├── README.md
├── data_functions.py
├── models.py
└── es_rnn_example.ipynb


/README.md:
--------------------------------------------------------------------------------
 1 | # ES-RNN-Pytorch
 2 | 
 3 | # Overview
 4 | This is a work in progress Pytorch implementation of the recently proposed ES-RNN by Slawek Smyl, winner of the M4 competition.
 5 | 
 6 | ## Progress: 
 7 | 
 8 | - [x] Implement Single Series Model based on Additive season and level in Pytorch 
 9 | - [x] Implement toy datset, and Pytorch data functions
10 | - [x] Test and Debug SSM on Toy Dataset
11 | - [ ] Implemente multiplicative season and trend
12 | - [ ] Implement Multi Series Model with shared Trend LSTM
13 | - [ ] Implement Multi Series Multi Output Model 
14 | - [ ] Compare performance on benchmark dataset
15 | 


--------------------------------------------------------------------------------
/data_functions.py:
--------------------------------------------------------------------------------
 1 | #Dataset with random shuffeling: We have to check, that seasonality gets shifted apropriatly 
 2 | 
 3 | import numpy as np 
 4 | import pandas as pd
 5 | 
 6 | import random
 7 | from torch.utils.data import Dataset,sampler,DataLoader
 8 | import torch
 9 | import torch.nn as nn
10 | from tqdm import tqdm
11 | 
12 | 
13 | class sequence_labeling_dataset(Dataset):
14 |     
15 |     def __init__(self, input,max_size=100,sequence_labeling=True,seasonality=12,out_preds=12):     
16 |         
17 |         self.data=input
18 |         self.max_size=max_size
19 |         self.sequence_labeling=sequence_labeling
20 |         self.seasonality=seasonality
21 |         self.out_preds=out_preds
22 |         
23 |     def __len__(self):
24 |         
25 |         return int(10000)
26 |     
27 |     def __getitem__(self, index):
28 |         
29 |         data_i=self.data
30 |         
31 |         #we randomly shift the inputs to create more data
32 |         if len(data_i)>self.max_size:
33 |             max_rand_int=len(data_i)-self.max_size
34 |             #take a random start integer
35 |             start_int=random.randint(0,max_rand_int)
36 |             data_i=data_i[start_int:(start_int+self.max_size)]
37 |         else:
38 |             start_int=0
39 | 
40 |         
41 |         inp=np.array(data_i[:-self.out_preds])
42 |         
43 |         
44 |         if self.sequence_labeling==True:
45 |             #in case of sequence labeling, we shift the input by the range to output
46 |             out=np.array(data_i[self.out_preds:])
47 |         else:
48 |             #in case of sequnec classification we return only the last n elements we
49 |             #need in the forecast
50 |             out=np.array(data_i[-self.out_preds:])
51 |             
52 |         #This defines, how much we have to shift the season 
53 |         shift_steps=start_int%self.seasonality
54 |         
55 |         return inp, out,shift_steps
56 |     


--------------------------------------------------------------------------------
/models.py:
--------------------------------------------------------------------------------
  1 | import numpy as np 
  2 | import matplotlib.pyplot as plt
  3 | 
  4 | import random
  5 | from torch.utils.data import Dataset,sampler,DataLoader
  6 | import torch
  7 | import torch.nn as nn
  8 | import pandas as pd
  9 | from tqdm import tqdm
 10 | 
 11 | class holt_winters_no_trend(torch.nn.Module):
 12 |     
 13 |     def __init__(self,init_a=0.1,init_g=0.1,slen=12):
 14 |       
 15 |         super(holt_winters_no_trend, self).__init__()
 16 |         
 17 |         #Smoothing parameters
 18 |         self.alpha=torch.nn.Parameter(torch.tensor(init_a))
 19 |         self.gamma=torch.nn.Parameter(torch.tensor(init_g))
 20 |         
 21 |         #init parameters
 22 |         self.init_season=torch.nn.Parameter(torch.tensor(np.random.random(size=slen)))
 23 |         
 24 |         #season legnth used to pick appropriate past season step 
 25 |         self.slen=slen
 26 |         
 27 |         #Sigmoid used to norm the params to be betweeen 0 and 1 if needed 
 28 |         self.sig=nn.Sigmoid()
 29 |         
 30 |     def forward(self,series ,series_shifts,n_preds=12,rv=False):
 31 |         
 32 |         #Get Batch size
 33 |         batch_size=series.shape[0]
 34 |         
 35 |         #Get the initial seasonality parameter
 36 |         init_season_batch=self.init_season.repeat(batch_size).view(batch_size,-1)
 37 |         
 38 |         #We use roll to Allow for our random input shifts.
 39 |         seasonals=torch.stack([torch.roll(j,int(rol)) for j,rol in zip(init_season_batch,series_shifts)]).float()
 40 |         
 41 |         #It has to be a list such that we dont need inplace tensor changes. 
 42 |         seasonals=list(torch.split(seasonals,1,dim=1))
 43 |         seasonals=[x.squeeze() for x in seasonals]
 44 |         
 45 |         #Now We loop over the input in each forward step
 46 |         result = []
 47 |         
 48 |         #rv can be used for decomposing a series./normalizing in case of ES-RNN
 49 |         if rv==True:
 50 |             value_list=[]
 51 |             season_list=[]
 52 |         
 53 | 
 54 |         for i in range(series.shape[1]+n_preds):
 55 |             
 56 |             #0th step we init the parameter 
 57 |             if i == 0: 
 58 | 
 59 |                 smooth = series[:,0]
 60 |                 value_list.append(smooth)
 61 |                 season_list.append(seasonals[i%self.slen])                
 62 |                 result.append(series[:,0])
 63 | 
 64 |                 continue
 65 | 
 66 |             #smoothing
 67 |             #its smaller here, so smoothing is only for one less than the input? 
 68 |             if i <series.shape[1]:
 69 | 
 70 |                 val = series[:,i]
 71 | 
 72 |                 last_smooth, smooth = smooth, self.sig(self.alpha)*(val-seasonals[i%self.slen]) + (1-self.sig(self.alpha))*(smooth)
 73 | 
 74 |                 seasonals[i%self.slen] = self.sig(self.gamma)*(val-smooth) + (1-self.sig(self.gamma))*seasonals[i%self.slen]
 75 |                 
 76 |                 #we store values, used for normaizing in ES RNN 
 77 |                 if rv==True:
 78 |                     value_list.append(smooth)
 79 |                     season_list.append(seasonals[i%self.slen])
 80 | 
 81 |                 result.append(smooth+seasonals[i%self.slen])
 82 |             
 83 |             #forecasting would jsut select last smoothed value and the appropriate seasonal, we will do this seperately 
 84 |             #in the ES RNN implementation
 85 |             else:
 86 |     
 87 |                 m = i - series.shape[1] + 1
 88 | 
 89 |                 result.append((smooth ) + seasonals[i%self.slen])
 90 |                 
 91 |                 if rv==True:
 92 |                     value_list.append(smooth)
 93 |                     season_list.append(seasonals[i%self.slen])
 94 |                 
 95 |             #If we want to return the actual, smoothed values or only the forecast
 96 |         if rv==False:
 97 |             return torch.stack(result,dim=1)[:,-n_preds:]
 98 |         else:
 99 |             return torch.stack(result,dim=1),torch.stack(value_list,dim=1),torch.stack(season_list,dim=1)
100 |                 
101 |             
102 | class es_rnn(torch.nn.Module):
103 |     
104 |     def __init__(self,hidden_size=16,slen=12,pred_len=12):
105 |       
106 |         super(es_rnn, self).__init__()
107 |         
108 |         self.hw=holt_winters_no_trend(init_a=0.1,init_g=0.1)
109 |         self.RNN=nn.GRU(hidden_size=hidden_size,input_size=1,batch_first=True)
110 |         self.lin=nn.Linear(hidden_size,pred_len)
111 |         self.pred_len=pred_len
112 |         self.slen=slen
113 |         
114 |     def forward(self,series,shifts):
115 |         
116 |         #Get Batch size
117 |         batch_size=series.shape[0]
