Skip to content

Instantly share code, notes, and snippets.

@gitu

gitu/autoblogger.ipynb

Created April 10, 2020 22:08
Show Gist options
  • Save gitu/fbcb6c0f9413849bcae42f36e274d27e to your computer and use it in GitHub Desktop.
Save gitu/fbcb6c0f9413849bcae42f36e274d27e to your computer and use it in GitHub Desktop.
Download ZIP
autoblogger - train gpt-2
Display the source blob
Display the rendered blob
Raw
autoblogger.ipynb
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[33mWARNING: You are using pip version 19.3.1; however, version 20.0.2 is available.\n",
"You should consider upgrading via the 'pip install --upgrade pip' command.\u001b[0m\n"
]
}
],
"source": [
"!pip install -q gpt-2-simple"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"WARNING:tensorflow:\n",
"The TensorFlow contrib module will not be included in TensorFlow 2.0.\n",
"For more information, please see:\n",
" * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md\n",
" * https://github.com/tensorflow/addons\n",
" * https://github.com/tensorflow/io (for I/O related ops)\n",
"If you depend on functionality not listed there, please file an issue.\n",
"\n",
"WARNING:tensorflow:From /usr/local/lib/python3.6/site-packages/gpt_2_simple/src/sample.py:17: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
"Instructions for updating:\n",
"Use tf.where in 2.0, which has the same broadcast rule as np.where\n",
"WARNING:tensorflow:From /usr/local/lib/python3.6/site-packages/gpt_2_simple/src/memory_saving_gradients.py:62: get_backward_walk_ops (from tensorflow.contrib.graph_editor.select) is deprecated and will be removed after 2019-06-06.\n",
"Instructions for updating:\n",
"Please use tensorflow.python.ops.op_selector.get_backward_walk_ops.\n",
"Loading checkpoint models/774M/model.ckpt\n",
"INFO:tensorflow:Restoring parameters from models/774M/model.ckpt\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
" 0%| | 0/1 [00:00<?, ?it/s]"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Loading dataset...\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|██████████| 1/1 [00:00<00:00, 2.57it/s]\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"dataset has 28417 tokens\n",
"Training...\n",
"[1 | 23.07] loss=2.90 avg=2.90\n",
"[2 | 24.07] loss=2.95 avg=2.93\n",
"[3 | 25.06] loss=2.83 avg=2.89\n",
"[4 | 26.05] loss=2.95 avg=2.91\n",
"[5 | 27.05] loss=3.20 avg=2.97\n",
"[6 | 28.04] loss=3.05 avg=2.98\n",
"[7 | 29.03] loss=2.91 avg=2.97\n",
"[8 | 30.03] loss=2.73 avg=2.94\n",
"[9 | 31.02] loss=3.05 avg=2.95\n",
"[10 | 32.01] loss=3.66 avg=3.03\n",
"[11 | 33.01] loss=2.50 avg=2.98\n",
"[12 | 34.00] loss=2.83 avg=2.96\n",
"[13 | 34.99] loss=2.88 avg=2.96\n",
"[14 | 35.99] loss=2.85 avg=2.95\n",
"[15 | 36.98] loss=2.71 avg=2.93\n",
"[16 | 37.97] loss=3.12 avg=2.94\n",
"[17 | 38.97] loss=2.94 avg=2.94\n",
"[18 | 39.96] loss=3.66 avg=2.99\n",
"[19 | 40.95] loss=2.68 avg=2.97\n",
"[20 | 41.94] loss=3.30 avg=2.99\n",
"[21 | 42.94] loss=2.89 avg=2.98\n",
"[22 | 43.93] loss=3.04 avg=2.99\n",
"[23 | 44.93] loss=3.23 avg=3.00\n",
"[24 | 45.92] loss=3.49 avg=3.02\n",
"[25 | 46.92] loss=2.64 avg=3.00\n",
"[26 | 47.92] loss=2.85 avg=3.00\n",
"[27 | 48.91] loss=2.46 avg=2.97\n",
"[28 | 49.91] loss=3.24 avg=2.99\n",
"[29 | 50.90] loss=2.74 avg=2.98\n",
"[30 | 51.90] loss=2.58 avg=2.96\n",
"[31 | 52.89] loss=3.14 avg=2.97\n",
"[32 | 53.88] loss=2.47 avg=2.95\n",
"[33 | 54.88] loss=3.00 avg=2.95\n",
"[34 | 55.87] loss=2.49 avg=2.93\n",
"[35 | 56.87] loss=2.99 avg=2.94\n",
"[36 | 57.86] loss=2.18 avg=2.91\n",
"[37 | 58.86] loss=3.42 avg=2.93\n",
"[38 | 59.85] loss=2.56 avg=2.92\n",
"[39 | 60.84] loss=2.15 avg=2.89\n",
"[40 | 61.84] loss=3.03 avg=2.90\n",
"[41 | 62.84] loss=2.36 avg=2.88\n",
"[42 | 63.83] loss=2.50 avg=2.87\n",
"[43 | 64.83] loss=3.51 avg=2.89\n",
"[44 | 65.82] loss=3.46 avg=2.90\n",
"[45 | 66.82] loss=1.97 avg=2.88\n",
"[46 | 67.82] loss=2.16 avg=2.86\n",
"[47 | 68.81] loss=3.19 avg=2.87\n",
"[48 | 69.81] loss=2.85 avg=2.87\n",
"[49 | 70.80] loss=2.86 avg=2.87\n",
"[50 | 71.80] loss=3.28 avg=2.88\n",
"[51 | 72.79] loss=2.68 avg=2.87\n",
"[52 | 73.79] loss=1.94 avg=2.85\n",
"[53 | 74.79] loss=3.14 avg=2.86\n",
"[54 | 75.78] loss=2.73 avg=2.85\n",
"[55 | 76.78] loss=2.24 avg=2.84\n",
"[56 | 77.77] loss=3.24 avg=2.85\n",
"[57 | 78.77] loss=1.90 avg=2.83\n",
"[58 | 79.76] loss=2.49 avg=2.82\n",
"[59 | 80.76] loss=2.39 avg=2.81\n",
"[60 | 81.76] loss=2.49 avg=2.80\n",
"[61 | 82.75] loss=3.02 avg=2.81\n",
"[62 | 83.75] loss=2.76 avg=2.81\n",
"[63 | 84.75] loss=1.85 avg=2.79\n",
"[64 | 85.74] loss=2.54 avg=2.78\n",
"[65 | 86.74] loss=2.67 avg=2.78\n",
"[66 | 87.73] loss=1.85 avg=2.76\n",
"[67 | 88.73] loss=2.39 avg=2.75\n",
"[68 | 89.72] loss=1.64 avg=2.73\n",
"[69 | 90.72] loss=2.37 avg=2.72\n",
"[70 | 91.72] loss=2.77 avg=2.72\n",
"[71 | 92.72] loss=2.48 avg=2.72\n",
"[72 | 93.71] loss=2.36 avg=2.71\n",
"[73 | 94.71] loss=2.33 avg=2.70\n",
"[74 | 95.71] loss=2.05 avg=2.69\n",
"[75 | 96.71] loss=2.82 avg=2.69\n",
"[76 | 97.71] loss=2.24 avg=2.69\n",
"[77 | 98.70] loss=2.14 avg=2.68\n",
