-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
809 lines (605 loc) · 19 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
<!DOCTYPE html>
<html>
<head>
<title></title>
<!-- 2018-01-31 Wed 22:38 -->
<meta charset="utf-8" />
<meta htto-equiv="X-UA-Compatible" content="chrome=1" />
<meta name="generator" content="Org-mode with org-ioslide" />
<meta name="author" content="fatfingererr @ Sukki 2018 二月聚會" />
<!--<meta name="viewport" content="width=device-width, initial-scale=1.0, minimum-scale=1.0">-->
<!--<meta name="viewport" content="width=device-width, initial-scale=1.0">-->
<!--This one seems to work all the time, but really small on ipad-->
<!--<meta name="viewport" content="initial-scale=0.4">-->
<meta name="apple-mobile-web-app-capable" content="yes" />
<link rel="stylesheet" media="all" href="theme/css/default.css" />
<link rel="stylesheet" media="only screen and (max-device-width: 480px)" href="theme/css/phone.css" />
<link rel="stylesheet" media="all" href="theme/css/small-icon.css" />
<base target="_blank"> <!-- This amazingness opens all links in a new tab. -->
<script data-main="js/slides" src="js/require-1.0.8.min.js"></script>
<script src="js/jquery-1.7.1.min.js" type="text/javascript"></script>
<script src="js/mathjax/MathJax.js?config=TeX-AMS-MML_HTMLorMML,local/local" type="text/javascript"></script>
</head>
<body style="opacity: 0">
<slides class="layout-widescreen">
<slide class="title-slide segue nobackground">
<aside class="gdbar"><img src="images/sukki-icon.png"></aside>
<!-- The content of this hgroup is replaced programmatically through the slide_config.json. -->
<hgroup class="auto-fadein">
<h1 data-config-title><!-- populated from slide_config.json --></h1>
<h2 data-config-subtitle><!-- populated from slide_config.json --></h2>
<p data-config-presenter><!-- populated from slide_config.json --></p>
</hgroup>
</slide>
<slide id="sec-" class=" segue dark quote nobackground" style="background-image: url(nil)">
<aside class="gdbar right bottom"><img src="images/sukki-icon.png"></aside><hgroup class="">
<h2 class=" "><div id="outline-container-orgea4441c" class="outline-2">
<h2 id="orgea4441c">TensorForce 基本介紹</h2>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="flexbox vleft auto-fadein" id="text-">
</article>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgccab18b" class="outline-4">
<h4 id="orgccab18b">TensorForce 簡介</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>開源的 Reinforcement Learning (RL) library</li>
</ul>
<ul>
<li>建立在 TensorFlow 上，目前與 Python 2.7 和 3.5 相容</li>
</ul>
<ul>
<li>沒有對輸入和狀態有限制，可自由建構代理與環境</li>
</ul>
<ul>
<li>在 Agent Logic 和 Update Logic 做了嚴格分離，以便在真實環境中使用</li>
</ul>
<ul>
<li>主打將 RL Logic 呈現在 TensorFlow 的圖表上頭，減少對 Python 的依賴</li>
</ul>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgd36a163" class="outline-4">
<h4 id="orgd36a163">懶人安裝法</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>安裝詳細請見官網 <a href="https://github.com/reinforceio/tensorforce#installation">GitHub - reinforceio/tensorforce : Installation</a></li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="bash">
# 已有 tensorflow
pip install tensorforce
# 沒有 tensorflow + 與 tensorflow 一同安裝
pip install tensorforce[tf]
# 務必更新, 否則可能會有 error
pip install --upgrade tensorforce</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org94131e5" class="outline-4">
<h4 id="org94131e5">開門見山來 DEMO</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>如果沒有 <code>gym</code> 請先安裝，詳細見官網 <a href="https://github.com/openai/gym#installation">GitHub - openai/gym</a></li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="bash">
pip install gym</pre>
</div>
<ul>
<li>再來 demo , <code>TensorForce</code> 有提供 example</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="bash">
git clone https://github.com/reinforceio/tensorforce.git
cd tensorforce</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org16bcad9" class="outline-4">
<h4 id="org16bcad9">開門見山來 DEMO</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>測試執行請留意添加輸出資料夾的參數 <code>--monitor</code> ，方便檢視：</li>