118 |         result,smoothed_value,smoothed_season = self.hw(series,shifts,rv=True,n_preds=0)
119 |         
120 |         de_season=series-smoothed_season
121 |         de_level=de_season-smoothed_value
122 |         noise=torch.randn(de_level.shape[0],de_level.shape[1])
123 |         noisy=de_level#+noise
124 |         noisy=noisy.unsqueeze(2)
125 |         #noisy=noisy.permute(1,0,2)
126 |         #take the last element in the sequence t agg (can also use attn)
127 |         feature=self.RNN(noisy)[1].squeeze()#[-1,:,:]
128 |         pred=self.lin(feature)
129 |         
130 |         #äthe season forecast entail just taking the correct smooothed values 
131 |         season_forecast=[]
132 |         for i in range(self.pred_len):
133 |             season_forecast.append(smoothed_season[:,i%self.slen])
134 |         season_forecast=torch.stack(season_forecast,dim=1)
135 |         
136 |         #in the end we multiply it all together and we are done!
137 |         #here additive seems to work a bit better, need to make that an if/else of the model
138 |         return smoothed_value[:,-1].unsqueeze(1)+season_forecast+pred
139 |        


--------------------------------------------------------------------------------
/es_rnn_example.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import numpy as np \n",
 12 |     "import pandas as pd\n",
 13 |     "\n",
 14 |     "import matplotlib.pyplot as plt\n",
 15 |     "%matplotlib inline\n",
 16 |     "\n",
 17 |     "import random\n",
 18 |     "from torch.utils.data import Dataset,sampler,DataLoader\n",
 19 |     "import torch\n",
 20 |     "import torch.nn as nn\n",
 21 |     "from tqdm import tqdm"
 22 |    ]
 23 |   },
 24 |   {
 25 |    "cell_type": "code",
 26 |    "execution_count": 2,
 27 |    "metadata": {},
 28 |    "outputs": [],
 29 |    "source": [
 30 |     "from data_functions import *\n",
 31 |     "from models import es_rnn,holt_winters_no_trend"
 32 |    ]
 33 |   },
 34 |   {
 35 |    "cell_type": "code",
 36 |    "execution_count": 3,
 37 |    "metadata": {
 38 |     "collapsed": true
 39 |    },
 40 |    "outputs": [],
 41 |    "source": [
 42 |     "series = [30,21,29,31,40,48,53,47,37,39,31,29,17,9,20,24,27,35,41,38,\n",
 43 |     "          27,31,27,26,21,13,21,18,33,35,40,36,22,24,21,20,17,14,17,19,\n",
 44 |     "          26,29,40,31,20,24,18,26,17,9,17,21,28,32,46,33,23,28,22,27,\n",
 45 |     "          18,8,17,21,31,34,44,38,31,30,26,32]"
 46 |    ]
 47 |   },
 48 |   {
 49 |    "cell_type": "code",
 50 |    "execution_count": 4,
 51 |    "metadata": {
 52 |     "collapsed": true
 53 |    },
 54 |    "outputs": [],
 55 |    "source": [
 56 |     "train=series[:-12]\n",
 57 |     "test=series"
 58 |    ]
 59 |   },
 60 |   {
 61 |    "cell_type": "code",
 62 |    "execution_count": 5,
 63 |    "metadata": {
 64 |     "collapsed": true
 65 |    },
 66 |    "outputs": [],
 67 |    "source": [
 68 |     "sl=sequence_labeling_dataset(train,1000,False)\n",
 69 |     "sl_t=sequence_labeling_dataset(test,1000,False)\n",
 70 |     "\n",
 71 |     "train_dl= DataLoader(dataset=sl,\n",
 72 |     "                      batch_size=512,\n",
 73 |     "                      shuffle=False)\n",
 74 |     "\n",
 75 |     "test_dl= DataLoader(dataset=sl_t,\n",
 76 |     "                      batch_size=512,\n",
 77 |     "                      shuffle=False)"
 78 |    ]
 79 |   },
 80 |   {
 81 |    "cell_type": "code",
 82 |    "execution_count": 6,
 83 |    "metadata": {
 84 |     "collapsed": true
 85 |    },
 86 |    "outputs": [],
 87 |    "source": [
 88 |     "hw=es_rnn()"
 89 |    ]
 90 |   },
 91 |   {
 92 |    "cell_type": "code",
 93 |    "execution_count": 7,
 94 |    "metadata": {
 95 |     "collapsed": true
 96 |    },
 97 |    "outputs": [],
 98 |    "source": [
 99 |     "opti = torch.optim.Adam(hw.parameters(), lr=0.01)#,weight_decay=0.0001"
100 |    ]
101 |   },
102 |   {
103 |    "cell_type": "code",
104 |    "execution_count": 8,
105 |    "metadata": {},
106 |    "outputs": [
107 |     {
108 |      "data": {
109 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD8CAYAAABn919SAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noSuKomQJauiKoihZwv8DzLnilJu+Qv8AAAAASUVORK5CYII=\n",
110 |       "text/plain": [
111 |        "<Figure size 432x288 with 1 Axes>"
112 |       ]
113 |      },
114 |      "metadata": {},
115 |      "output_type": "display_data"
116 |     }
117 |    ],
118 |    "source": [
119 |     "#Initial Prediction \n",
120 |     "\n",
121 |     "overall_loss=[]\n",
122 |     "batch=next(iter(test_dl))\n",
123 |     "inp=batch[0].float()#.unsqueeze(2)\n",
124 |     "out=batch[1].float()#.unsqueeze(2).float()\n",
125 |     "shifts=batch[2].numpy()\n",
126 |     "pred=hw(inp,shifts)\n",
127 |     "plt.plot(torch.cat([inp[0],out[0,:]]).detach().numpy(),\"g\")\n",
128 |     "plt.plot(torch.cat([inp[0],pred[0,:]]).detach().numpy(),\"r\")\n",
129 |     "plt.show()"
130 |    ]
131 |   },
132 |   {
133 |    "cell_type": "code",
134 |    "execution_count": 14,
135 |    "metadata": {},
136 |    "outputs": [
137 |     {
138 |      "data": {
139 |       "text/plain": [
140 |        "tensor(2.9005, grad_fn=<PowBackward0>)"
141 |       ]
142 |      },
143 |      "execution_count": 14,
144 |      "metadata": {},
145 |      "output_type": "execute_result"
146 |     }
147 |    ],
148 |    "source": [
149 |     "#Initial Loss RMSE \n",
150 |     "(torch.mean((pred-out)**2))**(1/2)"
151 |    ]
152 |   },
153 |   {
154 |    "cell_type": "code",
155 |    "execution_count": 15,
156 |    "metadata": {},
157 |    "outputs": [
158 |     {
159 |      "data": {
160 |       "text/plain": [
161 |        "tensor(9.5917)"
162 |       ]
163 |      },
164 |      "execution_count": 15,
165 |      "metadata": {},
166 |      "output_type": "execute_result"
167 |     }
168 |    ],
169 |    "source": [
170 |     "#Baseline Loss Predicting last value at each step\n",
171 |     "(torch.mean((inp[0][-1]-out)**2))**(1/2)"
172 |    ]
173 |   },
174 |   {
175 |    "cell_type": "code",
176 |    "execution_count": 16,
177 |    "metadata": {
178 |     "collapsed": true
179 |    },
180 |    "outputs": [],
181 |    "source": [
182 |     "overall_loss_train=[]\n",
183 |     "overall_loss=[]"
184 |    ]
185 |   },
186 |   {
187 |    "cell_type": "code",
188 |    "execution_count": 12,
189 |    "metadata": {},
190 |    "outputs": [
191 |     {
192 |      "name": "stderr",
193 |      "output_type": "stream",
194 |      "text": [
195 |       "  1%|          | 1/100 [00:02<03:27,  2.10s/it]"
196 |      ]
197 |     },
198 |     {
199 |      "name": "stdout",
200 |      "output_type": "stream",
201 |      "text": [
202 |       "6.8607206\n",
203 |       "7.3957877\n"
204 |      ]
205 |     },
206 |     {
207 |      "name": "stderr",
208 |      "output_type": "stream",
209 |      "text": [
210 |       "\r",
211 |       "  2%|▏         | 2/100 [00:04<03:25,  2.10s/it]"
212 |      ]
213 |     },
214 |     {
215 |      "name": "stdout",
216 |      "output_type": "stream",
217 |      "text": [
218 |       "4.0958314\n",
219 |       "4.4489546\n"
220 |      ]
221 |     },
222 |     {
223 |      "name": "stderr",
224 |      "output_type": "stream",
225 |      "text": [
226 |       "\r",
227 |       "  3%|▎         | 3/100 [00:06<03:22,  2.09s/it]"
228 |      ]
229 |     },
230 |     {
231 |      "name": "stdout",
232 |      "output_type": "stream",
233 |      "text": [
234 |       "2.9072347\n",
235 |       "1.7616297\n"
236 |      ]
237 |     },
238 |     {
239 |      "name": "stderr",
240 |      "output_type": "stream",
241 |      "text": [
242 |       "\r",
243 |       "  4%|▍         | 4/100 [00:08<03:21,  2.10s/it]"
244 |      ]
245 |     },
246 |     {
247 |      "name": "stdout",
248 |      "output_type": "stream",
249 |      "text": [
250 |       "2.90861\n",
251 |       "0.27803552\n"
252 |      ]
253 |     },
254 |     {
255 |      "name": "stderr",
256 |      "output_type": "stream",
257 |      "text": [
258 |       "\r",