"[78 | 99.70] loss=1.47 avg=2.65\n",
"[79 | 100.70] loss=1.98 avg=2.64\n",
"[80 | 101.69] loss=2.50 avg=2.64\n",
"[81 | 102.69] loss=1.96 avg=2.63\n",
"[82 | 103.69] loss=2.77 avg=2.63\n",
"[83 | 104.69] loss=2.19 avg=2.62\n",
"[84 | 105.69] loss=1.89 avg=2.61\n",
"[85 | 106.70] loss=1.78 avg=2.59\n",
"[86 | 107.70] loss=2.48 avg=2.59\n",
"[87 | 108.70] loss=3.02 avg=2.60\n",
"[88 | 109.70] loss=2.33 avg=2.59\n",
"[89 | 110.70] loss=3.08 avg=2.60\n",
"[90 | 111.70] loss=2.28 avg=2.60\n",
"[91 | 112.70] loss=2.90 avg=2.60\n",
"[92 | 113.70] loss=1.94 avg=2.59\n",
"[93 | 114.70] loss=1.89 avg=2.58\n",
"[94 | 115.70] loss=2.43 avg=2.58\n",
"[95 | 116.70] loss=3.23 avg=2.59\n",
"[96 | 117.70] loss=3.08 avg=2.60\n",
"[97 | 118.70] loss=2.53 avg=2.59\n",
"[98 | 119.70] loss=2.54 avg=2.59\n",
"[99 | 120.70] loss=1.27 avg=2.57\n",
"[100 | 121.70] loss=2.72 avg=2.58\n",
"======== SAMPLE 1 ========\n",
" is a new feature that allows you to see the details of every HTTP request you've made with your Web application so far.\n",
"I'm pleased to announce that we have just released our new High-Resolution Web Profile!\n",
"In general terms, this new feature shows a high-resolution web representation of your Web application in the browser, so that it's evident that you're only a few MB away from a complete web experience, even if you're at a very low resolution (e.g., 1920 x 1080). This gives you an idea of the performance and scalability of a typical site.\n",
"As with many new features, high-resolution web profiling wasn't planned from the beginning. In fact, its early implementation was a direct reaction to the need to understand the impact of high-resolution web profiling as a toolset for improving user experience.\n",
"From our feedback, we know that people really value being able to see precisely what's happening during Web application development by actually experiencing the Web applications at low-resolution (e.g., lower than 1280 x 720). In fact, even more people value this feature when implemented in a way that supports multiple user agents; to that end, we've been working hard for the last two years to enable multiple user agents to support high-resolution web profiling.\n",
"Now, with the launch of the High-Resolution Web Profile, high-resolution profiling is available for all major user agents, including:\n",
"Safari 4.0+\n",
"IE 7+ (Internet Explorer 7 does not support the High-Resolution Profile functionality)\n",
"Apple Safari (only Safari 2.1 and up)\n",
"With this launch, all user agents can now create and submit web profiling reports to Jenkins, in formats that support multiple user agents. With the High-Resolution Web Profile in place, the results reported to Jenkins will be in an easily legible form and will include additional granularity to better describe the performance and resource consumption of your applications.\n",
"As you can see from the image above, the High-Resolution Web Profile details the detail provided by every (single source line of code) request to your application.\n",
"The Jenkins JIRA profile is also fully compatible with the new JSON-based Web Report format, which allows you to submit your Web report to a wide range of applications. You will notice in the image above that the Web Report will now include a detailed breakdown of the requests, in the form of JSON objects.\n",
"We're really excited to be able to provide users this powerful new capability for improved web performance analysis. As you'll see below in the section \"What is the Google Web Console Data Quality Toolset Like?\" above, we have a range of other new reports available as well, including Google Cloud Console Alerts, Azure Resource Manager Alerts, Google Cloud BigQuery Analytics, Google Cloud Datastore Dashboards, Google Cloud Dataflow Reports, MapReduce dashboards, WebLogic Enterprise Dashboards, and JSON. Analytics Logs, and JSON-based Web Report are the most visible new capabilities.\n",
"Here's a look at all the capabilities available with the new JSON-based Web Report:\n",
"Annotations (the raw JSON data that describe the HTTP requests you've recorded) and Caching Information (how many records we've retained per response) are available in an XML-based form.\n",
"A full list of the reports you can submit can be found on the Google Web Console Data Quality Toolset page.\n",
"Google Web Console Data Quality toolset\n",
"Analytics Logs (an in-depth breakdown of every single page view)\n",
"Google Cloud Dataflow Reports (analyze and summarize all the data transformations done via Cloud Datastore)\n",
"Web Report (explain the detailed HTML, JS, XML, JSON, and more that drive each request and response)\n",
"Cloud Dataflow Dashboards provide a more granular look at the details of Cloud Datastore data flows\n",
"JSON-based Web Reports (report the granular details as to how the JSON content is transmitted between Cloud Datastore and your application),\n",
"MapReduce Dashboards provide detailed insights into your Cloud Data Factory workloads and data warehouse operations\n",