</ul>
<ul>
<li>我則是在 tensorforce 資料夾底下開一個 <code>results</code> (windows)</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="batch">
mkdir results
python examples/openai_gym.py CartPole-v0^
-a examples/configs/vpg.json^
-n examples/configs/mlp2_network.json^
-e 100^
-m 5000^
--monitor results</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org6d7402b" class="outline-4">
<h4 id="org6d7402b">DEMO 結果</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>輸出會有兩個 <code>json</code> 檔案</li>
</ul>
<article class="flexbox vcenter">
<div class="figure">
<p><img src="images/tensorforce-results.png" alt="tensorforce-results.png" width="800px" />
</p>
</div>
</article>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgc05bcd7" class="outline-4">
<h4 id="orgc05bcd7">DEMO 結果</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li><code>epsiode_batch.json</code> 內容大致如下</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="json">
{"initial_reset_timestamp": 1517301297.9757686,
"timestamps": [1517301297.9968548, 1517301298.0198836, ... , 1517301299.1702468],
"episode_lengths": [14, 22, 29, 16, 15, ... , 28, 22, 24, 17, 21, 23],
"episode_rewards": [14.0, 22.0, 29.0, 16.0, ..., 21.0, 24.0, 40.0, 23.0],
"episode_types": ["t", "t", "t", "t", ... , "t", "t", "t", "t"]}</pre>
</div>
</article>
</slide>
</slide>
</slide>
<slide id="sec-" class=" segue dark quote nobackground" style="background-image: url(nil)">
<aside class="gdbar right bottom"><img src="images/sukki-icon.png"></aside><hgroup class="">
<h2 class=" "><div id="outline-container-org6dad1ed" class="outline-2">
<h2 id="org6dad1ed">TensorForce 流程說明</h2>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="flexbox vleft auto-fadein" id="text-">
</article>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org783fff8" class="outline-4">
<h4 id="org783fff8">TensorForce 流程</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>TensorForce 的執行流程非常簡單</li>
</ul>
<ul>
<li>是在 TensorFlow 和 Oepn AI Gym 之間的連結橋梁</li>
</ul>
<ul>
<li>OpenAI Gym 提供 <code>Environment</code> 也就是 <code>gym_id</code></li>
</ul>
<ul>
<li>而 <code>TensorFlow</code> 被定義成 <code>Model</code> 讓 <code>Agent</code> 來初始化</li>
</ul>
<ul>
<li>透過建立一個集成物件 <code>Runner</code> 來進行 RL 訓練</li>
</ul>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org538ffc8" class="outline-4">
<h4 id="org538ffc8">TensorForce 流程</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<article class="flexbox vcenter">
<div class="figure">
<p><img src="images/tensorforce-intro.png" alt="tensorforce-intro.png" width="900px" />
</p>
</div>
</article>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org5542297" class="outline-4">
<h4 id="org5542297">TensorForce 流程 - 以 openai-gym.py 為例 (1/7)</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>我們可以讀 code : <a href="https://github.com/reinforceio/tensorforce/blob/master/examples/openai_gym.py">tensorforce/openai-gym.py - GitHub</a></li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="python">
import argparse # 1. 為了在 command-line 搭配不同 arg 執行
import json # 2. 讀取 Agent 以及 Network 設置
import logging # 3. 每個 epsiode 進行紀錄
import os # 4. 操作檔案路徑
import time # 5. 操作運算時間
from tensorforce import TensorForceError # 丟 Error
from tensorforce.agents import Agent # 建立 Agent
from tensorforce.execution import Runner # 建立 Runner
from tensorforce.contrib.openai_gym import OpenAIGym # 建立 Env</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgc9ecfe6" class="outline-4">
<h4 id="orgc9ecfe6">TensorForce 流程 - 以 openai-gym.py 為例 (2/7)</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li><code>monitor</code> 相關參數主要是為了 OpenAI Gym 的設置</li>
</ul>
<ul>
<li>建議不要輸出 Video ，可能遇到不知名錯誤而停止 (GUI Window 問題)</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="python">
# 建立環境 Environment
environment = OpenAIGym(
gym_id=args.gym_id, # Gym ID 就是你的特定環境
monitor=args.monitor, # 是否要輸出 Gym Results
monitor_safe=args.monitor_safe, # 是否要避免蓋掉之前的 Results
monitor_video=args.monitor_video # 是否要每隔幾步輸出影片(危險!)