259 |       "  5%|▌         | 5/100 [00:10<03:17,  2.08s/it]"
260 |      ]
261 |     },
262 |     {
263 |      "name": "stdout",
264 |      "output_type": "stream",
265 |      "text": [
266 |       "2.8909032\n",
267 |       "0.07848331\n"
268 |      ]
269 |     },
270 |     {
271 |      "name": "stderr",
272 |      "output_type": "stream",
273 |      "text": [
274 |       "\r",
275 |       "  6%|▌         | 6/100 [00:12<03:14,  2.07s/it]"
276 |      ]
277 |     },
278 |     {
279 |      "name": "stdout",
280 |      "output_type": "stream",
281 |      "text": [
282 |       "2.9040785\n",
283 |       "0.025534514\n"
284 |      ]
285 |     },
286 |     {
287 |      "name": "stderr",
288 |      "output_type": "stream",
289 |      "text": [
290 |       "\r",
291 |       "  7%|▋         | 7/100 [00:14<03:11,  2.06s/it]"
292 |      ]
293 |     },
294 |     {
295 |      "name": "stdout",
296 |      "output_type": "stream",
297 |      "text": [
298 |       "2.9152913\n",
299 |       "0.02120905\n"
300 |      ]
301 |     },
302 |     {
303 |      "name": "stderr",
304 |      "output_type": "stream",
305 |      "text": [
306 |       "\r",
307 |       "  8%|▊         | 8/100 [00:16<03:08,  2.05s/it]"
308 |      ]
309 |     },
310 |     {
311 |      "name": "stdout",
312 |      "output_type": "stream",
313 |      "text": [
314 |       "2.9007642\n",
315 |       "0.015477379\n"
316 |      ]
317 |     },
318 |     {
319 |      "name": "stderr",
320 |      "output_type": "stream",
321 |      "text": [
322 |       "\r",
323 |       "  9%|▉         | 9/100 [00:18<03:05,  2.03s/it]"
324 |      ]
325 |     },
326 |     {
327 |      "name": "stdout",
328 |      "output_type": "stream",
329 |      "text": [
330 |       "2.903886\n",
331 |       "0.013104449\n"
332 |      ]
333 |     },
334 |     {
335 |      "name": "stderr",
336 |      "output_type": "stream",
337 |      "text": [
338 |       "\r",
339 |       " 10%|█         | 10/100 [00:20<03:03,  2.04s/it]"
340 |      ]
341 |     },
342 |     {
343 |      "name": "stdout",
344 |      "output_type": "stream",
345 |      "text": [
346 |       "2.9012113\n",
347 |       "0.01664723\n"
348 |      ]
349 |     },
350 |     {
351 |      "name": "stderr",
352 |      "output_type": "stream",
353 |      "text": [
354 |       "\r",
355 |       " 11%|█         | 11/100 [00:22<03:00,  2.03s/it]"
356 |      ]
357 |     },
358 |     {
359 |      "name": "stdout",
360 |      "output_type": "stream",
361 |      "text": [
362 |       "2.9098902\n",
363 |       "0.013082137\n"
364 |      ]
365 |     },
366 |     {
367 |      "name": "stderr",
368 |      "output_type": "stream",
369 |      "text": [
370 |       "\r",
371 |       " 12%|█▏        | 12/100 [00:24<02:58,  2.03s/it]"
372 |      ]
373 |     },
374 |     {
375 |      "name": "stdout",
376 |      "output_type": "stream",
377 |      "text": [
378 |       "2.9086177\n",
379 |       "0.013396879\n"
380 |      ]
381 |     },
382 |     {
383 |      "name": "stderr",
384 |      "output_type": "stream",
385 |      "text": [
386 |       "\r",
387 |       " 13%|█▎        | 13/100 [00:26<02:58,  2.05s/it]"
388 |      ]
389 |     },
390 |     {
391 |      "name": "stdout",
392 |      "output_type": "stream",
393 |      "text": [
394 |       "2.9079268\n",
395 |       "0.013137279\n"
396 |      ]
397 |     },
398 |     {
399 |      "name": "stderr",
400 |      "output_type": "stream",
401 |      "text": [
402 |       "\r",
403 |       " 14%|█▍        | 14/100 [00:28<02:55,  2.04s/it]"
404 |      ]
405 |     },
406 |     {
407 |      "name": "stdout",
408 |      "output_type": "stream",
409 |      "text": [
410 |       "2.904755\n",
411 |       "0.011554496\n"
412 |      ]
413 |     },
414 |     {
415 |      "name": "stderr",
416 |      "output_type": "stream",
417 |      "text": [
418 |       "\r",
419 |       " 15%|█▌        | 15/100 [00:30<02:53,  2.04s/it]"
420 |      ]
421 |     },
422 |     {
423 |      "name": "stdout",
424 |      "output_type": "stream",
425 |      "text": [
426 |       "2.9042325\n",
427 |       "0.01075287\n"
428 |      ]
429 |     },
430 |     {
431 |      "name": "stderr",
432 |      "output_type": "stream",
433 |      "text": [
434 |       "\r",
435 |       " 16%|█▌        | 16/100 [00:32<02:53,  2.07s/it]"
436 |      ]
437 |     },
438 |     {
439 |      "name": "stdout",
440 |      "output_type": "stream",
441 |      "text": [
442 |       "2.8992906\n",
443 |       "0.014193204\n"
444 |      ]
445 |     },
446 |     {
447 |      "ename": "KeyboardInterrupt",
448 |      "evalue": "",
449 |      "output_type": "error",
450 |      "traceback": [
451 |       "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
452 |       "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
453 |       "\u001b[0;32m<ipython-input-12-a53d2957661b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     24\u001b[0m         \u001b[0mout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfloat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;31m#.unsqueeze(2).float()\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     25\u001b[0m         \u001b[0mshifts\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 26\u001b[0;31m         \u001b[0mpred\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mhw\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minp\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mshifts\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     27\u001b[0m         \u001b[0;31m#loss=torch.mean(torch.abs(pred-out))\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     28\u001b[0m         \u001b[0mloss\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpred\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m**\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m/\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
454 |       "\u001b[0;32m~/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
455 |       "\u001b[0;32m~/AI/repos/ES-RNN-Pytorch/models.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, series, shifts)\u001b[0m\n\u001b[1;32m    116\u001b[0m         \u001b[0;31m#Get Batch size\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    117\u001b[0m         \u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mseries\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 118\u001b[0;31m         \u001b[0mresult\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0msmoothed_value\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0msmoothed_season\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhw\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mseries\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mshifts\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mrv\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mn_preds\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    119\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    120\u001b[0m         \u001b[0mde_season\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mseries\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0msmoothed_season\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
456 |       "\u001b[0;32m~/anaconda3/envs/pytorch/lib/python3.6/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    487\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    488\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 489\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    490\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    491\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