"If you're using Google Cloud Datastore in Google Cloud Platform, you'll find that the JSON-based reports are already incorporated into your API and Cloud Platform dashboards.\n",
"In the next few weeks, we'll also begin to release annotated JSON and XML logs from Cloud Datastore API calls onto GitHub, so that you can understand exactly what happened behind the scenes. As always, please comment with any questions or suggestions in the Google Cloud Platform High-Resolution Web Profile forum thread.\n",
"Why is the Google Web Console Data Quality Toolset useful?\n",
"The Web Console presents data in a number of useful formats. The first one that comes to mind is JSON, so in a sense the Web Console is the perfect place to have a deeper understanding of JSON data. However, the JSON available in all the Web Console reports is somewhat inconsistent, which means that you may\n",
"\n",
"[101 | 155.16] loss=2.21 avg=2.57\n",
"[102 | 156.15] loss=2.62 avg=2.57\n",
"[103 | 157.16] loss=2.33 avg=2.57\n",
"[104 | 158.16] loss=2.62 avg=2.57\n",
"[105 | 159.15] loss=2.04 avg=2.56\n",
"[106 | 160.15] loss=1.67 avg=2.55\n",
"[107 | 161.15] loss=2.25 avg=2.54\n",
"[108 | 162.15] loss=2.17 avg=2.54\n",
"[109 | 163.15] loss=2.24 avg=2.53\n",
"[110 | 164.15] loss=2.02 avg=2.52\n",
"[111 | 165.16] loss=2.52 avg=2.52\n",
"[112 | 166.16] loss=2.21 avg=2.52\n",
"[113 | 167.16] loss=1.79 avg=2.51\n",
"[114 | 168.16] loss=2.04 avg=2.50\n",
"[115 | 169.17] loss=2.33 avg=2.50\n",
"[116 | 170.16] loss=2.01 avg=2.49\n",
"[117 | 171.17] loss=2.48 avg=2.49\n",
"[118 | 172.17] loss=2.45 avg=2.49\n",
"[119 | 173.17] loss=2.00 avg=2.48\n",
"[120 | 174.17] loss=1.76 avg=2.47\n",
"[121 | 175.17] loss=1.53 avg=2.46\n",
"[122 | 176.17] loss=1.97 avg=2.45\n",
"[123 | 177.17] loss=1.55 avg=2.44\n",
"[124 | 178.18] loss=1.82 avg=2.43\n",
"[125 | 179.18] loss=1.97 avg=2.43\n",
"[126 | 180.18] loss=1.26 avg=2.41\n",
"[127 | 181.18] loss=2.01 avg=2.40\n",
"[128 | 182.19] loss=1.98 avg=2.40\n",
"[129 | 183.19] loss=1.36 avg=2.38\n",
"[130 | 184.19] loss=2.30 avg=2.38\n",
"[131 | 185.19] loss=1.34 avg=2.37\n",
"[132 | 186.20] loss=1.21 avg=2.35\n",
"[133 | 187.20] loss=1.80 avg=2.34\n",
"[134 | 188.20] loss=2.15 avg=2.34\n",
"[135 | 189.21] loss=3.04 avg=2.35\n",
"[136 | 190.21] loss=1.11 avg=2.33\n",
"[137 | 191.21] loss=1.68 avg=2.33\n",
"[138 | 192.22] loss=1.89 avg=2.32\n",
"[139 | 193.22] loss=1.70 avg=2.31\n",
"[140 | 194.23] loss=1.56 avg=2.30\n",
"[141 | 195.23] loss=1.18 avg=2.29\n",
"[142 | 196.23] loss=2.43 avg=2.29\n",
"[143 | 197.23] loss=1.55 avg=2.28\n",
"[144 | 198.24] loss=2.52 avg=2.28\n",
"[145 | 199.24] loss=1.87 avg=2.28\n",
"[146 | 200.25] loss=1.06 avg=2.26\n",
"[147 | 201.25] loss=1.03 avg=2.25\n",
"[148 | 202.26] loss=1.98 avg=2.24\n",
"[149 | 203.26] loss=2.06 avg=2.24\n",
"[150 | 204.27] loss=1.04 avg=2.22\n",
"[151 | 205.27] loss=0.87 avg=2.21\n",
"[152 | 206.28] loss=1.89 avg=2.20\n",
"[153 | 207.28] loss=1.73 avg=2.20\n",
"[154 | 208.28] loss=0.76 avg=2.18\n",
"[155 | 209.29] loss=1.24 avg=2.17\n",
"[156 | 210.29] loss=1.23 avg=2.15\n",
"[157 | 211.29] loss=1.34 avg=2.14\n",
"[158 | 212.30] loss=0.60 avg=2.13\n",
"[159 | 213.30] loss=1.10 avg=2.11\n",
"[160 | 214.31] loss=0.94 avg=2.10\n",
"[161 | 215.32] loss=1.65 avg=2.09\n",
"[162 | 216.32] loss=0.72 avg=2.08\n",
"[163 | 217.33] loss=1.44 avg=2.07\n",
"[164 | 218.34] loss=1.58 avg=2.06\n",
"[165 | 219.34] loss=2.33 avg=2.06\n",
"[166 | 220.35] loss=1.57 avg=2.06\n",
"[167 | 221.36] loss=1.76 avg=2.05\n",
"[168 | 222.36] loss=0.90 avg=2.04\n",
"[169 | 223.37] loss=1.25 avg=2.03\n",
"[170 | 224.38] loss=0.91 avg=2.02\n",
"[171 | 225.38] loss=1.73 avg=2.01\n",
"[172 | 226.39] loss=0.83 avg=2.00\n",
"[173 | 227.40] loss=1.01 avg=1.99\n",
"[174 | 228.40] loss=2.07 avg=1.99\n",
"[175 | 229.41] loss=0.68 avg=1.97\n",
"[176 | 230.42] loss=1.91 avg=1.97\n",
"[177 | 231.43] loss=1.54 avg=1.97\n",
"[178 | 232.44] loss=0.72 avg=1.95\n",
"[179 | 233.44] loss=1.20 avg=1.94\n",
"[180 | 234.45] loss=1.10 avg=1.93\n",
"[181 | 235.46] loss=1.35 avg=1.93\n",
"[182 | 236.46] loss=0.62 avg=1.91\n",
"[183 | 237.47] loss=0.77 avg=1.90\n",
"[184 | 238.48] loss=0.75 avg=1.88\n",
"[185 | 239.49] loss=1.62 avg=1.88\n",
"[186 | 240.50] loss=1.22 avg=1.87\n",
"[187 | 241.50] loss=1.30 avg=1.87\n",
"[188 | 242.51] loss=0.47 avg=1.85\n",
"[189 | 243.52] loss=1.20 avg=1.84\n",
"[190 | 244.52] loss=0.57 avg=1.83\n",
"[191 | 245.53] loss=0.71 avg=1.81\n",
"[192 | 246.54] loss=0.83 avg=1.80\n",
"[193 | 247.54] loss=1.54 avg=1.80\n",
"[194 | 248.55] loss=0.39 avg=1.78\n",
"[195 | 249.56] loss=1.31 avg=1.78\n",
"[196 | 250.57] loss=0.33 avg=1.76\n",
"[197 | 251.58] loss=1.41 avg=1.76\n",
"[198 | 252.59] loss=1.01 avg=1.75\n",
"[199 | 253.60] loss=1.17 avg=1.74\n",
"[200 | 254.60] loss=0.40 avg=1.72\n",
"======== SAMPLE 1 ========\n",
" audit tool that helps organizations better manage digital transformation.\n",
"To be notified of upcoming speaker dates and times, please sign up for updates here.<|endoftext|>This is a rush transcript. Copy may not be in its final form.\n",
