \n",
+       "5   features.module.12.weight  (161, 244, 3, 3)      353556       353556   \n",
+       "6   features.module.14.weight  (237, 161, 3, 3)      343413       343413   \n",
+       "7   features.module.17.weight  (341, 237, 3, 3)      727353       727353   \n",
+       "8   features.module.19.weight  (311, 341, 3, 3)      954459       954459   \n",
+       "9   features.module.21.weight  (369, 311, 3, 3)     1032831      1032831   \n",
+       "10  features.module.24.weight  (390, 369, 3, 3)     1295190      1295190   \n",
+       "11  features.module.26.weight  (116, 390, 3, 3)      407160       407160   \n",
+       "12  features.module.28.weight  (303, 116, 3, 3)      316332       316332   \n",
+       "13          classifier.weight         (10, 303)        3030         3030   \n",
+       "14            Total sparsity:                 -     5780922      5780922   \n",
+       "\n",
+       "   Cols (%) Rows (%)  Ch (%)  2D (%)  3D (%)  Fine (%)   Std      Mean  \\\n",
+       "0         0        0     0.0     0.0     0.0       0.0  0.29 -6.32e-03   \n",
+       "1         0        0     0.0     0.0     0.0       0.0  0.10 -1.46e-02   \n",
+       "2         0        0     0.0     0.0     0.0       0.0  0.08 -9.57e-03   \n",
+       "3         0        0     0.0     0.0     0.0       0.0  0.06 -1.11e-02   \n",
+       "4         0        0     0.0     0.0     0.0       0.0  0.05 -6.46e-03   \n",
+       "5         0        0     0.0     0.0     0.0       0.0  0.04 -5.12e-03   \n",
+       "6         0        0     0.0     0.0     0.0       0.0  0.04 -6.94e-03   \n",
+       "7         0        0     0.0     0.0     0.0       0.0  0.02 -2.80e-03   \n",
+       "8         0        0     0.0     0.0     0.0       0.0  0.02 -1.36e-03   \n",
+       "9         0        0     0.0     0.0     0.0       0.0  0.01 -1.17e-03   \n",
+       "10        0        0     0.0     0.0     0.0       0.0  0.01 -1.11e-04   \n",
+       "11        0        0     0.0     0.0     0.0       0.0  0.01  3.76e-05   \n",
+       "12        0        0     0.0     0.0     0.0       0.0  0.01 -1.54e-04   \n",
+       "13        0        0     0.0     0.0     0.0       0.0  0.08 -1.55e-05   \n",
+       "14        0        0     0.0     0.0     0.0       0.0  0.00  0.00e+00   \n",
+       "\n",
+       "    Abs-Mean  \n",
+       "0   2.22e-01  \n",
+       "1   7.41e-02  \n",
+       "2   6.07e-02  \n",
+       "3   4.74e-02  \n",
+       "4   3.85e-02  \n",
+       "5   3.07e-02  \n",
+       "6   2.89e-02  \n",
+       "7   1.83e-02  \n",
+       "8   1.24e-02  \n",
+       "9   1.05e-02  \n",
+       "10  8.75e-03  \n",
+       "11  8.66e-03  \n",
+       "12  8.87e-03  \n",
+       "13  6.21e-02  \n",
+       "14  0.00e+00  "
       ]
      },
+     "execution_count": 102,
      "metadata": {},
-     "output_type": "display_data"
+     "output_type": "execute_result"
     }
    ],
    "source": [
-    "def add_results(df, cmap):\n",
-    "    # create data\n",
-    "    x = df['MACs'].tolist()\n",
-    "    y = df['Top1'].tolist()\n",
-    "    z = df['NNZ'].tolist()\n",
-    "    z = [n/30000 for n in z]\n",
-    "    plt.scatter(x, y, s=z, c=x, cmap=cmap, alpha=0.4, edgecolors=\"black\", linewidth=2)\n",
-    "\n",
-    "# Change color with c and alpha. I map the color to the X axis value.\n",
-    "plt.figure(figsize=(20,10))\n",
-    "add_results(df, cmap=\"Blues\")\n",
-    "add_results(df2, cmap=\"Reds\")\n",
-    "add_results(df3, cmap=\"Greens\")\n",