)</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org4a1eb9d" class="outline-4">
<h4 id="org4a1eb9d">TensorForce 流程 - 以 openai-gym.py 為例 (3/7)</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>接著是從 JSON 讀取相關設置， <code>spec</code> 結尾的函數是對 <code>dict</code> 字典資料的處理</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="python">
if args.agent_config is not None:
<b>with open(args.agent_config, 'r') as fp:</b>
<b>agent_config = json.load(fp=fp)</b>
else:
raise TensorForceError("No agent configuration provided.")
if args.network_spec is not None:
<b>with open(args.network_spec, 'r') as fp:</b>
<b>network_spec = json.load(fp=fp)</b>
else:
network_spec = None
logger.info("No network configuration provided.")</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org722ffec" class="outline-4">
<h4 id="org722ffec">TensorForce 流程 - 以 openai-gym.py 為例 (4/7)</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>其中 Agent Config 的 JSON 大概長成這樣：</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="json">
{
"type": "vpg_agent", # RL Agent 名稱
"batch_size": 4000, # TensorFlow 中的 batch size
"optimizer": { # TensorFlow 中的 optimizer
"type": "adam", # TensorFlow 中的 optimize type
"learning_rate": 1e-2 # TensorFlow 中的 learning rate
},
"discount": 0.99, # TensorFlow 中的 discount factor
"entropy_regularization": null, # TensorFlow 中的... (略)
# ...(略)
}</pre>
</div>
<ul>
<li>總之這裡的設置都是 TensorFlow 的基本參數設置</li>
</ul>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org121145c" class="outline-4">
<h4 id="org121145c">TensorForce 流程 - 以 openai-gym.py 為例 (5/7)</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>而 <code>networ_spec</code> 讀取的 JSON 就是 TensorFlow Model 的設置：</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="json">
{
"type": "conv2d", "size": 32, "window": 8, "stride": 4
},
... (略)
{
"type": "flatten"
},
{
"type": "dense", "size": 512
}</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org186ff09" class="outline-4">
<h4 id="org186ff09">TensorForce 流程 - 以 openai-gym.py 為例 (6/7)</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>接著你需要定義一個 <code>epsiode_finished</code> 後面講 runner 會提到</li>
</ul>
<ul>
<li>主要是方便你可以在每個 epsiode 使用 <code>logger</code> 輸出迭代資訊</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="python">
def episode_finished(r):
if r.episode % report_episodes == 0:
steps_per_second = r.timestep / (time.time() - r.start_time)
logger.info("Finished episode {} after {} timesteps. Steps Per Second {}"
.format(r.agent.episode, r.episode_timestep, steps_per_second
))
return True</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgd79ba32" class="outline-4">
<h4 id="orgd79ba32">TensorForce 流程 - 以 openai-gym.py 為例 (7/7)</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>最後就是執行 <code>rnuner.run</code> 即可，並且搭配 <code>close</code> 完成整個流程</li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="python">
runner.run(
timesteps=args.timesteps,
episodes=args.episodes,
max_episode_timesteps=args.max_episode_timesteps,
deterministic=args.deterministic,
episode_finished=episode_finished
)
runner.close()</pre>
</div>
</article>
</slide>
</slide>
</slide>
<slide id="sec-" class=" segue dark quote nobackground" style="background-image: url(nil)">
<aside class="gdbar right bottom"><img src="images/sukki-icon.png"></aside><hgroup class="">
<h2 class=" "><div id="outline-container-org90295f7" class="outline-2">
<h2 id="org90295f7">TensorForce 深入剖析</h2>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="flexbox vleft auto-fadein" id="text-">
</article>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgf732655" class="outline-4">
<h4 id="orgf732655">Agent 以及 Model 的設置</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>在 TensorForce 的 RL 框架中，有兩個東西要留意，分別是 <code>Runner</code> 以及 <code>Model</code></li>
</ul>
<ul>
<li>Agent (代理) 並不是直接與 Environment (環境) 交互，是透過 Runner</li>
</ul>
<ul>
<li>Agent (代理) 可以有多個 Model ，例如 <code>Double-DQN</code> 就有兩個 <code>Q-Model</code></li>
</ul>
<article class="flexbox vcenter">
<div class="figure">
<p><img src="images/agent-and-model-view-in-tensorforce.png" alt="agent-and-model-view-in-tensorforce.png" width="800px" />
</p>
</div>
</article>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgff076ee" class="outline-4">
<h4 id="orgff076ee">Agent 類</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>Agent 是一個 class ，所有的 Agent 都繼承自這個 Class</li>
</ul>
<ul>
<li>在 TensorForce 中，大部分的 Agent 是指一種 RL 方法，例如 <code>DQNAgent</code></li>
</ul>
<ul>
<li>有些 Agent 要使用 Model 的歷史資訊（例如 RNN ）則要繼承自 <code>MemoryAgent</code></li>
</ul>
<ul>
<li>有些 Agent 是在 Model 的每個 Batch 做 Replay 則要繼承自 <code>BatchAgent</code></li>
</ul>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgb2645a6" class="outline-4">
<h4 id="orgb2645a6">Agent 類</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<article class="flexbox vcenter">
<div class="figure">
<p><img src="images/agent-class.png" alt="agent-class.png" width="800px" />
</p>
</div>
</article>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-org35f7765" class="outline-4">
<h4 id="org35f7765">Agent 類 - 以 DQNAgent 為例</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li><code>Agent</code> 本身主要放參數，詳細請見 <a href="https://github.com/reinforceio/tensorforce/blob/master/tensorforce/agents/dqn_agent.py">DQNAgent.py - GitHub</a></li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="python">
class DQNAgent(MemoryAgent):
def __init__(
# Agent 的參數
self, states_spec, actions_spec, batched_observe=None, scope='dqn', ...
# Learning 的參數
summary_spec=None, network_spec=None, device=None, ...
# DQNAgent 的特殊參數
target_sync_frequency=10000, target_update_weight=1.0,
double_q_model=False, huber_loss=None, ...</pre>
</div>
</article>
</slide>
</slide>
<slide id="sec-" >
<hgroup class="">
<h2 class=" "><div id="outline-container-orgeacb254" class="outline-4">
<h4 id="orgeacb254">Agent 類 - 以 DQNAgent 為例</h4>
</div>
</h2>
<h3></h3>
</hgroup>
<article class="" id="text-">
<ul>
<li>透過 <code>Agent</code> 來可以定義 Model 初始化函數 <code>initialize_model</code></li>
</ul>
<div class="org-src-container">
<pre class="prettyprint" data-lang="python">
def initialize_model(self):
return QModel(
states_spec=self.states_spec,
actions_spec=self.actions_spec,
network_spec=self.network_spec,
...
double_q_model=self.double_q_model,
huber_loss=self.huber_loss,
random_sampling_fix=True
)</pre>
</div>
</article>
</slide>
</slide>
</slide>
<slide id="sec-" class=" thank-you-slide segue nobackground" style="background-image: url(nil)">
<aside class="gdbar right"><img src="images/sukki-icon.png"></aside><article class="flexbox vleft auto-fadein" id="text-">
<h2>
<p>Thank You !</p>
</h2>
<br>
<p class="auto-fadein" data-config-contact>
</p>
</article>
</slide>
<slide class="backdrop"></slide>
</slides>
<!--[if IE]>
<script src="http://ajax.googleapis.com/ajax/libs/chrome-frame/1/CFInstall.min.js"></script>
<script>CFInstall.check({mode: 'overlay'});</script>
<![endif]-->
</body>
</html>