"\n",
"AMY GOODMAN: With the election a week away, Democracy Now! is going deeper into the state of the presidential race. We're in New Hampshire, the home state of presidential hopeful, Senator Bernie Sanders. And we're joined by Professor Larry Bartels and Dr. Jill Lepore, the co-founders of the Presidential Campaign Risk Assessment Project.\n",
"\n",
"We welcome you to Democracy Now! Dr. Jill, thanks so much for joining us. You are a professor of political science and public affairs at Loyola Maryland and director of the Presidential Campaign Risk Assessment Project. Can you talk about the start of your study? Have you gone back from October, when you started, to see what has changed?\n",
"\n",
"DR. JILL LEPORE: Sure. So, we're in New Hampshire, which is not that different from last fall. But we had a lot more data than we had last fall. So, we had, for example, a lot more surveys and focus groups because we did a very large-number survey for the PBS NewsHour and we did focus groups with all the presidential candidates. So, we have a lot more data. And the most significant thing we noticed was that we didn't have as many polls out then. And the reason we don't have as many polls is that most people don't pay attention to polls anymore, which is why we need a lot more samples.\n",
" the POLLS DON'T MATTER study released today. So, we have a new study published today, which is based on an amazing study by Professor Alan Abramowitz and my colleague Jonathan Page, who did the first studies showing that polls don't matter. This latest study shows that the polls matter way more than people think they do. But that is what the media wants you to take away from the report. They want you to believe the polls are bad, when actually they're good. They mean well, but they're just not being used in practice. They say that the new study is better than the first one because it's peer reviewed. And they actually have the study, which was done by four distinguished political scientists. They say, \"You know what? We think this study is more convincing than the first one,\" which is a little bit depressing for poll advocates.\n",
"\n",
"JUAN GONZÁLEZ: Well, professor Lepore, you said that you had a lot more data from this study. Tell us about that.\n",
"\n",
"DR. JILL LEPORE: Well, we had all our surveys in December, January, February, and March. And so, we had all the polls. So, we had CNN/ORC International, Gallup International, Pew Research, Pew Hispanic Center, Rasmussen Reports. We also have all the polls from the last seven or eight days. And we showed that polls don't matter. We showed that only 25 percent of people say that their opinion of their local polling organization is very important or rather that they're somewhat or very important to them. Of course, that's lower than the average percentage who say that, but the point is that in every group except for those who haven't participated in polling in a long time, less than half agree with that statement.\n",
"\n",
"AMY GOODMAN: Talk about this quote, \"The poll is the best predictor of the future as far as I'm concerned.\"\n",
"\n",
"DR. JILL LEPORE: This is from a 2011 survey where a total of 23,000 people were asked, \"Is your assessment of how influential the news media are in your area somewhat or very important to you?.\" And it showed that when you control for other factors like education, party identification, gender and income, the poll says, \"The poll is the best predictor we have of the future importance of the news media in your area.\" And what that means is that what news organizations do, what we see happen, is that what happens before is very very predictable.\n",
"\n",
"So that leads us to the second thing that happens. What most people fail to realize is in order to have a poll you have to have people talk to people. People can't do that if they can't see the other people and they can't hear the other people. People can't take care of their own business if they don't see it and have people back of it. And polls tell you in a very clear and simple way if a lot of other people around you care about what you care about. It's so clear in fact, that if you ask a group of people whether or not it's very important to them that a major political office holder in their community have a college degree, you know, in the vast majority of cases those people, when they look at the poll, will say yes.\n",
"\n",
"AMY GOODMAN: And,\n",
"\n",
"[201 | 281.91] loss=0.47 avg=1.71\n",
"[202 | 282.92] loss=0.95 avg=1.70\n",
"[203 | 283.93] loss=2.26 avg=1.71\n",
"[204 | 284.93] loss=1.53 avg=1.71\n",
"[205 | 285.93] loss=1.15 avg=1.70\n",
"[206 | 286.93] loss=0.95 avg=1.69\n",
"[207 | 287.94] loss=1.09 avg=1.68\n",
"[208 | 288.94] loss=1.06 avg=1.68\n",
"[209 | 289.95] loss=1.16 avg=1.67\n",
"[210 | 290.95] loss=0.93 avg=1.66\n",
"[211 | 291.95] loss=0.87 avg=1.65\n",
"[212 | 292.96] loss=1.04 avg=1.65\n",
"[213 | 293.97] loss=0.65 avg=1.64\n",
"[214 | 294.97] loss=0.62 avg=1.62\n",
"[215 | 295.98] loss=0.96 avg=1.62\n",