457 |       "\u001b[0;32m~/AI/repos/ES-RNN-Pytorch/models.py\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, series, series_shifts, n_preds, rv)\u001b[0m\n\u001b[1;32m     37\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     38\u001b[0m         \u001b[0;31m#We use roll to Allow for our random input shifts.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 39\u001b[0;31m         \u001b[0mseasonals\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mroll\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mj\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrol\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mj\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mrol\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minit_season_batch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mseries_shifts\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfloat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     40\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     41\u001b[0m         \u001b[0;31m#It has to be a list such that we dont need inplace tensor changes.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
458 |       "\u001b[0;32m~/AI/repos/ES-RNN-Pytorch/models.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m(.0)\u001b[0m\n\u001b[1;32m     37\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     38\u001b[0m         \u001b[0;31m#We use roll to Allow for our random input shifts.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 39\u001b[0;31m         \u001b[0mseasonals\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mroll\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mj\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrol\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mj\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mrol\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minit_season_batch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mseries_shifts\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfloat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     40\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     41\u001b[0m         \u001b[0;31m#It has to be a list such that we dont need inplace tensor changes.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
459 |       "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
460 |      ]
461 |     }
462 |    ],
463 |    "source": [
464 |     "for j in tqdm(range(20)):\n",
465 |     "    loss_list_b=[]\n",
466 |     "    train_loss_list_b=[]\n",
467 |     "    #here we use batches of past, and to be forecasted value\n",
468 |     "    #batches are determined by a random start integer\n",
469 |     "    for batch in iter(train_dl):\n",
470 |     "\n",
471 |     "        opti.zero_grad()\n",
472 |     "        inp=batch[0].float()#.unsqueeze(2)\n",
473 |     "        out=batch[1].float()#.unsqueeze(2).float()\n",
474 |     "        shifts=batch[2].numpy()\n",
475 |     "        #it returns the whole sequence atm \n",
476 |     "        pred=hw(inp,shifts)\n",
477 |     "        loss=(torch.mean((pred-out)**2))**(1/2)\n",
478 |     "        train_loss_list_b.append(loss.detach().cpu().numpy())\n",
479 |     "        \n",
480 |     "        loss.backward()\n",
481 |     "        opti.step()\n",
482 |     "\n",
483 |     "\n",
484 |     "    #here we use all the available values to forecast the future ones and eval on it\n",
485 |     "    for batch in iter(test_dl):\n",
486 |     "        inp=batch[0].float()#.unsqueeze(2)\n",
487 |     "        out=batch[1].float()#.unsqueeze(2).float()\n",
488 |     "        shifts=batch[2].numpy()\n",
489 |     "        pred=hw(inp,shifts)\n",
490 |     "        #loss=torch.mean(torch.abs(pred-out))\n",
491 |     "        loss=(torch.mean((pred-out)**2))**(1/2)\n",
492 |     "        loss_list_b.append(loss.detach().cpu().numpy())\n",
493 |     "    \n",
494 |     " \n",
495 |     "    print(np.mean(loss_list_b))\n",
496 |     "    print(np.mean(train_loss_list_b))\n",
497 |     "    overall_loss.append(np.mean(loss_list_b))\n",
498 |     "    overall_loss_train.append(np.mean(train_loss_list_b))"
499 |    ]
500 |   },
501 |   {
502 |    "cell_type": "code",
503 |    "execution_count": 13,
504 |    "metadata": {},
505 |    "outputs": [
506 |     {
507 |      "data": {
508 |       "text/plain": [
509 |        "[<matplotlib.lines.Line2D at 0x7f60e9fc27b8>]"
510 |       ]
511 |      },
512 |      "execution_count": 13,
513 |      "metadata": {},
514 |      "output_type": "execute_result"
515 |     },
516 |     {
517 |      "data": {
518 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW4AAAD8CAYAAABXe05zAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
519 |       "text/plain": [
520 |        "<Figure size 432x288 with 1 Axes>"
521 |       ]
522 |      },
523 |      "metadata": {},
524 |      "output_type": "display_data"
525 |     }
526 |    ],
527 |    "source": [
528 |     "#Plot of Train and Validatiaon Loss, we nicely converge\n",
529 |     "plt.plot(overall_loss,\"g\")\n",
530 |     "plt.plot(overall_loss_train,\"r\")"
531 |    ]
532 |   },
533 |   {
534 |    "cell_type": "code",
535 |    "execution_count": 17,
536 |    "metadata": {},
537 |    "outputs": [
538 |     {
539 |      "data": {
540 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD8CAYAAABn919SAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
541 |       "text/plain": [
542 |        "<Figure size 432x288 with 1 Axes>"
543 |       ]
544 |      },
545 |      "metadata": {},
546 |      "output_type": "display_data"
547 |     }
548 |    ],
549 |    "source": [
550 |     "#Forecasting on the Validation set\n",
551 |     "\n",
552 |     "batch=next(iter(test_dl))\n",
553 |     "inp=batch[0].float()#.unsqueeze(2)\n",
554 |     "out=batch[1].float()#.unsqueeze(2).float()\n",
555 |     "shifts=batch[2].numpy()\n",
556 |     "pred=hw(torch.cat([inp,out],dim=1),shifts)\n",
557 |     "\n",
558 |     "#plt.plot(torch.cat([inp,out,pred],dim=1)[0].detach().numpy(),\"r\")\n",
559 |     "\n",
560 |     "\n",
561 |     "plt.plot(torch.cat([inp[0],out[0,:]]).detach().numpy(),\"g\")\n",
562 |     "plt.plot(torch.cat([inp[0],pred[0,:]]).detach().numpy(),\"r\")\n",
563 |     "plt.show()"
564 |    ]
565 |   },
566 |   {
567 |    "cell_type": "code",
568 |    "execution_count": 34,
569 |    "metadata": {},
570 |    "outputs": [
571 |     {
572 |      "data": {
573 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD8CAYAAABn919SAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzsvXeUW+d57vt86L236ZVthk0USVmS\n5S7JiavimuIozomdoyjnJnHqPWedknjlxlnnxCk3UhwnTiLfOHF3lMSxKFlWtyQOJVEkZ1hmhpyG\nGQx678B3//iwBwMCmEEZYO8Z7t9aXMRgbwAvh8CDZ7/f+70voZRCRERERGT3I+E7ABERERGRnUEU\ndBEREZE9gijoIiIiInsEUdBFRERE9giioIuIiIjsEURBFxEREdkjiIIuIiIiskcQBV1ERERkjyAK\nuoiIiMgeQdbNF7PZbHR4eLibLykiIiKy63nttdf8lFL7dud1VdCHh4dx7ty5br6kiIiIyK6HELLY\nyHliykVERERkjyAKuoiIiMgeQRR0ERERkT2CKOgiIiIiewRR0EVERET2CKKgi4iIiOwRREEXERER\n2SPc0oJ++Wuv48JfvcR3GCIiIiI7Qlc3FgmN7K/8OnRpP/DQDN+hiIiIiLTNLSvotEgxGLsEDU2g\nkC1AqpDyHZKIiIhIW9yyKRffpXWYaQhKZOF5zc13OCIiIiJtc8sK+sqZ6Y3b3pfneYxEREREZGe4\nZQU99kpZ0GPn53iMRERERGRnuGUFncxMI0TMyEKO4lVR0EVERHY/t+yiqHF1Bkv6wzCk1qF0iykX\nERGR3c8t6dBZhcs0In0T8JvGYQmIDl1ERGT3c0sKuveCB2YaAp2YRLJ3HL3pedAi5TssERERkba4\nJQXd/RTbSGS4cxIYG4MecfhnvDxHJSIiIij+4A+AX/gFvqNoioYEnRCyQAi5SAg5Twg5V7rPQgh5\nihAyW/rb3NlQdw6uwqXv3glojo4DANZeENMuIiIim3jxReC55/iOoimacejvpJQep5SeLP38ewCe\nppTuA/B06eddAZmZRpBYYD/shO0tTNCjb4gLoyIiIpuIRIBwmO8omqKdlMuHADxWuv0YgA+3H053\nMLmnsayfBJEQ9N09jAIkyF8RHbqIiMgmIhH2p1jkO5KGaVTQKYAnCSGvEUI+W7rPSSldK932AHDW\neiAh5LOEkHOEkHM+n6/NcNuHFikG4jOI9E8CABQ6BVZlg5AviQ5dRERkE+EwQCkQjfIdScM0Wof+\nVkqpmxDiAPAUIeTK5oOUUkoIqVkmQin9MoAvA8DJkyd5LyXxXvDASUOghyY27vPpx2D0iw5dRERk\nE5EI+zsUAkwmfmNpkIYcOqXUXfrbC+B7AE4DWCeE9ABA6e9dUSbifpItiBrunNy4L+4aR29SFHQR\nEZES2SyQTrPbuyiPvq2gE0K0hBA9dxvAfQAuAfhXAA+WTnsQwOOdCnInib3KShb77y8LenF0HBYa\nRPhGiK+wREREhATnzoG9JehgufEXCSFvAjgL4PuU0icAfAHAvYSQWQDvKf0seMjMNALECtuEY+M+\n1eQYAGD1BTGPLiIigkoRD+0eo7dtDp1Seh3AsRr3BwC8uxNBdRKTexor+glYJWTjPstpVroYfm0e\n+PmT9R4qIiJyq7CHHfqegRYpBuPTGxUuHH33jAIAsjNiHl1ERASVgr6LHPotJejeCx6YaBh0olLQ\ntQ4tPJIeSBdEQRcREUGlKxcdujCpVeHCsa4bh8Er5tBFREQgOvTdQLw0mcj1tv1Vx6KOMTjjokMX\nERFBWdAtFtGhC5Xisht5SOE46qo6Vhgeh6u4hoQ3wUNkIiIigiISAQgBBgdFhy5UpOtu+CQuSBXS\nqmOKCVbp4n7herfDEhERERrhMKDXiw5dyKiDbgTVvTWPmW5ntejBs2LaRUTklicSAYxGwGwWBV2o\nGOOriBn6ah7rvYcJenpGdOgiIrc8nKCbTGLKRajYsm5kbLUF3ThkQhpKYH29y1GJiIgIjkiEibno\n0IVJKphic0R7ags6kRAEpXZIQ/y3+BUREeGZcLjs0FMpIJPhO6KGuGUE3Xt+FQAgG6ydQweAiMIO\nVWRXNI0UERHpJJtTLsCucem3jKCHp90AAM2+2g4dABIaBzRJ0aGL7A5+/GvfwHMf+0u+w9ibbF4U\nBXZNHr3RARe7nsQ1JujGifqCnjbY4Yxc7VZIIiJtofja36EnugDgV/kOZW9BaTmHLjp0YZK9wQTd\ndrR+yiVvccCSF1MuIrsDVToMXX53CM2uIpUCcrld6dBvGUHH2iri0MLQb6h/js0OLZJI+pPdi0tE\npEU0uQiMNARa5H2y496C2/Yv5tCFi8Lnhk/eB7KpD/rNSF12AEDwqphHFxE+2nwECuSQCqb4DmVv\nwQk6V7YIiA5daOjCboS19fPnAKAcYFOMonNi2kVE+OiLTHgiC7tDbHYNnBsXHbpwMafcSJrq588B\nQDPEHHpiQXToIsImG89CA+bME+7dITa7hs0pF5WK/REFXTjQIoUjv4qcY2uHbhhnDj2zLDp0EWET\nXS73606u7g6x2TVsFnRgV23/vyUEPTgbgBJZkP6tBd28nzn0/Jro0EWETWK1LOjptd0hNruGzTl0\nYFdt/78lBN3/JitZVAxvnXLRuXSsn4tPFHQRYZNcKwt6zrc7xObZd/0BXtj/n/gOY3s259AB0aEL\njegVtu1fd2Brh04kBAGpA7KgmHIRETaptbKI5327Q2xMr/8IE3OP8x3G9kQigFQKaLXsZ9GhC4v0\nPHPoliNbCzoARBV2KKOiQxcRNllf2aHT0O4QG13aBysNILIo8HgjEcBgYBOLANGhC43CEhN0x7Ge\nbc9NaOxiPxcRwZPzbxpiHN4dYmPMss/V6gsCH8bObfvnEB26sJB43PARB+Qa+bbnpo0OGNNiykVE\n2BRDTNDj0EIaE77YFLIFWKkfABCaEvhUMK51LofJxO4rFvmLqUFuCUFXBVfhV22fbgGAvNkOc0F0\n6CLChkuzrCsGIY8LX9BD80FIwFoUZGcELuhcp0UOk4mJeTzOX0wNcksIuiHmRkzfmKDD7oAWSSS8\nic4GJSLSBiQaQQw6xFVWKFPCT7mEZ8smSbq4C1IumwV9F23/vyUE3Zp2I23ZumSRg+vnEromunQR\n4SKJRxCTmpBWm6HOCN+hx6+zNGYOMhjWd4FD35xD30Xb//e8oGeiGdipD8U6o+duRtnPBD06Lwq6\niHCRJyJIyIzIaU3QZYXvHFNL7PM0qzkOR1zgDv3mHDrn0EVB5x/fRQ8AQDrYmKBrhtn2/8QNcWFU\nRLgoUmGkFEYU9GboC8IXmuwK+zz5xu9ET3FVuClNSoFotDqHDogpFyEQusRKFtXjjQm6YYw59MyK\n6NBFhIsqE0FGZQQ1mmBEBIVsge+QtoR62edJdvcdAAD3C9f5DKc+8ThbABUdujCJX2WCbjjYWA7d\ncpA59Pyq6NBFhIsmF0FWYwLMzD3G3FGeI9oaEvAhSCyw3HUQABA6J9C0CyfatXLookPnn+wC2/Zv\nO9aYQ9c6tEhBJfZzERE02nwEBa0RMhtzj7FlYbtHRciLkNyB3nvGAACpiwJdGL250yJ3mxDRoQsB\nuuJGGkqYxywNnU8kBEGpHbKQ8AT9hV/4Cp597xf4DkOEZ2iRwkjDKOqNkNuZe4wvC9s9quM+xFV2\nGIdMCBArJDcE6tBrCbpEwloBiA6df+ReN7yy3i1Hz91MROmAMiq8lIv521/GqTOfRyaa4TsUER5J\nh9NQgA0xVrqYQ097hO0edWkfUjq2PrWmGYfWs4scOrBrtv83LOiEECkh5A1CyL+Xfh4hhLxKCJkj\nhHyDEKLoXJitow27EdI0uKmoREJjhzYhPIfek5yHFknM/M1LfIciwiOxFSY6EosJ6h7m0NMeYbtH\nU9aLrImtT0VsY3BEBSrotXLo3M97zKH/GoDLm37+YwB/SikdBxACIMhGx8bkKhLG5gQ9Y7DDkBGW\noIdvhGClAQBA5JtneI5GhE/ibiboUosRugHm0PMC7oleyBZgoQEUrcyh54bG0ZtfQjae5TmyGtwK\nDp0Q0g/gfQD+tvQzAfAuAN8unfIYgA93IsB2oEUKR86N7Daj524mb3bAUhBWyoXrUJeFHD0XnuA5\nmsZYevY65h6f5juMhll+/gbm/nWG7zC2hRs5p7AboR9gTrLgF6575Pq4ECdz6LJD45CiCPdLC/wG\nVot6gr7HHPqfAfgdAFy7MSuAMKU0X/p5BUBN1SSEfJYQco4Qcs7X5cqR6EoUWiQB1/Ztcyuw26FB\nSlCbH8KvMUGfGvoYDqQvYP38Gs8RbY/3ow9B8vGP8B1Gw7g/+TnkPvlzfIexLRkvEx2lwwidS4cC\nJIJ2j+FrzBzJe5lDN97GKl38rwgw7RKJAAoFGwy9mb3i0Akh7wfgpZS+1soLUEq/TCk9SSk9abfb\nW3mKlglMs12i8sHmBF3aw5xE8IpwXDrXoU7/u78CAJh95Ek+w2mIgdAFjGavInxD+M4GAHTRNdjT\ny3yHsS3ccAt1jwkSmQQRYgKJClds4jeYkVMPss+/663jAIDkRQFWunDb/slNRRRcC12B04hDvxvA\nBwkhCwC+DpZq+XMAJkKIrHROPwB3RyJsg9jcOgBAPeJq6nFC7OciXZiDR9KDw5+5E16JE+QpYefR\nQ/NBOIvsC3X+G+d4jqYx9Bk/LDSAXDLHdyhbkg8wQdf2srRATGqCLCrcL83kIjNGulFmlGyH7IhB\nB8wJ1KHfnG4BmKAnEkBO2O+NbQWdUvp/U0r7KaXDAD4J4EeU0p8F8AyAj5ZOexCA4IYFJq4zQdGP\nO5t6nGaICXpyUTiCbvDOY103DolMgmvD9+HA0lMo5oXbcH/pB+XceexHUzxG0jjGQgASUASuCOf/\nvRaFAHOKuj4mPAmFGYqkcN1jzs1+n6Z97HNFJARu9TjUawJ06PUEfZds/2+nDv13AXyOEDIHllP/\nys6EtHPklpigWyaac+jGfcxJpJfKKZfnJx/Ci2M/v3PBNYkzPoeok12qkvvvh436ceWfXuctnu2I\nvMwWFyMwQn3xLM/RbE8+nYeJsg9r6LKH52i2IRJBEQQ6lw4AkFaaoBJwT3Suj4t53LpxX8gyDltY\noA795pJFYNds/29K0Cmlz1JK31+6fZ1SeppSOk4p/RilVHC7XeiaBznIGt4lymHez5xEwcPeiGvn\n3Lhr5m8wvPjcjsfYCAlvAq7iGgpDbDFp/8P3AgC8XxVutQu9NI0YdLg09D4MeoXv0EPzwY3b8Tlh\nCzqJRhAlRkhk7OOb0ZihyQrXOUr8XgSIFTKVbOO+TP8Y+rPXhddU7ObWuRy3gEMXPFL/OvwS58Yb\nv1G0Di2SUG/0c7n6m1+GDAU4C6u8pDm4znSKCebQ7ZMOXFafgHlKuHl0/fI0lrSHkL/9DvQUV7F2\nTnBLLBVEbwQ2bqcX13mMZHuk8QjikrLo5HQm6PLCdY7ysA9heWVBhPTAOBTIYW1qhaeo6rBVDh3Y\nWw59t6EKexBWNpc/Bzb1cwl6kY1ncejFLyMPKeTI85JfDZ5ll6am28c27ls/fj8moy8jshSp9zBe\n6Y9MI9gzCcv9pwAAS98RtkuPL/g3bhfcwnbo8mQYCXlZdIoGMwxF4TpHro/LZvS3MXPi/bHA0i63\ncA5d8OjiHsR1zeXPOaJKO5QxH879t+/BWfTgxwc+DQAIXFzdyRAbIn2Jvem5TnUAYP7p90KGAi4/\n8qOux7MdwdkAHMV1FA5MYvyjx5GDDKnnhJ1HT7vLDp2sC1vQlakIUsrK9q5qpJEOp/kLagv0KS+S\nekfFfY472Xs5/qaAFkYLBSAWu3Vy6LsNU8aDtKk1QU9oHdAmvNA89iiWZCMwf+4XAQDRy91PHUhu\nzCNArDCNmDfuO/TptwAA0mcvdD2e7Vg+wxZEtacnobaoMac+Av1VYTv07Cpz6AloIA8KW9BVWTbc\ngoNYmNhEl4TpHo05H3LGSofuur0PGShQnBPQoItoqaf8Vg5dFHR+KOaLsBW9KNiaT7kAQEZvx0hy\nGscjz+P6fQ/BdtsAACA9331B13nmsKYZq7hPoVMgAiNIMFDnUfwR+TErWex59wQAwDd0CmOhc4Iu\nsyx42e9xUTMBTVTYOXRNLoKcpiw6cjsTm/iK8AR9o4+LrdKhS2QSeGQDkHuWeIqsBvW2/QOAWs3+\nBIT3edvMnhX00HwQcuRBelpz6HmzHWqkkYIKR//sF2GbdKIIguJy9wXdHptHxDZedX9YZoMs4q/x\nCH7hKlz67hwEAJA7TsNEw1h8WmD50s34/UhBhYhlBIaUsB26vhBGQbdZ0JlDT6wIzz0GZ1ltP3FU\n7xIP6gahDy7yEFUdthJ0ALDbBT/4Zs8KenCG2/bfmqDDwRzFufFPwrLPCrlGDp/ECel6d3PomWgG\nvfkl5IaqBT2mtEIZF55j0K/MYFE7sdGD3vl+tjC6+rhw0y7SSAAhqQ1ZiwvWrHAFnRYp9DQKaijn\nedW9wu2JHpljAsj1cdlM3DYEe0JAgl6vdS6HwwF4hdMOpBZ7VtBjs+xDqR1tTdAVB4ZRBIHtf/7q\nxn0BVR9Uwe46dPdLC5CiCNmBsapjKY0N2qTwHHp/ZBqh3smNn0ffP4EENMj/WLgLo4qoHzG5FdTp\nghERwS4wJrwJyFCocJGaXiZA2XXhOfTYPBNA9ZCj6lihbwjO4qpwWi2IDl24pBZYHrTZbf8cp//o\np7DwH5dx6Odu37gvZuiDMd5dQQ+cZVUAxturHXpGb4U+KyyHvrnChUOmkmHOcAKWeeE6dHUygITa\nBmkve7/4p4WZR98YbmEui45+sNQTPSA8h55aZgKoG6l26NKRQUhA4TknkFr05VJjNkf1lw8AUdD5\nJLfMHLp1sjWHLlPJMPoTByruy1h6Yc10N+WSvMDyzq67qx16wWiFqSAsh778BFsQ1Z6aqLg/tO80\nxuNvCMeN3YQuE0BaZ4VqmL1fwleEmXZJuJloy6ybBL2f3aZB4Qn6zX1cNqOdGAIABN4QyMLo1BQT\n7f7+2se5lAul3Y2rCfasoMPjQQoqGPoNO/aUxZ4+WGmgu5fjc3OIQQfbRLVroFYbdEgIasYo18Ol\n997JivsVd52CGmnMP36Jj7C2xZjzI2e0QTfOBD0xL0xBT3mYQ1c4ynlelUmFFFSCLKmj614UQSr6\nuHBYjrNF8/i0QPLoU1PA6dPVrXM57HYgnWZdFwXKnhV0WcADv8zV1HDo7ZAOshke3vPdc+matXm4\n1eM1/x0SO/uQhOaEk3ahl6YRhR69dwxU3N/3YbYw6v2+8NIuhWwBJhoCNVthOsgEPbMoTEFPrzNB\nVzkr87wRiRmSmPAcuiTgQ4hYKvq4cLhOsfdI/roAHHosBly+DJw6Vf8cLhUj4LTLnhV0dWQdkRa2\n/W+FZh8T9NCl7uXRbeE5hCzV6RYAkPfYAADR68JJuxiWp7G0qcKFY+BtI7yVfW5HZDEMKYogNuvG\nlVBxTZg59Jy/NNzCVSnoCZkJ8rjwHDrr41I7J60yqeCVOCFZEYBDf+01lko5fbr+OdyAHgFXuuxZ\nQdcnPUjoWyxZrIPhYC8AIDHbHYdeyBbQl7uBTH/1gigAqPqYQ08sCcehswqXiar7JTIJYjCARITn\nIiPX2e9P5rJBrpEjQKyQeIXp0PN+9vvjhltwJJRmKAXYE10d9yKmrj+pzKsegsYvAIc+Vbpy3Mqh\nc4IuOvTuY854kDHvrKDbjzOHnl3ojstcO7sMBXKQHqgt6Noh5tDTK8Jw6IGrftipt6LCZTMxqRHS\nuPCaiXGNuZS97AsyqHBBIdDt/8UQ+/0ZBitrpTMqE9Rp4Tl0fcqHlK6+oMdMg7DEBODQz54FRkYA\nm63+OWLKhR/y6Tys1I+iY2cF3ThkYm113d0RdK4Tne5Y7ZSLYYQJUM4jDIe+8iRbENXdUVvQk3Ij\nZEnhCXpqhf3+NIPswxzTuKCNCVPQEYkgBxnUFnXF3VmtGZqc8By6MedDzlSnDBBAxjUEV3YJtMhz\n5cjU1NbuHBBTLnwRuOKDBBSSnp3NoRMJgU/WC4WvOymX+KUbAAD76ZGax7nKgaJPGIIe+tYPUQTB\nwPuP1TyeUpigTAlPdDJu5tD1w+z3mTI4YUgLM4dOYmy4xc1rFHm9CfqCsH63G31crPUdOhkahAYp\nBK7yeJXp9QKLi1vnzwFAq2X9XESH3l24EWKKVrf9b/Xc2j5ow91x6MUl9jqO4701jysNSsSgAwnw\nn3LJxrOYeOnLOOf4STiP99Q8J6MyQpURnkPnGnMZR0tXPDYXbHkP/66xBrJYGHFp9U5GajTDSMOC\naoC2VR8XDtUBVovuneIx7dJI/pxD4Nv/96Sgb2z7H9t5QU+Y+mBKdkfQJR43fMQBhU5R95yI1App\nhH+Hfu6/fheO4jrIww/XPSenMUKTE56gIxBAFnLoe/XsZ6cLWiQR98T5jasG8lQESXntft1SFAUV\nc/gaEz55f/2Ui/EIq0WPXuJxYXRqCpBIgBMntj9X4LtF96SgZ5bY5bJx/86mXAAgZ++DM+/uintT\nB1fhV/VteU5UaYMyxr9D1z72CBZlY7j9v95f95y8zghdQXiCLgn5EZJYN9IYsgFmBALTwsujK9MR\npJXVDl1qLfVEXxTOwmj8BhM+9UB9h+48zRx6+hqPDv3sWWBiAtDptj9XFPTuk18pbfuf2HlBR18v\nVMggfKPzHxxDzI2YfmtBT6qtUCf5dejXvn0Bx6Iv4sZ7H9pyfis1mGCkYcGlMhTRAKLycnWDeoi9\nb6Kzwsujq7MRZNTVgi53sH4uXGsAIZCcY2tN+rH6Dt00YkYcWpAlnhw6pY0tiHKIKZfuQ9Y9iEIP\nrUO748+tGGEC6zvf+bSLNeNG2lI7f86R0dmgz/Dr0D3/41GkoMKxP/v01icajZChgKQ/2Z3AGkSd\n8COhKm9N1+8T7vZ/bS6MnLY65aJ0svtSq8Jx6IVnnkMEBgzdu7/uOURC4FEOQenhyaEvLgJ+//YL\nohycQxdoP5c9KejygAcB+c7nzwFAt58JeqdH0WWiGdioH8WerR163miFKc+fQ48sRXDi8j/i3L6f\nhnnMsuW5XIdArmOgUNBmAkhry4JuPsTeO9kl4Qm6rhBBUVft0Lme6Jl1YTh0WqQYnT2Dy73vgVwj\n3/LckH4QhghPDv1sqaVzMw5dwP1c9qSgq2PriKo7kG4BYDnCBDY139nSRd+FNQDl/jH1oFYbjIjw\n1sXw/G88Bh0SsP2P+ouhHFyHwMSqsATdmPUjayinXCz7rChAAromLEEv5ovQIwZqqBZ0bR9z6Dmf\nMAT9xhNX0V9YQvad9ddUOFL2IThSPDn0qSlAqQSOHGnsfIHXou9JQTckPUju8LZ/DvtRVpJXWOqs\nQw9eZM+vHt9a0ImNOcvw9WBH46kFLVIM/PujuKS9o6JvfD24DoHJVWGIDsD+DWYaQNFcduhShRR+\niQMSv7By6LHVGCSggKla0A1DpZ7ofmGkXJb/5gkAwOhD2wt6sX8QNupHwsuD6z17Fjh+HFDUrySr\nQODb//ekoFtzHmQtnRF0pUEJP7FBstZZQY9fY8/P9Y+pB9egi+tH0k1iqzGMZq/C/7afauh8pYMJ\nUcYrHIceXY6wCUD2yi3fIaULqpCwHHpsmX0RSi3VOXR9nwEFSAA//yWsAKB58QyuKw6g/+6hbc+V\nj7Nz1s8tdzqsSopF1pSr0XQLUH/7/+c/D/zSL+1cbC2y5wQ9HU7DRMOgzs4IOgAElH1QBTor6LkF\nltKxHdvaoW806Frs/sJoYi0KAJDat86dc3AdArM+4Qh6eL7UmMtR2a87pnVBFxeWoCfX2O9NZqt2\n6BKZBEFigyTAv3NMh9OY9D+H5UPbu3MA0E2wWvTgG11Ou3g8LBc+Ud1Mri71Ui7f/S7wve/tXGwt\nsucEPXCZ/aK5UWKdIKrvhT7W2Rw6XXEjDeW2C42agdJ29ZXuO7MNQbc0NkSE6xDIdQwUAvFF9ntT\n9lU69IzRCVNGWILODbdQ2mvPvAwr7FBE+Bf06S+9AA1S0DzQmKDbbmcOPXm1ywujXKnk4GDjj6mV\ncikUgCtXgGCQ/eGRPSfo3Ogw5VDnHHrK0gdrprMOXe5zY13Wt+2ADsMoE6LsWvcd+sb0HGtjgs51\nCOQ6BgqB5BL7van7Kx163uaCrbAuqJp5LlV183ALjrjaAU2c/8W62LefQBpKTDz09obOd97Wizyk\nKF5vzKF71ihe+ce59gtNFkuvN7R9WmgDrRbQaCoF/fp1VvkCAPPzbQbVHntO0ONzTNC5UWKdoOjq\ng63o7WhliTa8irBm6/w5AJjGSg26vN136Bkfc+hcbnw7NDYN8pCWp6sLgMwa+73phioFnfS4oECu\nKxvIGoXbAa0dqH3VltLboU/z79D7Lp3BtOWehveByFQyeKR9kK015tCv/vG/4NSnDmD1lTYdPSfo\nzTh0gLn0zSmXmZny7VqCfv488NBD5SHUHWTPCTo3Osx0oHMpF8lgHySg8F3s3CW5KelGwrh1/hxg\nIpmEGgh0X9BzASboamdjDp1ICOsUGBOOoOc9zKGbxitTLnIBbv8vvvY6otBj4G21u2/mTHaY8vwK\n+trUCvZlphG7s7F0C0dAOwhdoDGHXjx/AVIUMUjbzLkvLQEmE2Bocu7wzdv/p6fLt+fmqs//8Y+B\nL32p/qzSHWTPCTo3Osx6qP5243ZRjzLnzJUW7jS0SOHIuZF1bC/oABCWWiENdz/lkg8yQde4Gv9A\nxKQmyIQ0+9IfQAESGAYqrzLUw8wQcI3ehID9+lnMm07Wba9AbQ6YaYi3PQkAMPfokwCA3k83J+hx\nyxCsicYct3yJiaYy2uaX1+Jic+kWDoejWtAHB4He3toOfXqafWn0NfZ5boc9J+jE60GQWKA0KDv2\nGoZD7D8mdrUzgh5ZDEODFNCzfcoFAKIKGxSx7jv0QikXru1pXNCTciPkAhpyIQn6ESTWKpE07GcO\nPbUgjFr0TDSD8eSbiByov0Vd4mQLdsFr/LWCkP3wCaxJerHvgcNNPS7XMwhXfgWFbGHbc42+kgtu\ntxZ8aan5dAtQO+UyMQGMjdWDIEr0AAAgAElEQVR26Nxx0aE3j8q/gpC8c+kWALAeZYKe6VCHOP8F\nVkHD9Y3ZjqTKCk2i8kP87Ie+iGfv+q87HlsFEebQN9rONkBaaYQyLRxBl0cDiMqtVfdbJ7nt/2vd\nDqkm89+7AAVyUN5Tv2aaa1PLta3tNsV8EQfdP8Tc6H3bLubfjGRkCHLkMf9vM9ue60qUXHC7gt6u\nQ6e0XOEyOQmMj9cW9OlpdrwL7ClBD80HcdTzJNyjb+3o61gP2HBdcQC2p7/RkeePzDDnr93fmKCn\ndVboMpUOfegHf43bXn6kswMPYlFEod+yw+LNZFVGqLLCEXRVIoCEslrQjUMm5CADfPy3JgYA3/dZ\nz5Ghj9Z36JpB5tATC/zk0cM3QjDTEIqHa0+s2orx/+snESQWSH/mE4gs1X9/RJYisNPSv68dQY9E\n2J9WHTrXz4WrcOEEnatt5/D52B+hCDohREUIOUsIeZMQMk0I+f3S/SOEkFcJIXOEkG8QQhrcO9s5\n3vzcP0CNNJy/v31fkXYgEoLl9/8KDifOYuar53b8+ZOzTNDNhxsT9JzRBmO+LDyRxTBGctdgRBQL\nT83ueHwc0ngUCUlzC0o5rQlaAc2+1Kb9SGmrBwMTCUGYWCAJCWPnpfT1KXglTvSc6q97jm6ECXp6\nmR9BD11lVway3ubXr3rvGMDSF7+DwewsZk98Avl0vuZ5qy9sylG3I+hcDXorDn3z5iKuwoVLuQCV\nefTNx7tAI9YqA+BdlNJjAI4DeC8h5C0A/hjAn1JKxwGEAPynzoW5PcV8ESP/8Sgu6O/GgY837xCa\n5fifPog4tAh8/pEdf+7CMku5cH1jtoOarTDR0Eb+cf4b5S+Z1cfP7nh8HLJEBElZc4Je0BmhF9CQ\nC302gKy+2qEDQERhgyIqDIfes3wWC7ZTW6YyzAeYkObc/KRcuIEWqi0GWmzF8V9/B17+1KM4GTiD\nl+78rZrnhM4xsSxodO0Jeqsli0Dl9n+uwmVigjl0oDLtwh0XikOnDG6ulbz0hwJ4F4Bvl+5/DMCH\nOxIhgB//xjfxwqf/bstzXv/CkxjKzyP+YGfdOYdx0IjXD/0cTs59HcHZnXVxkjU3gsRSNdm9HsRu\ngwQUkUXmfKNPsxmJKahQeHlqR2PbjDwdRVLRWA06BzUYoUdUELMvaZHCUvSjYK526ACQUFqhSvDv\n0KMrUYxkryB5eOue3aYRM/KQgnr5cejJJfa6utHWK8ze9tXP4Lnjv4a3n/9zvPDg31Ydz84wsaQn\nT/Hv0DlBHxwE9Pr6Dr1LFS5Agzl0QoiUEHIegBfAUwDmAYQppdx10QqAmhETQj5LCDlHCDnna/E/\nQPr1r8Hx9T/f8pzi//sIfMSBk3/0kZZeoxVcn38YaqRx4Tf+fkefVxlww69s/A0gc5Y6Ls4xN6m6\neBY35Ptw1Xga1vnOOXRlJoqMsskaXqMRElDEVmOdCaoJEt4ElMgC1toOPaW1QZvm36HPf/M1SECh\ne+fWTaQkMglCxMpbP5fcCrsyMI635tA57n75/+BNw1sx8rXPVx2TLMzDK3FCNj7SvkNXKABnCwUU\nN6dcuHSKycTeSzc79C5VuAANCjqltEApPQ6gH8BpAAcbfQFK6ZcppScppSft9tb+ozMmFyxb9NVY\nfv4GTnq/j+m7PrPlQOWdZv9HjuBNwz0YOfNXO+o49dFVRHSNCzrXhySxxNzk0PoUVntPITx+CuOJ\n88jGszsW22bUuShyquYEnZt9GXfzn3bhvgClztoOPWuwwpDj36FHfsiuskY/sX1XwLDCAUWEpyqX\ndSawlv21f5+NIlPJEHrHA+gvLMF3qbJs1LA+B492vP3JQYuLwMAAGw7dLJyOeTzlCheO8fFqh96l\ndAvQZJULpTQM4BkAdwIwEUJkpUP9ADrW3KRod8JKfXUXSuZ/+0ugINj/J7/cqRDqkviFhzGUv47X\n/vCJHXtOS9qNtLmxGnSg3KArueSH5/VV9BTdyJ84DcU9p6FCBvOPX9qx2DajzUWQ1zQn6EIachFb\nYGKt6Knt0ItmGyxFP+/9XJQXzmJRNgrLvtpxbiautkMT58ehE78PYWLaEVNlvpd9ed34ZmXK0Bmb\nQ9RREvRcDohGW3uBpaXW0i1AuZ/Lq6+WK1w4Ntei+/3MxXdpQRRorMrFTggxlW6rAdwL4DKYsH+0\ndNqDAB7vVJCk1wUJaM0NE+lwGkemvoKpng+h946BToVQl5N/+AC8EifwyM4sjubTediL6yi4Gnfo\n5QZdASx+m30ALPefwsAD7EPBlbztNNpiFAV9czl0RalTYHqdf0HnWg7f3JhrA6sVcuR5Tw8NeKbg\n7m1s5mVK74A+zY9Dl4e9CMnaS7dwjH/8BAqQIPlcWdBTwRQzK0Nj7Q+aWFxsbUGUw24Hnn+e3d4s\n2OPj7Msik+n6gijQmEPvAfAMIeQCgCkAT1FK/x3A7wL4HCFkDoAVwFc6FaRykG3yCF2uTru88Yff\nh5UGoPj1X+nUy2+JQqfAzF2fwe2+H8A33f4HyXfRAymKkAw0LujGUSZIBW8AqeenkIcUYx85jv63\nDiNArCDndn5htJgvwoAYoG/OoXOdAtMe/ksX0wulRm7DtVME0lKPdD6Gh3D4Lq2zUW7HGxvCkDPZ\nYeapn4s65kNMtTMtN7QOLeZVk9DOlM2I+4XrAADFxHh7gp7NAmtrrTt0gFW6hEvv4c2CPjbG0kAL\nC+WSRSEJOqX0AqX0NkrpUUrpYUrpH5Tuv04pPU0pHaeUfoxSmulUkNoxJui1+mpkLrHLm/0//5ZO\nvfy2qO46wa4gLrXfI517DtVo4ykXnUuHLOSA3w/95bOYUx+BxqYBkRBct52Ga2nnHXrcUyp8MjYn\n6Jre0uxLP/8OnT77LILEgqH37Kt5nFubiN3gb2GUSzlwKYjtoDY7TDTcsXWTrdClvEjqdsahA8D6\nwCmMBqY2Ul6BV9ln3XSyTUFfWWGi246gc6/PVbhwbC5d7GIPF45dsVPUsI+tRHOOajNk1Y0IDNC5\ndN0OawNVL2tnmlxpv7l97ApbitAfbPxNQCQEQYkN0pAfY6EpeIfKH/7EoVMYzcyUBXiH4BY1pabm\nBF3XVxpyEeRX0GmRYt/1M7jS/x5IFdKa53CpmOQyfw49+dwUCpBg/OMnGjpf4mIOmY9+LqasD1nj\nzgk6PXUaVhrAyosLAIDUNFts7HlrmymXdmrQObha9Jvz45ygz893tYcLx64QdOsEE/S8u7pRktLv\nhl/RvW/AWmj6maCnV9sX9Mx1JujbjZ67mZjcCtfiqzDRMMjpcr5V+87TkKKI+W+93nZsm0l62GIU\nt8jZKPp+dj7lecjF7HcvwlVcQ+Hd9bsCcqmYjJs/h66dOctSDw32Fpf3MaGLzHU37VLMF2GhfhSt\nO9fl1P6TzJgsf4ddYUrm5xAiZjbFqx1Bb6cGfSO40uvfnE6x2wGdruzQu5huAXaJoOtcOsSgA1mv\nduja6CrCTZT4dQL9EBP03Hr7gk7dq8hBBuvB5pxOXG3DvgyrZnG8r+zQhz/Gboee3Nk8emqdCbq8\nwWlFHCqTCmkoy/lHnlj9+zMAgPGH6ws6tzaRX+fHodMixWhgCuuDjS2IAvX7ubz63/4VL/zizu6X\n2ExoPggpiiCOnXPo4w8cQRpKZF9i713t2hzW1KXNOxpN9eSgRuEcen/9NgrbUk/QCWEu/ZVXul7h\nAuwSQQeAgNwFeaBa0C0pN1KmxvPNncA4bAYAFP3tC7ps3Q2vtKephlcAkNaW0gNQY+yD5TeZfdKB\nFekQFG/ubB59Y1qRvcmNRQBixAgJz0MuDC+fwazyMHpO1jcDxiETCpCA+vkR9MBVP6w0gOLkkYYf\nox9jDjm1WLlAr/mLL2D0q/9zR+PbTHiWCau8b+cculwjx6zuNphn2XvXFp1H2DZePuHmvuSNsrgI\nuFyAStV6cNyGpFqCPTYGTJUMlOjQaxNVOaGOVgp6MV+Eo7CGvJNfh662qJnrDLU/rkwTciOobv7f\nkzOw9MCc/gRkKlnFsWXXKfSv7axDz5YWNRudVrSZuMwIWYI/QU94E5gMvQD34a2HMEgVUoSJGZIg\nPykX32ssNaA60HhqwLSPOcf8WlnoaJFiMD6NnsIK0uH0zgZZot0+LvUIjp7CePQ1pIIp9OcXkB3c\nJOg3Tw5qlFb7oG/mgQeARx8FTtVYrB7fFKPo0GuTMLhgSFbm0P0zXsiRB+nnV9CJhCAssUASad+h\nGxOriDcweu5mimbm0IP7qi/PM8dOYzB/A4GrOydMhdK0Im1vczl0AEjJjZAn+Uu5zDz6LJTIwvDx\n9257bkRugyzKj0OPXGCpAePRJgSd6+eyXnbontdXYUQUElC4X7yx43ECQHKBvZ52eGcFXXbXaWiR\nxJt//ASkKEJ2YKx8sFVBb7UP+mb0ejYntNZOU07QDYb20jotsGsEPWtxwZKrdOiBi6zETznCb8oF\nAGJyCxSx9gXdlnUja2vhC8rOHLrirmrHYCqVvF3/xs659EK4JOhNTCviSKlMUPE45CL5vTNIQo2J\nz27fNz+utEIV58ehZ2aZQ3eeblx8JDIJghIbJMGy0K0+VZ55yZX+baaYL+Lady62ESmQW2WvZ9q/\ns6Mfez7I3rv0n/4ZAGA4cZND9za594PS9naJNgLXpKvLFS7ALhJ06nDBTEPIRMvl7q2U+HWKpNIC\nZbI9QfdNe2FEFHSg+ctB9eQYspBj8JN3VR0b+/jtSEMJyZ//ad32CU1TmlbUSrloVm2EOsefoA9e\nPoNp+zugMm2fQ01prNCmeCpbXFxEHFqYRsxNPSwit0MRKQt67NWyoKcuVc+8fOU3v4Xxjx7bKA9s\nhZ3q43Izw/fuQwRGHF/5NwCA6601Ui7N9HPxetkuznZTLlvBOfQup1uAXSTokl62CBG4XP5GTs8z\nQbcc4V/Q0xozNOn2BP3aI08BABwfe3vTjz39+Q8gcnEZ/XdXOw99rx5TP/+XuD34FF56y2+2FSMH\niUaanlbEkdcYoeVJ0Jefv4GR3DUk39rYEOOswQZDlh+HrvQswqMcanqcW0zjgCZe/pxIL0/DT2yI\nwADMVzv07Musm6P/tdZHKhK/FyFihlwjb/k5aiGRSTBvPgk10ohDC/vhTd0RN08OahSuwqWTDr2v\nD7j3XuDDHesoXpddI+iqIbZbNHylnHYprqyiAEnlfzJPZHUW6HLtLYrSJ56An9hw8Gca20SyGSIh\nW/4e7nnsl/Dsid/A29/8Czz/s3/dTpgAAEk8iri0+fw5ABT0RuiL/OTQb3yJlSsOfGb7/DkAFIxW\nmIr8OHRDZAkhffNOMq2zQ58uO3TT6gxWDJNYVY9Ds1bt0LULzMEnr9fvaLod8rAP4R3q43Iz0YNs\nXcitGqv8cmulFp2rQe+kQ5dIgCefBD7wgc69Rr2X7vortohunAl6fK78ppN63PBJnFVVHXxQMFpg\nLLTu0Iv5Ig4sPImrg/e15Hob4Z6X/zem7D+JO//pV/HGn/yoreeSJaNNTyvawGiCDomdS/80geLZ\nM1iRDmHk/v2NPcBmgxppJP3JjsZVC0dqESl7805ycz8XrsIlMjCJkG0c9ki1Q3eFWM+R3HLrgs76\nuHRG0FWlwdghy3jlgVYEvRsOnUd2jaCbDjJBTy+W33TqoBtBFf/pFgCA2QIdEhU5/ma49q03Yade\n0PsaSwW0glQhxYHX/xmLyv0Y/u2PbjmMdzvk6ShS8tYEnZiYs4+5W2x92iKFbAETa0/j+r77G05j\ncA26QrONp12uffsCLhjubqtLY9KfhI36Uexv3klSu2Ojn4vnNTeMiAKTk8j2j6Evt1DxRRr3xDGQ\nX2CP81TvxG4U1sdlZxdEOYY+xhx6um+s8kCrDl2nY8Mo9iC7RtCth9ibpbhaftMZ427EDMIQdImN\n7RaNLLSWdvE8xlIB+3/1vh2LqRaGfgPC//2LMNMQrn+79XYAqnSk+WlFJaSWkqAvdzft4n1zjXWI\nvP32hh+j6C016FpoPO2y9s/P4mjsx1h+6krTMXJ4zrLUgHy8eScpcTKhC17zw/0kS6cY75yE9MA4\n5Mhj9ZWljXOXzlzeuC31t+7Qd7qPy2Z6Tvbh+U8+ipEv3DTvoBVBn58Hhoe7Xn3SLXaNoCsNSoSI\nGcRbftNZs6vIWPkvWQQAmZ1VIkQXWku7GF89g6uqY3Acde1kWDXp/wm28zD68vQ2Z9ZHnYsip24t\nh871f0mudXdh1HeOXW5rDjbuelV9zKFzvdMbga6wxfrN6cFmCb3JRFc30bxDV/Qz8xOZ9SI+xdIp\n/fdNbJT8+V4up12CL7D3QJiYoAq3Fm+5j0tnBB0A3vbPD2HoXW06dEqBc+eA227b2eAExK4RdAAI\nyl1QBNmbLhVMwUKDKPYIw6Ere0odF93NO/TYagyHwy9i7VhjC3Xt4jzegzAxgcy0LuiafLTpaUUc\nSie73O32kIvYJSbo5uONu17tIBP0zGrjDl3uZYKeWWo9hZG4zGK13d68Q1eX+rnEb/ggvTwNH7HD\ndsgOx51MEBMXyguj+TenkYYSc5Y7oIu3JugbfVycnUm51EWnA5TKxgV9ZQVYX6+9u3OPsKsEPapx\nQVva/u89zzYVyYaEIejqvtY7Ls488gzkyMP48c7lzzdDJARLugkY3DMtP4e2GEVB15qgc0MuMt7u\nCnpunrle1+nGXa9xjKVccmuNO3RNmAl6wd26Qy/eWEIeUjhva/4KVD/KBD297INpdRorRtZPxHm8\nB0moUZwtO3Tt4gwWVQeRsvTBlGntC2ijj0tv5xx6TQhpbrco11/ldOPNznYbu0rQUwYnDGn2pgtP\nsw+NelwYgq4bZIKe9TQv6Ol/PYM4tJj87N07HVZdwn2TGIhOtzQvs5AttDStiEPTwwQ95+tuDl2y\nvIgAsTbcihYAa9UKNNWgy5RgZmNzerBZZKuL8Ej7WqrgMh9gTjnnXsdgfAbRASboEpkEbuUo1O6y\noPcEp+F3TqJgd8FeXG9p2PlGH5fBLjt0oHlBl8mAY8c6GxOP7CpBz1ldsJa2/ydm2YfGeEgYOXTj\nCPvg533NC/rwlTOYcbxzR4brNgo9NAkrDcA/0/zYvI1hGabWcuj6AZZyKXR5yIXatwSvqrmctEwl\nY+mpYGOCTosU9hwzG1x6sBV0gSUEtK3VShuHTMhDCtnF86zCZdOOxaB5HJYQS7nEPXH0FxaR2zcB\n0uOCDAUEZ5uvuef6uOhGuuzQgeYE/exZ4OjR9rosCpxdJejU6YIeccQ9cWRvsA+N/bgwHLq+z4AC\nJECwuRz64tNzGMrPI/X27uTPOfRvYa5t5UzzefTEGis3bHZa0cZr97HH0XB3Bd0UXUTE1HxOOiK1\nQhZpLOUSXYlCB7ZzURNrPYduTSwibmmtVprr5zK8+CwAVuHCkeobR39mHsV8EYs/YCk39clJyAfY\nprTgTPNfQlwfF+O4gAW9WGQLons43QLsMkGX9bMKkMDMOrDqRgIaGAZac4k7jUQmQYSYQMLNOfSF\nv2blisO/3J38OUfve5hri77SfB59Y1qRpTVBl2vkSEADRMqCnkvmkEvmWnq+RqBFCldmERlH8643\nprRBGa90rvl0vubGKP8FduWYhRzGVGsOvZAtwJVfQa6n9d2MYYUD/QW2ZjDw3rKgS/aNQY001s+v\nbVS4ON81Ce1oaePefPNfQkUPc+g73celIRoV9NlZIBrd0wuiwC4TdNUwe9NFZ9eh8K3CJ+9tus9F\nJ4nKLJBFmxN05Ys/xKJsFEPvHt/+5B2knUqXlIcJcbPTijYTkxghjZVz6Of2/wwu9XWuBj+yGIYe\ncdAWdgimNFZok5UO/eWjv4zz/e+vfp0ZduU4rz68kR5sFu+ba5AjD8lI67sZ42rmln3EDuuBstBq\nj7H32fpLcyhcnEEaSgy8fRSG/eyzlbrRfMwk4OtIH5eGsNuBeJz1dNkKbkFUFHThoB9nl4WJeQ+0\nETfCGmGkWzjiCguU8cYFnRYpRtZfwfJg9xZDOVilyySMK80LOjetiKtWaYWEzLQx5GLp2eu4w/0d\nHAs/B//lzszCXJ9iblW5v3mRzOit0GcrHfr4/BnsC7xStaicnGWC7hu8HXrEkfA20TiqhP91Fqvm\nUOuCntazBUquwoXDcRcT9Nj5eWgXp7GoOgipQgrrJBP0Vrb/y8M+hOQ8LIgCjdeinz3LRtYdOtT5\nmHhkVwm6+RB702WWPDAn3UiYhCXoKbUF6iY6Lnpec8NZ9KBwgp+8XrhvEgOx5itdsoGSoDtad+hJ\nhRGKFBP067/zJUhAIQHF1b98quXn3Irwm6VhEYebT2PkTTYYC2VB97y+ip6iG0ZEEJqv/P/OLzJB\nx22swVpgpvkUBlcvbzraesolZ2JCx1W4cPScHkAOMhSuzm1UuACsI2cKKsDTvKCrY96O9XHZlkYF\nfWqK7RCW8d/3qZPsKkG3HrSzGY+rHjjyq8g5hCXoWa0F2mzji6KL32aXgdb38nMZSA9NwEKDTVe6\nbEwramG4BUdGaYQqE0YqmMLRc1/BKz0PIECsoE+cafk5tyJ9jble+8kWXK/FCj3iG316uP83AFh9\nvrLZFfGsIkxM0EyOAAAiV5sXyOwsE/Rm6uVvhtqY0JGbZlrKVDKsyEegnX0D/YUl5PaxtRQiIfDL\nXJAFm/8C0qV8SGl5EnRH6cpgK0HP5YDz5/d8ugXYZYIuVUgRkNihvD4DFTIgfcIoWeTIG8ww5Bt3\n6OnnzyIHGcY/wk9dbKuVLly5YTuCntUYoclG8NrvfAMWGoTyt/4Lrgzeh/0LZ1qqhd4OurCIFFSw\nHWpeeCQOloMOzTGXnnquPHA78nplO1ql3w2fog/6fexqMjHfvKBLVpYQJJaWhodsPIeLCZ3hLdVD\nFvzGMRzxPwOAVbhwhFUuqCPNx2vK+pA1CTjlcukSy7GLgi48Qgon+tdfAwAoRoTl0KnJAhMNNSxI\nhqtTmNUca2hyTifou499mJvu6RKNogjSluDktSZoCxGYv/4o5hWHcPzX3wF67/1wFNcx+50LLT9v\nPZSeRXjkgy0tostdbPt/9DpbGNVfmcK84iCKIMhdrnTo+qgbEV1fRXqwWdTeRayr2mvvOvLL9+Gl\n4Z/FgU9Vp/OSPeNQgV1tON9VFvSE3gVDorl4u9HHZUs4Qd9qFN3Z0hewKOjCI6Z1bbT71B0QlqAT\nqwUSUESXt6+vLuaLGAtNwT/M35vMcdTVUqULiUURa3FaEUdRb4SjuI7JxBRWPvgrIBKCfQ+zKpe1\nf9j5tIs+tIRgC8MiAEA9wBx6YikAWqQYC03BPfo2tpNzsdKhW1KrSJl7YdlvQxEEdK35FIYpuoSo\nsb0BDANvG8HdN/4Raou6+mBpRBpX4cKRNTlhzjYn6Bt9XBw8CbrRCMjlWzv0qSnAYgFGR+ufs0fY\ndYKeNpa7EZonhZVykdpLLXRvbJ92WXhqFkZEQe7gb6PDRqVLkz1dJPEoEtLW0y0A2AcRQAw63Pan\nPw+AlVJeVR2F8ZUn2nvuGthTi0jaWnO9mgHm0NMrfiw+PQcTDYOcPoV1/TgMvrJDL2QLsBc9KDjZ\nlv0AsUPS5PZ/WqRwZhaRdnZuAIP6MGvSxVW4cBQdLlhooKn9ABt9XPp4SrkQAths2wv6qVN7tmXu\nZnadoOdtZUF3HBeWoCtcTNATK9svjK4+zi4DXR/g9zKwlUoXWTLS+rSiEtyQizcmPwVDf/m51o7c\nj8nwS+X2AjtAJpph1UR9LW6l5xp0eQJw/wtbEHV+4DRiznH0xMuC7ru0DhkKkAywK8eg0gVlky1p\no8sRGBBraVB4o9jewhw6V+HCIel1QQKKwJXGS0dj8yzVoRrgyaEDW28uSiSA6elbIt0C7EJBh4vV\novuIvau9TxpB1cN6oqfc2zv0wstTiEOL0ffxWxdLJyZhoUH4LjWeGlCkokjJ29uhqzk8igwU6P3D\nhyvuN37ivVAgh5lHnmnr+TfjmVoGAMjGWnO95nHm0Iu+API/Posk1Bh9/wSKw2OwU+/GZKLgJbZL\nVDnKBD2mdUEba07QPa+yCpdW6uUbpe+tIwgRM/Kn7qy4X1ma2xu63HjMqSUmpLpRnhw6sLWgv/EG\nUCjs+S3/HLtO0OXc9n+hjJ7bhHaAOfTM2vaCbrk+hTnj7RWXvHygv4NVQTRT6aLMRFueVsRx8r//\nBFLXPRj/UKVLnPjM3UhAg9TjO5dHD55nJYvaQ625XoVOgRh0IAE/zPNTmNOfgEwlg3KSOd3VF1ge\nPXaF1aDrD7Arx7TRCVO6uRx65CKL1dBCvXyjKA1KSJcW8NavPVRxv2aEmaVmBnNwfVxM+3h06E5n\n/fr5K6WpUUeOdC8eHtl1gq4p9ZyI6YSVbgEAwzAT9Jx3a0HPxrPYF38D4XH+LwO5SpfYq43n0dm0\nojZTLhIC04i56n6lQYlpxzsxfHnn8ujxaeZ6rSdad71hmQ3ygAf7Yq8juI+5PctpJujBsyztkr7O\nBN16lJmNvNUFW8HTVDordYXF6jjV2SHGhn5DlZkoz+1t/EuI6+PCXcXwQl8fsLrKJhLdjLu00atX\neHrRCXadoHP1vSmr8By6cZgJFPVvLejzj1+CChko7uH/MtBx1MVG+zVR6aLJR1qeVtQIqXvux1B+\nHos/mt/+5AYo3FhCEQSuk/0tP0dMYcXIyvNQIw3FXeyLuPcetriYmWFx0mU38pDCNlFKP/S4oEKm\noaonDrq4hDSUsB7svuO1TjCH3sxgDhLwIUgs/PRx4ejtBTIZIFjjc+d2s81HCmGlZzvFrhN0rudE\n0SU8QecuzRHeelHU9322IDrwAP8OnUgIlvT1K11mvnoOVzS3VWxx1xWjKOg71+Vy6JdZK+Ebf/D/\n7cjzSd2LWJf0tLXmklTb0FNkOfK+B9gXsb5XDx9xQHKDOXSpdxU+iWvD+cr7mEAGphsXSOXaAtbk\ng22VhLaKxqZBFHqQ9RK/cNAAABnYSURBVMbjVQTXEZbzmG4BmEMHmEu/mdXV8vFbgF0n6OYxC174\n9N/hwP/+Jb5DqUlEaoE0srVDJ+em4Cc29L91uDtBbUPMNgJrYrHmMe93X8DB1HnM/MkPALDSPD3i\nLU8raoShd4/j5b6P4p7nPo+p3/+Ptp9P61+CX9NeTjqjYymFILFg8B3lemaPdgz6dSbomqAbAXVZ\nPNQj5e6gjeJaP4912+T2J3aIgNwFeRODOZy+i/BZDnYwogbgBJtLr2zG7b5l0i1AA4JOCBkghDxD\nCJkhhEwTQn6tdL+FEPIUIWS29Hd1QrRD3PN3n0bPSWF+6ybkZii26bjoWjqLG7ZTgmn9m3f0wVlY\nrb3DtTTBnj7Bctob5YTGzgk6kRAcff0fcE19DAf+1ycx93jrw6wBwBJfRMzcXk46b2Kli9ctlf9v\nEcc4HDGWcjEm3Igbyu9LriVt8npjAhm+EcJIbhbpI/yl4qJqFzTRxr6AIothjGavInWY5ytNTrDr\nCbro0CvIA/hNSukEgLcAeJgQMgHg9wA8TSndB+Dp0s+3PEmVBapkfUFPeBMYzcwgMcF//pyD9PVC\njjwCV6un8ih87ENyYOFJFPPFtqcVNYrWoYXxmceRkmih+OgHasbWCMV8Ea7cMrKu9hw6tTCHHj9U\nKV75oXH0FZaRCqZgy64iYyuLh2WiuZa08984BwAwvJs/gUwaXDA0OJhj/pusBYf+3Ty/l+sJeibD\nyhlFQS9DKV2jlL5euh0DcBlAH4APAXisdNpjAD7cqSB3E2mtBdpMfUGf++brkKII7Tv4z59zcHXT\ngQvVDkcTYXlJO/Xi2rfeRGKVLfDJrJ2fFNV7xwD8f/s47Pk1rJz+KRSyhaafwz/jZY3chttz6MTG\nBF3zjkrxkh9kC6M3/n0aJhoG7Slf3ptGzMhBBrrWmEDGfsQ2LY194mRbsbZD1uyEtcHt/9EfsrUg\nPuMFwBY87fbqHDpXyigKem0IIcMAbgPwKgAnpXStdMgDwFnnMZ8lhJwjhJzzNTrMdReT01ugy9Vf\nFI2cZXWxvfcLpy6W64kTvVwt6OaEG2+Y3gEA8Dx2Bmkvc+gKW2cdOsfkp09j6lN/gWPRF3Dx0Rea\nfrx3iq0NqA+2J+iOD74F11RHsO8XKoeRGG9npYve7zwPAJANlcVDIpPAL3FCGmgshaG+eBY35Pth\nHDK1FWs7UKcLRkSQDm8zAQiA6uIUbsj31Sw/7Tq9vdUO/RYrWQSaEHRCiA7AdwD8OqU0uvkYpZQC\nqFlsSyn9MqX0JKX0pN3O82p4FygaLTAVg3Vrj4vrLHVgOSCc34XlcGkjzPVKh0OLFI68G5Gx23FV\ndQymV57YmFaktHdH0AHg+P/zCeQgQ/gbzW82ik2zjTrtDIsAgEM/dzv2py7APGapuL/3bUzQ1Wef\nAwBo9lW6wZDKBVWD2/8HvVNw9/ObvpD2sTSRf3r7L6Gh9bNY7RXIlWZfX31BFx16JYQQOZiYf41S\n+t3S3euEkJ7S8R4AzU1J2KuYzVAhg1QwVft4IIAENLW74PGE/YgLRRAUlys/EOEbIaiRBvr7sHb0\nfkxGXkJqjp3TzrSiZjH0GzBtvAuuN5vfbJS5xhy683RnNuqYxyyIwIj96+zqwTRZKR4JrRO6+PaC\nvnbOjZ7iKgq38SuQ3Nze8JWtY+amNuV5mrZVBbe5aDPcz6KglyGEEABfAXCZUvrFTYf+FcCDpdsP\nAnh858PbfUi26bgoC/sRlvK4q64Gco0cPokTUk+loHMT7BUjfTB+4r2QIw/lD/4FAKDt7XwOfTPh\n0/fjYOp8Uz1nAIAsLyECA4yDnYmXSAjcmnGYKUuz2Y5WXt6nTS6YM9sL+tJ3WP7ccj+/gq4dLc/t\n3QpuahPf8W7Q18d6ouc2dYp0uwGlkrXOvUVoxKHfDeBTAN5FCDlf+vOTAL4A4F5CyCyA95R+vuWR\nO9mbJ7ZYW9AVsQCiClvNY3wSUPVBFawUdG6CvW5fLyY/ezfi0OK4l6U9dL3dc+gA4HyQbTa6+pdP\nNvU4w9IleFTDHYioTNhaGrwMXUXnSAAo2F2wFb3bDj1JPT/Fpld99HjH4myEjcEci1sLeur5KeQh\nxfjHbutGWNvT28u2/q+tle/jatBvgba5HI1UubxIKSWU0qOU0uOlP/9BKQ1QSt9NKd1HKX0PpbTx\n2Wt7GFUPE/Sku/bCqDrpR1ItLIcOADF9LwzxyktWLr1iPtwHhU6BGcc7oUAORRBoHdquxnfgE8fh\nI3aQM43n0ReemsVt4WewdudHOhgZkB1glS4+RfWlPelxQYYCgrOBqmOb0V85izn1Ed5TcdZDrG1B\ncZvBHPrLwoh3g1qbi26xGnRgF+4UFTqafibo6dXa32+6TAAZnfAcesbaB1um0qEXltjPXN/51Nvu\nB4C2pxW1gkQmwbWh+7B/8cmGR/wt/N5fIQcZDn3xMx2NTXqAOfSwtlo85AMshRGcqe942fSqc/AN\n85+PVugUCBDrloM5aJFiNHQO3iGBpFuA2tv/b7Ft/4Ao6DuOfqjUcXG9tqCb8n7kjMJz6MWePlhp\noKJcTeJZRYBYoTQoAQDD/5mlPeLS7ubPOej998NOfbj69Te2PTfpT+L4G3+PqYGfgvN4T0fjMpxg\ngp40VZfHabnuoLP1BXLzFCQhEFI4odhi+//SM/Mw0xCIkHqM37y5iNJbbts/IAr6jmMYYjW5BV+1\noBeyBRhpGNQsPEGXDrA3vu9COQepDLjhV5YdztC7x7EoG217WlGrHPhVNnN0/avbp11e+61/homG\nof3th7c9t12cd7GUS95R7Qa57f9btaRdfbw0Ben9whD0qGbrwRwr32UbihzvE0a8ANgYOoWiLOiR\nCJBMig5dpD20Di2ykNds5Rm+EYIEFMQuvJSLepy98UOXymkXQ9SNqL7yA7Hy2c9j7YHOi2Qt7Ied\nuKy+DaazWws6LVLYvvkIZpWHcfThezoel/N4D569/XNw/pePVx3juoNutf0///IUEtBg9P0THYux\nGVIGJwzp+lXI+ZenkIQaYx/kr4lYFYQwN86lXG7BkkUAkPEdwF6DSAjCEgskkepF0ci8H1YAMqfw\nHLpxgr3x41fLgm7JrMI3eKLivLsf+ZmuxnUz68fux92v/B9EV6JVFSUcl/72FRxJvYHnf/qvsK8L\nDdCIhOAd5/6k5jF9rx4pqOpP1AFgmTuLOcMJHFMJ4+OYN9lhXqwv6Oa5s5jTn8BRgcS7webNRbfg\nLlFAdOgdISazQB6rdujxRVbpoOoXnkO3H2eCnl1kziaXzMFeXEfRKawPhOkT90OOPC4/Wn/maPSP\nHkEUepz44s91MbLaEAmBX+aCPFBb0HPJHMbjbyC0Tzj5aGqzw4AYMtFM1bFcMofx2BsbU5sExebt\n/7fgLlFAFPSOkFBaoExUC3pqmW371wwI0KEPmZiTLLXL9V30QAK6McFeKEz80l2IQYfM47V3jfqm\nvTi58C28ceRB6Fy6LkdXm7DKBVWdlrTzj1+qmIIkBCQuVroYvFrde+n6v01Dg5Sg4t2Ac+jcgigg\nOnSR9klprdCmqj8M2TXm0PUjwnPoRELglfVttMsNXmR/c7l1oaDQKTDjfCeGr52p2S9n5n99E0pk\n0ff5/8xDdLVJ6F0wJNZqHvP/6AIAoOd9J2oe5wNFH+szFL5WnXbxPXMRAOD6CYFsKNpMXx+QSACx\nGMuhWyyAWiB18l1CFPQOkDH3wJqpHodV8DJBN40Jz6EDQEjbB22YCXn8Goufm2AvJNLveC8G8zew\n+PRc1TH1809gUTaG8Q8JZ8EuY+2FPVtj+AKA/MIKAMB1aqCbIW2JZpg59MRCtSnJL7EvJseJ1uez\ndozNpYu3YMkiIAp6Ryj29MFG/dU5SL8fGfz/7Z1rbJvXecd/D0WRIiVRJCXqRsWS747TzklgrCk8\nFGmaxu5atAF2a7EB3tChH5Zt3dBhyLZPAzbsgmE3oBiQXrZ8KLoWWbcG/eAtSNJ1K4Z0bhI7vsX2\nbEcWdacs6kpRl7MP56VJWpYlpKbfQ/L5fRFfkkaeHDznz+f9v+c8J/TAd1nulKWOfuJLVshX7jjB\n3iWGvmA3OL33QuVql5W5FR6ZfJ0bh477EdaWmL40cTPL0vTSps8CYxlmJOnOjkugfY+t0FdG7tLu\nemKcBVqdsbMqKN8t2oC7REEFvSo07bKJVL6mG6BpNstMoMuZo+fupJBK072WwWwYzEiGAs10HnTP\nHhp8ai/vBfcS+UGloF/4yg9pZYnIsyd8iuzuBHd5a/zPbL5rC2dHyYbcqiQTB6ygr45utlyasxNk\ng70POqSdUb5btAF3iYIKelWI7LUTtOhDFwnNTTMfctNuAWAgTYQ8s9dvEZwaZbKp35fT53fCjUPH\neWTytYq7oNy3TlGgmcPPfdTHyDZztzX+RdrnMuTa3BKejsE4qwRhcnOFHpkbJxe561k2/lO0WIaH\n7TJRtVyU+0HsYTtBF69UVmQtS1kWW9yreIuEBu0EmD47SvRWhlsRdydE5NkTtLLEha/88PZ7fe/8\nO+fjx5yzA4pr/Jeubq7Qk/kMy0m3BF0CQjaQomlms6DHlsZZane0Qo9GIR6HN9+EjQ2t0JX7w+01\n3dcrK7K2/DT5Vncr9OJRdLkLGeKLGRY63J0Qh5/7KAWayX3b2i4Tb49xMH+W3BNu2S2wdT6s5ddI\nbUyw3uveOOdCKUK5zZZLsjDOStJRQQcr4j/6Uel1g6GCXgXiuxMs04IZqZzAsdUsqx3uVuiJD9gJ\nsHw1Q6qQoZByd0K09bZxPn6MvrNW0K982fZJ7/tVtx6Igj1xaZEoMlqZD1PnJmhig8CAe3dCC9Fu\noouVFXphoUDSzGC6HRf0Bt1UBCroVcGu6e6neao0gTfWNkiYGTYcbMxVpNgmd/3Cu7SzUHGCvYvk\nPnScg/kzTLw9RuA/TjEZ6GH/z/2U32FtQgLCZHOa5ulKy6X4jKVlr3vCk29PEctXCnr2oq3YA32O\neuhQ6Zurh67cL2ajaVpnSxN47maOIOvQ5a6gh2NhstJJx2Xb/a95yD2hKaf3pK3GL//9KQ7efIXL\nQ884+xB3tjVN22xlhT5/yV63H3JvnNcS3STWKi2XWxdt+4LwoOMVOkAwCN3d/sbiA25mfx2wGE8T\nXypN4Nw1u6ko2Ouu5QIwHU6zd/Y0AK0H3BOacg78whEmAz30f+Mv6TRZ5IR7/nmRpY5+EsuVgr5y\n3f7gd37QvUrybv1cFq5aQW/bVwOC3tcHgcaTt8b7P35AFFL9dK9mbm9Pn79u+7iE+9yt0AHm2tPE\nmAc2n2DvGoFggMtDz7C3cIkNhIO/+XG/Q9qS1e403WujFe0KzEiGVYJ0HXavkgz02LXo5f1cVoZt\nP5r4IYcFvWizNKDdAiro1SOdJsoyufdmAVgesRV6dJfbFXo+UZoIqSPuTwo5bm2XS9HH6Xo45XM0\nWyMDacIUKs4WDU5kmGzqc9ImCj9kf2TK+7msZ2yFXjx31EmKFXoDPhAFFfSqEfL85+mz3lb6UTuR\n2wbdrtCLS+hydDjboqCcg7/1DAWamTz6Sb9DuSeh3V4+nCnZLpHZUWYibgpPZJf9cSzv5yIT48xK\nnJZ4i19hbY8KulINytd0A6xNWMslvs/tCr3YLncqXBsTouvhFMP/9hYf+tfn/Q7lnrQdsHc7cxdL\ngh5fyLAYc/MuqNjPJX+zVKGHZsbJNjtst4B9EHrkCBw75nckvuDYkSP1Q/ywnahLV7wJPJ1lleCW\np+y4QnGbeu4uJ9i7ikudFbeiuMY/f6208qmrkGGk62m/QronyUPWVlkbK1Xo0fkJ5l3d9l+kqQne\nftvvKHxDK/QqcXtN97AV9MDMNLOSdLYxV5Fiu9zlhJuVY63SfaQPKOXDwvgCHcxh+t384Yw91GHP\nxi3r5xJbHmcp5niF3uCooFeJlngLWekkMG4rsub5LLmQ23YLlNrlrvW4KTS1SqgtxJSkCIx7J0J5\nnReDg27+cN7u55ItWS6dq+Osdqqgu4xaLlUkG+4nnPV2Ay5mWQy7/UAUrCf9/WN/RPpLn/U7lLoj\n25KmZcYK+ex5mxet+9394ZwLpQjN2Qp9cXKRGPOYHhV0l1FBryK5tjTtc3biti1Pk03u9zmi7ZGA\n8OR//4nfYdQlc+1pYvM2H4rPVoqdGF1kIZq63c8le2GCVqCp33EPvcFRy6WK5JNpOvPe9u7VLIWY\n+xW6Uj3K82H1hv3r8lr/fHs3sby1XHKX7aailiGt0F1GBb2KrPemSW1MsLq0SnJjmo2E+x66Uj02\nevtJmSkKCwVkbJQ52mnvb/c7rC1ZS6RIrHmWy//ZTUXt+1XQXUYFvYoEBvoJYBh+7SohVqFTK/RG\npng04eSZMULTGaZC7totACbVTYx58rN5CsNW0BMPq6C7jAp6FSm2RZ145SwATd0q6I1MMR9uncvQ\nlss4v9Y/0Fvq57IxOs4GQuchd9srKCroVaV4FF3htBX0cFotl0am+AB04d0MieVRlhJuC3o4bcU7\nd3WKwNQEWeki2KLrKFxGBb2KFNuiRq9aQY8MaIXeyBTzYeXaCD3ro6x1u/tAFCA6ZHeLLt6YInxr\nnFthtVtcRwW9inQeSlGgmf6sFfS2Ia3QG5nk/k7yhAmeO0Mza8iA2xV6eT+X1vlx5qO6ZNF1thV0\nEfm6iEyKyLmy95Ii8oqIXPH+JqobZm0SCAaYbOpjYH0YgI49WqE3MsWjCftu2kOMw3vcFvTEASvo\na2NTxPPjLHdohe46O6nQ/wm48yiY54FXjTH7gVe9a+Uu3Ip4PV0I0DEY9zkaxW9mI/3sLlwCSh0Y\nXeV2P5eJSTrXJ1jTbf/Os62gG2N+AMzc8fZngBe91y8Cz97nuOqGhQ5bhc1KgqZQk8/RKH6zEE8T\nwJ5aVOyb4yrFfi7hkatEyEOvCrrrvF8PvccYM+a9HgfUXNuCQpedtHNBtVsUKKRsPqwTIPVB9wUy\nF+6md/IdAJoHdJq7zk/8UNQYYwCz1eci8gUROS0ip6emprb6Wt1SbI8636IPRBXAy4epQE9NLAFc\njKQYXL0CQGS3+z9Ajc77FfQJEekD8P5ObvVFY8wLxpijxpijqVTjbUpo9tqjLke1QlcgNGTzYabF\nbf+8SL49ddsi0m3/7vN+Bf1l4KT3+iTw3fsTTv0R9dqjFmJaoSvQesC7Y4u57Z8XWUuWDoROHlZB\nd52dLFv8JvA/wEERGRGRzwN/DnxcRK4AT3vXyl2IP+J5ph1aoSulfFjprA1BN13e0kWaSOxN+hyN\nsh3bmnjGmM9t8dHH7nMsdUn3Y2mWiCC7h3yORHGB7kf7WSTKxj73e+NDqZ/LVKCHvqDuQ3Qd95/K\n1Dit3a3c/M/zPPFobXimSnWJJCMMv/4OH368Nir08IC1XGbDvfT5HIuyPSroD4CHPrLb7xAUh9j1\n5B6/Q9gx0UFboS+06ZLFWkDvoRRF2ZLYPluh5+P6QLQWUEFXFGVL4vtthb7eqRV6LaCWi6IoWxIb\niPH943/GwG982u9QlB2ggq4oypZIQHjylPbeqxXUclEURakTVNAVRVHqBBV0RVGUOkEFXVEUpU5Q\nQVcURakTVNAVRVHqBBV0RVGUOkEFXVEUpU4Qe4LcA/qPiUwB773Pf94FTN/HcOoRHaOdoeO0PTpG\n2/Mgx2jQGLPtkW8PVNB/EkTktDHmqN9xuIyO0c7QcdoeHaPtcXGM1HJRFEWpE1TQFUVR6oRaEvQX\n/A6gBtAx2hk6TtujY7Q9zo1RzXjoiqIoyr2ppQpdURRFuQc1IegickJE3hWRqyKizZkBEXlIRF4X\nkQsicl5Evui9nxSRV0Tkivc34XesfiMiTSLyloh8z7veLSJvePn0LREJ+R2j34hIXEReEpFLInJR\nRD6suVSJiPyuN9fOicg3RaTFtVxyXtBFpAn4MvAJ4DDwORE57G9UTrAGfMkYcxh4AnjOG5fngVeN\nMfuBV73rRueLwMWy678A/sYYsw+4BXzel6jc4u+AU8aYQ8AR7HhpLnmISBr4beCoMeYDQBPwWRzL\nJecFHfhp4Kox5poxpgD8M/AZn2PyHWPMmDHmTe/1PHYCprFj86L3tReBZ/2J0A1EZAD4JPBV71qA\np4CXvK/oGIl0AB8BvgZgjCkYY2bRXLqTIBARkSAQBcZwLJdqQdDTwM2y6xHvPcVDRIaAx4A3gB5j\nzJj30TjQ6Kf7/i3w+8CGd90JzBpj1rxrzSfYDUwB/+hZU18VkVY0l25jjMkAfwUMY4U8B/wYx3Kp\nFgRduQci0gb8C/A7xpi58s+MXcLUsMuYRORTwKQx5sd+x+I4QeBx4B+MMY8Bi9xhr2guSQJ7x7Ib\n6AdagRO+BnUXakHQM8BDZdcD3nsNj4g0Y8X8G8aY73hvT4hIn/d5HzDpV3wOcAz4tIjcwFp1T2G9\n4rh32wyaT2AryxFjzBve9UtYgddcKvE0cN0YM2WMWQW+g80vp3KpFgT9f4H93tPkEPZBxMs+x+Q7\nnhf8NeCiMeavyz56GTjpvT4JfPdBx+YKxpg/MMYMGGOGsHnzmjHml4HXgZ/3vtbQYwRgjBkHborI\nQe+tjwEX0FwqZxh4QkSi3twrjpFTuVQTG4tE5GexXmgT8HVjzJ/6HJLviMjPAP8FvEPJH/5DrI/+\nbWAXtrPlLxpjZnwJ0iFE5Eng94wxnxKRPdiKPQm8BfyKMWbFz/j8RkQexT44DgHXgF/DFnyaSx4i\n8sfAL2FXmL0F/DrWM3cml2pC0BVFUZTtqQXLRVEURdkBKuiKoih1ggq6oihKnaCCriiKUieooCuK\notQJKuiKoih1ggq6oihKnaCCriiKUif8P2/5UhmOtu5CAAAAAElFTkSuQmCC\n",
574 |       "text/plain": [
575 |        "<Figure size 432x288 with 1 Axes>"
576 |       ]
577 |      },
578 |      "metadata": {},
579 |      "output_type": "display_data"
580 |     }
581 |    ],
582 |    "source": [
583 |     "#Forecasting to the Future looks good.\n",
584 |     "\n",
585 |     "batch=next(iter(test_dl))\n",
586 |     "inp=batch[0].float()#.unsqueeze(2)\n",
587 |     "out=batch[1].float()#.unsqueeze(2).float()\n",
588 |     "shifts=batch[2].numpy()\n",
589 |     "\n",
590 |     "pred=hw(torch.cat([inp,out],dim=1),shifts)\n",
591 |     "plt.plot(torch.cat([inp,out],dim=1)[0].detach().numpy(),\"b\")\n",
592 |     "plt.plot(torch.cat([inp,out,pred],dim=1)[0].detach().numpy(),\"r\")\n",
593 |     "\n",
594 |     "\n",
595 |     "plt.show()"
596 |    ]
597 |   },
598 |   {
599 |    "cell_type": "code",
600 |    "execution_count": 37,
601 |    "metadata": {},
602 |    "outputs": [
603 |     {
604 |      "data": {
605 |       "text/plain": [
606 |        "array([0.4233447 , 0.37306103], dtype=float32)"
607 |       ]
608 |      },
609 |      "execution_count": 37,
610 |      "metadata": {},
611 |      "output_type": "execute_result"
612 |     }
613 |    ],
614 |    "source": [
615 |     "param_list=[]\n",
616 |     "for params in hw.parameters():\n",
617 |     "    param_list.append(params)\n",
618 |     "param_list=torch.sigmoid(params[0:2]).detach().numpy()\n",
619 |     "\n",
620 |     "param_list"
621 |    ]
622 |   },
623 |   {
624 |    "cell_type": "code",
625 |    "execution_count": null,
626 |    "metadata": {
627 |     "collapsed": true
628 |    },
629 |    "outputs": [],
630 |    "source": []
631 |   }
632 |  ],
633 |  "metadata": {
634 |   "kernelspec": {
635 |    "display_name": "Python (pytorchenv)",
636 |    "language": "python",
637 |    "name": "pytorch"
638 |   },
639 |   "language_info": {
640 |    "codemirror_mode": {
641 |     "name": "ipython",
642 |     "version": 3
643 |    },
644 |    "file_extension": ".py",
645 |    "mimetype": "text/x-python",
646 |    "name": "python",
647 |    "nbconvert_exporter": "python",
648 |    "pygments_lexer": "ipython3",
649 |    "version": "3.6.3"
650 |   }
651 |  },
652 |  "nbformat": 4,
653 |  "nbformat_minor": 2
654 | }
655 | 


--------------------------------------------------------------------------------