+    "checkpoint_file = \"../classifier_compression/logs/master___2018.07.25-205658/BEST_adc_episode_008_checkpoint.pth.tar\" \n",
     "\n",
-    "# Add titles (main and on axis)\n",
-    "plt.xlabel(\"Compute (MACs)\")\n",
-    "plt.ylabel(\"Accuracy (Top1)\")\n",
-    "plt.title(\"Network Space\")\n",
-    "plt.show()"
+    "model = models.create_model(pretrained=False, dataset=dataset, arch=arch)\n",
+    "apputils.load_checkpoint(model, checkpoint_file);\n",
+    "distiller.weights_sparsity_summary(model)"
    ]
   },
   {
diff --git a/examples/automated_deep_compression/presets/ADC_DDPG.py b/examples/automated_deep_compression/presets/ADC_DDPG.py
index 2d6ce113f2add532b0b5f24325018c92f58b1ebf..928aa7b785ff6d0afbdfbc6a7d3cd73d1d004f46 100755
--- a/examples/automated_deep_compression/presets/ADC_DDPG.py
+++ b/examples/automated_deep_compression/presets/ADC_DDPG.py
@@ -1,15 +1,15 @@
-from agents.ddpg_agent import DDPGAgentParameters
-from graph_managers.basic_rl_graph_manager import BasicRLGraphManager
-from graph_managers.graph_manager import ScheduleParameters
-from base_parameters import VisualizationParameters
-from core_types import EnvironmentEpisodes, EnvironmentSteps
-from environments.gym_environment import MujocoInputFilter, GymEnvironmentParameters, MujocoOutputFilter
-from exploration_policies.additive_noise import AdditiveNoiseParameters
-from exploration_policies.truncated_normal import TruncatedNormalParameters
-from schedules import ConstantSchedule, PieceWiseSchedule, ExponentialSchedule
-from memories.memory import MemoryGranularity
-from base_parameters import EmbedderScheme
-from architectures.tensorflow_components.architecture import Dense
+from rl_coach.agents.ddpg_agent import DDPGAgentParameters
+from rl_coach.graph_managers.basic_rl_graph_manager import BasicRLGraphManager
+from rl_coach.graph_managers.graph_manager import ScheduleParameters
+from rl_coach.base_parameters import VisualizationParameters
+from rl_coach.core_types import EnvironmentEpisodes, EnvironmentSteps
+from rl_coach.environments.gym_environment import MujocoInputFilter, GymEnvironmentParameters, MujocoOutputFilter
+from rl_coach.exploration_policies.additive_noise import AdditiveNoiseParameters
+from rl_coach.exploration_policies.truncated_normal import TruncatedNormalParameters
+from rl_coach.schedules import ConstantSchedule, PieceWiseSchedule, ExponentialSchedule
+from rl_coach.memories.memory import MemoryGranularity
+from rl_coach.base_parameters import EmbedderScheme
+from rl_coach.architectures.tensorflow_components.architecture import Dense
 
 
 ####################
@@ -30,6 +30,7 @@ agent_params.network_wrappers['actor'].middleware_parameters.scheme = [Dense([30
 agent_params.network_wrappers['critic'].input_embedders_parameters['observation'].scheme = [Dense([300])]
 agent_params.network_wrappers['critic'].middleware_parameters.scheme = [Dense([300])]
 agent_params.network_wrappers['critic'].input_embedders_parameters['action'].scheme = EmbedderScheme.Empty
+agent_params.network_wrappers['actor'].heads_parameters[0].activation_function = 'sigmoid'
 #agent_params.network_wrappers['critic'].clip_gradients = 100
 #agent_params.network_wrappers['actor'].clip_gradients = 100
 
@@ -55,6 +56,7 @@ env_params.level = '../automated_deep_compression/ADC.py:CNNEnvironment'
 
 
 vis_params = VisualizationParameters()
+vis_params.dump_parameters_documentation = False
 
 graph_manager = BasicRLGraphManager(agent_params=agent_params, env_params=env_params,
                                     schedule_params=schedule_params, vis_params=vis_params)