"[216 | 296.99] loss=0.91 avg=1.61\n",
"[217 | 297.99] loss=0.67 avg=1.60\n",
"[218 | 299.00] loss=0.57 avg=1.59\n",
"[219 | 300.01] loss=1.27 avg=1.58\n",
"[220 | 301.02] loss=0.93 avg=1.58\n",
"[221 | 302.02] loss=0.67 avg=1.57\n",
"[222 | 303.03] loss=0.45 avg=1.55\n",
"[223 | 304.03] loss=0.51 avg=1.54\n",
"[224 | 305.04] loss=0.46 avg=1.53\n",
"[225 | 306.05] loss=0.33 avg=1.52\n",
"[226 | 307.05] loss=1.22 avg=1.51\n",
"[227 | 308.06] loss=0.83 avg=1.50\n",
"[228 | 309.07] loss=0.49 avg=1.49\n",
"[229 | 310.08] loss=0.38 avg=1.48\n",
"[230 | 311.09] loss=0.49 avg=1.47\n",
"[231 | 312.10] loss=0.78 avg=1.46\n",
"[232 | 313.11] loss=1.85 avg=1.47\n",
"[233 | 314.11] loss=0.27 avg=1.45\n",
"[234 | 315.12] loss=0.62 avg=1.44\n",
"[235 | 316.13] loss=1.06 avg=1.44\n",
"[236 | 317.14] loss=0.24 avg=1.43\n",
"[237 | 318.14] loss=1.01 avg=1.42\n",
"[238 | 319.15] loss=0.28 avg=1.41\n",
"[239 | 320.16] loss=0.60 avg=1.40\n",
"[240 | 321.16] loss=0.27 avg=1.39\n",
"[241 | 322.17] loss=0.49 avg=1.38\n",
"[242 | 323.18] loss=0.87 avg=1.37\n",
"[243 | 324.19] loss=0.50 avg=1.36\n",
"[244 | 325.19] loss=0.64 avg=1.36\n",
"[245 | 326.20] loss=0.56 avg=1.35\n",
"[246 | 327.21] loss=0.72 avg=1.34\n",
"[247 | 328.22] loss=0.28 avg=1.33\n",
"[248 | 329.22] loss=0.39 avg=1.32\n",
"[249 | 330.23] loss=1.44 avg=1.32\n",
"[250 | 331.24] loss=0.26 avg=1.31\n",
"[251 | 332.24] loss=0.47 avg=1.30\n",
"[252 | 333.25] loss=0.19 avg=1.29\n",
"[253 | 334.27] loss=0.22 avg=1.28\n",
"[254 | 335.27] loss=0.35 avg=1.27\n",
"[255 | 336.28] loss=0.17 avg=1.25\n",
"[256 | 337.29] loss=0.44 avg=1.24\n",
"[257 | 338.30] loss=0.25 avg=1.23\n",
"[258 | 339.31] loss=0.69 avg=1.23\n",
"[259 | 340.32] loss=0.17 avg=1.22\n",
"[260 | 341.33] loss=0.39 avg=1.21\n",
"[261 | 342.34] loss=0.47 avg=1.20\n",
"[262 | 343.34] loss=0.25 avg=1.19\n",
"[263 | 344.35] loss=0.21 avg=1.18\n",
"[264 | 345.36] loss=0.34 avg=1.17\n",
"[265 | 346.37] loss=0.75 avg=1.17\n",
"[266 | 347.38] loss=0.61 avg=1.16\n",
"[267 | 348.39] loss=0.21 avg=1.15\n",
"[268 | 349.40] loss=0.33 avg=1.14\n",
"[269 | 350.40] loss=0.17 avg=1.13\n",
"[270 | 351.41] loss=0.71 avg=1.13\n",
"[271 | 352.42] loss=0.60 avg=1.12\n",
"[272 | 353.43] loss=1.20 avg=1.12\n",
"[273 | 354.43] loss=0.19 avg=1.11\n",
"[274 | 355.44] loss=0.31 avg=1.10\n",
"[275 | 356.45] loss=0.20 avg=1.09\n",
"[276 | 357.46] loss=0.18 avg=1.08\n",
"[277 | 358.46] loss=0.18 avg=1.07\n",
"[278 | 359.47] loss=0.36 avg=1.07\n",
"[279 | 360.48] loss=1.13 avg=1.07\n",
"[280 | 361.49] loss=0.53 avg=1.06\n",
"[281 | 362.50] loss=0.92 avg=1.06\n",
"[282 | 363.51] loss=0.26 avg=1.05\n",
"[283 | 364.52] loss=0.83 avg=1.05\n",
"[284 | 365.53] loss=0.39 avg=1.04\n",
"[285 | 366.54] loss=0.36 avg=1.03\n",
"[286 | 367.55] loss=0.78 avg=1.03\n",
"[287 | 368.56] loss=0.16 avg=1.02\n",
"[288 | 369.57] loss=0.41 avg=1.02\n",
"[289 | 370.58] loss=0.29 avg=1.01\n",
"[290 | 371.59] loss=0.53 avg=1.00\n",
"[291 | 372.59] loss=0.14 avg=0.99\n",
"[292 | 373.61] loss=0.11 avg=0.98\n",
"[293 | 374.62] loss=0.37 avg=0.98\n",
"[294 | 375.62] loss=0.18 avg=0.97\n",
"[295 | 376.63] loss=0.34 avg=0.96\n",
"[296 | 377.64] loss=0.64 avg=0.96\n",
"[297 | 378.65] loss=0.70 avg=0.96\n",
"[298 | 379.65] loss=0.59 avg=0.95\n",
"[299 | 380.66] loss=0.39 avg=0.95\n",
"[300 | 381.67] loss=0.46 avg=0.94\n",
"======== SAMPLE 1 ========\n",
" as we learn to anticipate and mitigate risks, and to make adjustments as events occur?then it will pay dividends.\n",
"To understand why, just consider another story: Imagine you operate a grocery business, and your competitors operate a restaurant business?and you want to gain a competitive advantage. Can you imagine how important it would be to gain a competitive edge over your competitors?by outspending them on advertising?or developing a superior menu, delivery system, or other key e nteractive RACE?or even better, getting their customers to splurge, forego products, and spend more?\n",
"How about a digital asset like a digital wallet for your e-debit or e-credit cards? Or a popular sports betting services?premium user base?Imagine all these opportunities if you could develop/march in the lead-time (i.e. invest in technology, not in a business model) quadrant of the asset class?M&As, where you have the best opportunity due to superior execution, lower cost competitors, and growth prospects.\n",
"Now, imagine we told you that your competitor also has a leading lead-time, digital asset division?and that your product does not perform as well?despite having the same key selling points?as theirs. How likely is it that you would enter an agreement to collaborate on a joint venture or joint venture-like arrangement?like a BrandX or Digimarc partnership?or even a cash acquisition?\n",
"The reality is, most companies do not know they do not currently have a need for digital asset management (DAM) services, or that they should seriously consider a standalone digital asset management offering. It appears many firms specialize in a specific use case, such as:\n",
"Regulation management - We either face new taxes or stricter KYC/AML rules, or we see more government regulation affecting financial services as a whole.\n",
"- We face new taxes or stricter KYC/AML rules, or we see more government regulation affecting financial services as a whole. Operational excellence - We want to continue to deliver high-quality service to our customers, even as we expand to a new function or deliver a new feature.\n",
"- We want to continue to deliver high-quality service to our customers, even as we expand to a new function or deliver a new feature. Training and development - We need to grow our employee base while retaining the skill sets we have been sowing for the last ten years.\n",
"- We need to grow our employee base while retaining the skill sets we have been sowing for the last ten years. Executives? day to day operations - We want to continue to deliver valuable service to our CEO, CFO, and other senior leadership, even as we expand our functions or deliver a new service.\n",
"- We want to continue to deliver valuable service to our CEO, CFO, and other senior leadership, even as we expand our functions or deliver a new service, experts? eyes wide open - We do not want to find ourselves serving an exec or chief marketing officer (CTO) with new questions or requirements, who has to learn a new technology application or programming language.\n",
"While our examples illustrate an extreme case, the point is that most businesses do not know they do not currently have a need for a digital asset management service. This is unfortunate, as the development of highly specialized and expensive tooling to support the business needs would signal to the business that the business does have needs, but that a more traditional IT approach is not delivering on those needs. On the other hand, the announcement that a business has a need for a digital asset management service shows that it is evaluating technology options, and provides some degree of confidence that the technology is mature, has well-defined use cases, and could be delivered in a cost-effective manner.\n",
"This business opportunity exists because firms know they have a data question. They know the data is not evenly distributed between the various business functions. They know there is a disparity between the types of data that are normally associated with certain uses and those that are not. They know that standards for the production and handling of some of the more salutary data assets (bonds, customer profiles, historical quotes, asset class profiles) have not kept up with the changing, ever-evolving nature of the business.\n",
"They have analyzed their data and know they need a service to move it from raw data into structured and unstructured data assets that can be easily analyzed and understood. However, they do not yet know how to approach the challenge of the business seeking help. They do not yet know how to structure the exercise to make it understandable and attractive to potential customers.\n",
"The point is, even though they know they have a data question, many firms do not yet know the right question to ask. They need to hire a team to ask the right questions and train employees on the right answers. This will go a long way towards restoring business to confidence in their data and enabling them to reassess their business operations.\n",
"In my\n",
"\n",
"[301 | 408.94] loss=0.29 avg=0.93\n",
"[302 | 409.94] loss=0.18 avg=0.93\n",
"[303 | 410.94] loss=0.16 avg=0.92\n",
"[304 | 411.95] loss=0.79 avg=0.92\n",
"[305 | 412.95] loss=0.29 avg=0.91\n",
"[306 | 413.95] loss=0.40 avg=0.91\n",
"[307 | 414.96] loss=0.65 avg=0.90\n",
"[308 | 415.96] loss=0.28 avg=0.90\n",
"[309 | 416.97] loss=0.62 avg=0.89\n",
"[310 | 417.97] loss=0.21 avg=0.89\n",
"[311 | 418.98] loss=0.38 avg=0.88\n",
"[312 | 419.98] loss=0.18 avg=0.87\n",
"[313 | 420.99] loss=0.14 avg=0.87\n",
"[314 | 421.99] loss=0.10 avg=0.86\n",
"[315 | 423.00] loss=0.10 avg=0.85\n",
"[316 | 424.00] loss=0.24 avg=0.84\n",
"[317 | 425.00] loss=0.43 avg=0.84\n",
"[318 | 426.01] loss=0.28 avg=0.83\n",
"[319 | 427.01] loss=0.35 avg=0.83\n",
"[320 | 428.02] loss=0.24 avg=0.82\n",
"[321 | 429.02] loss=0.24 avg=0.82\n",
"[322 | 430.03] loss=0.13 avg=0.81\n",
"[323 | 431.03] loss=0.14 avg=0.80\n",
"[324 | 432.04] loss=0.18 avg=0.80\n",
"[325 | 433.05] loss=0.71 avg=0.80\n",
"[326 | 434.05] loss=0.42 avg=0.79\n",
"[327 | 435.05] loss=0.11 avg=0.78\n",
"[328 | 436.06] loss=0.15 avg=0.78\n",
"[329 | 437.06] loss=0.24 avg=0.77\n",
"[330 | 438.06] loss=0.09 avg=0.76\n",
"[331 | 439.07] loss=0.19 avg=0.76\n",
"[332 | 440.08] loss=0.15 avg=0.75\n",
"[333 | 441.08] loss=0.35 avg=0.75\n",
"[334 | 442.09] loss=1.27 avg=0.75\n",
"[335 | 443.09] loss=0.21 avg=0.75\n",
"[336 | 444.10] loss=0.10 avg=0.74\n",
"[337 | 445.10] loss=0.17 avg=0.74\n",
"[338 | 446.11] loss=0.34 avg=0.73\n",
"[339 | 447.12] loss=0.26 avg=0.73\n",
"[340 | 448.12] loss=0.44 avg=0.72\n",
"[341 | 449.13] loss=0.29 avg=0.72\n",
"[342 | 450.13] loss=0.24 avg=0.71\n",
"[343 | 451.14] loss=0.11 avg=0.71\n",
"[344 | 452.14] loss=0.15 avg=0.70\n",
"[345 | 453.15] loss=0.20 avg=0.70\n",
"[346 | 454.15] loss=1.29 avg=0.70\n",
"[347 | 455.16] loss=0.18 avg=0.70\n",
"[348 | 456.17] loss=0.33 avg=0.69\n",
"[349 | 457.17] loss=0.15 avg=0.69\n",
"[350 | 458.18] loss=0.29 avg=0.68\n",
"[351 | 459.19] loss=0.67 avg=0.68\n",
"[352 | 460.19] loss=0.24 avg=0.68\n",
"[353 | 461.20] loss=0.11 avg=0.67\n",
"[354 | 462.21] loss=0.27 avg=0.67\n",
"[355 | 463.21] loss=0.21 avg=0.66\n",
"[356 | 464.22] loss=0.37 avg=0.66\n",
"[357 | 465.23] loss=0.39 avg=0.66\n",
"[358 | 466.24] loss=0.38 avg=0.66\n",
"[359 | 467.24] loss=0.17 avg=0.65\n",
"[360 | 468.25] loss=0.11 avg=0.65\n",
"[361 | 469.26] loss=0.12 avg=0.64\n",
"[362 | 470.27] loss=0.12 avg=0.63\n",
"[363 | 471.28] loss=0.17 avg=0.63\n",
"[364 | 472.28] loss=0.85 avg=0.63\n",
"[365 | 473.29] loss=0.21 avg=0.63\n",
"[366 | 474.30] loss=0.24 avg=0.62\n",
"[367 | 475.31] loss=0.18 avg=0.62\n",
"[368 | 476.31] loss=0.66 avg=0.62\n",
"[369 | 477.32] loss=0.48 avg=0.62\n",
"[370 | 478.33] loss=0.11 avg=0.61\n",
"[371 | 479.34] loss=0.28 avg=0.61\n",
"[372 | 480.35] loss=0.18 avg=0.61\n",
"[373 | 481.36] loss=0.15 avg=0.60\n",
"[374 | 482.36] loss=0.24 avg=0.60\n",
"[375 | 483.37] loss=0.33 avg=0.59\n",
"[376 | 484.38] loss=0.28 avg=0.59\n",
"[377 | 485.39] loss=0.20 avg=0.59\n",
"[378 | 486.40] loss=0.10 avg=0.58\n",
"[379 | 487.41] loss=0.27 avg=0.58\n",
"[380 | 488.41] loss=0.14 avg=0.57\n",
"[381 | 489.43] loss=0.17 avg=0.57\n",
"[382 | 490.43] loss=0.23 avg=0.57\n",
"[383 | 491.44] loss=0.21 avg=0.56\n",
"[384 | 492.45] loss=0.28 avg=0.56\n",
"[385 | 493.46] loss=0.11 avg=0.56\n",
"[386 | 494.47] loss=0.30 avg=0.55\n",
"[387 | 495.47] loss=0.19 avg=0.55\n",
"[388 | 496.48] loss=0.31 avg=0.55\n",
"[389 | 497.49] loss=0.30 avg=0.54\n",
"[390 | 498.50] loss=0.27 avg=0.54\n",
"[391 | 499.50] loss=0.13 avg=0.54\n",
"[392 | 500.51] loss=0.17 avg=0.53\n",
"[393 | 501.52] loss=0.24 avg=0.53\n",
"[394 | 502.53] loss=0.23 avg=0.53\n",
"[395 | 503.54] loss=0.13 avg=0.52\n",
"[396 | 504.55] loss=0.10 avg=0.52\n",
"[397 | 505.56] loss=0.12 avg=0.52\n",
"[398 | 506.57] loss=0.21 avg=0.51\n",
"[399 | 507.58] loss=0.19 avg=0.51\n",
"[400 | 508.59] loss=0.11 avg=0.50\n",
"======== SAMPLE 1 ========\n",
" business are not at risk, and are not likely to be adversely affected. Many people would expect to lose their jobs because the company has increased productivity or reduced costs, when in fact no one has lost their job and the company has not shifted jobs. The company may have changed names or leadership, but the employees have not been moved. Instead, employees are better off as a result of the decision to invest in digital technology, and the process for assessing digital initiatives has changed.\n",
"Digital transformation is a continuous process, and it involves identifying which digital assets to create and in what quantities and through what channels, when and how they should be used. It involves making quality digital assets available and easily discoverable, via data products and digital content; reducing digital design timescales; and automating the process of creating and distributing digital products and assets. It also involves improving efficiency and reducing complexity within the organizational structure.\n",
"Digital transformation requirements explain how a digital transformation should work, and these requirements describe how employees should be engaged and structured to implement the transformation. They do not dictate how a digital transformation will be accomplished.\n",
"All companies experience a period of adjustment after a major change like a new technology platform or organizational structure. This period of adjustment should be short-lived, as it gives employees time to adjust to the new approach and communications style.\n",
"Delayed transition to new processes and communication style can be damaging. When employees know that a digital transformation is under way, it can be hard to understand what's involved and how much it will change things. As a result, many employees fear the change and may delay reporting problems to see how things will be different.\n",
"It's important that employees understand that as part of a digital transformation, they must also accept some forms of change (such as using new technology or embracing new ways of working). As technology and business need change, so must the ways that employees be engaged and structured to take advantage of the changes.\n",
"Employee Digital Transformation Frameworks\n",
"It's important that employees understand what types of change are expected of them, if they are to make the most of the opportunities provided by digital transformation. These changes can include:\n",
"Making changes to how they work and interact with their customers.\n",
"Working in partnership with business to develop new digital solutions that meet the needs of customers\n",
"Developing and implementing new communication materials that reinforce digital solutions\n",
"Discussing and implementing new processes to manage and improve working relationships with customers\n",
"Creating new roles and working practices to better utilize digital solutions\n",
"Using a diversity of digital media, such as digital media and apps, to create and tap customer feelings and insights\n",
"Creating space for ethical issues around sharing customer data and analytics\n",
"A variety of employee feedback and management interactive tools can help create a culture of innovation and embrace change\n",
"Using these tools, employees can get help finding and responding to alternative views and opinions. They can ask for help and feedback from senior colleagues and managers, and share their ideas and experiences on the employee feedback website or in digital products.\n",
"Although some companies have specific guidelines around digital transformation, most workplaces will at least consider some type of digital transformation, and some may even have a stage or cycle defined by the change. For example, company policy may specify that IT and customer experience managers be created as digital transformation trainees, and that these roles be filled based on performance.\n",
"How digital transformation may work at KPMG\n",
"Today, we at KPMG offer a number of solutions to help businesses and their clients transform their business models and grow. These solutions can be applied to any business area, and typically include different software solutions, including:\n",
"A comprehensive support network that includes technical support and consultancies if necessary\n",
"A talented and growing staff who are ready and available to assist with all aspects of the business\n",
"A profitable business model based on increasing productivity and customer service both within the office and via its online sales force and customer service teams\n",
"Technical and business support contracts based on number of employee satisfied ratings and overall rating of customer service\n",
"Not only can our solutions help companies and their clients achieve positive realizations with regards to digital transformation, but we have found that working together as a company in support of our fellow partners and clients who are already involved in this area. This is because we all share a common interest in improving the customer experience across the board.\n",
"We at KPMG see digital change as a key pillar of our business, and so has the staff at our Digital Experience Customer Experience team. Our team works throughout the year to continually learn and improve our products and services to deliver the best experience for our customers across the full digital transformation lifecycle.\n",
"Our open-source digital transformation strategy\n",
"In 2013, we at KPMG started to take a more strategic view to our open-source initiative. While it may have started as a purely philanthropic initiative, it has become an important part of how we develop and execute our digital transformation strategy.\n",
"Our open-source initiative began in 2004 when founder and CEO Andrew J. McCabe joined other senior executives from large technology companies like Oracle, Compaq,\n",
"\n",
"[401 | 535.77] loss=0.23 avg=0.50\n",
"[402 | 536.77] loss=0.16 avg=0.50\n",
"[403 | 537.77] loss=0.10 avg=0.49\n",
"[404 | 538.78] loss=0.33 avg=0.49\n",
"[405 | 539.78] loss=0.31 avg=0.49\n",
"[406 | 540.79] loss=0.09 avg=0.49\n",
"[407 | 541.79] loss=0.12 avg=0.48\n",
"[408 | 542.80] loss=0.29 avg=0.48\n",
"[409 | 543.81] loss=0.20 avg=0.48\n",
"[410 | 544.81] loss=0.23 avg=0.48\n",
"[411 | 545.82] loss=0.14 avg=0.47\n",
"[412 | 546.82] loss=0.12 avg=0.47\n",
"[413 | 547.83] loss=0.20 avg=0.47\n",
"[414 | 548.83] loss=0.19 avg=0.46\n",
"[415 | 549.84] loss=0.24 avg=0.46\n",
"[416 | 550.84] loss=0.12 avg=0.46\n",
"[417 | 551.85] loss=0.10 avg=0.45\n",
"[418 | 552.86] loss=0.10 avg=0.45\n",
"[419 | 553.86] loss=0.06 avg=0.45\n",
"[420 | 554.87] loss=0.14 avg=0.44\n",
"[421 | 555.88] loss=0.23 avg=0.44\n",
"[422 | 556.88] loss=0.15 avg=0.44\n",
"[423 | 557.89] loss=0.09 avg=0.43\n",
"[424 | 558.90] loss=0.73 avg=0.44\n",
"[425 | 559.90] loss=0.12 avg=0.43\n",
"[426 | 560.91] loss=0.16 avg=0.43\n",
"[427 | 561.92] loss=0.16 avg=0.43\n",
"[428 | 562.92] loss=0.05 avg=0.42\n",
"[429 | 563.93] loss=0.11 avg=0.42\n",
"[430 | 564.94] loss=0.13 avg=0.42\n",
"[431 | 565.94] loss=0.11 avg=0.42\n",
"[432 | 566.95] loss=0.12 avg=0.41\n",
"[433 | 567.96] loss=0.28 avg=0.41\n",
"[434 | 568.97] loss=0.15 avg=0.41\n",
"[435 | 569.97] loss=0.13 avg=0.41\n",
"[436 | 570.98] loss=0.42 avg=0.41\n",
"[437 | 571.99] loss=0.12 avg=0.40\n",
"[438 | 573.00] loss=0.08 avg=0.40\n",
"[439 | 574.00] loss=0.18 avg=0.40\n",
"[440 | 575.01] loss=0.16 avg=0.40\n",
"[441 | 576.02] loss=0.16 avg=0.39\n",
"[442 | 577.03] loss=0.18 avg=0.39\n",
"[443 | 578.04] loss=0.20 avg=0.39\n",
"[444 | 579.04] loss=0.06 avg=0.39\n",
"[445 | 580.05] loss=0.15 avg=0.38\n",
"[446 | 581.06] loss=0.06 avg=0.38\n",
"[447 | 582.07] loss=0.30 avg=0.38\n",
"[448 | 583.07] loss=0.12 avg=0.38\n",
"[449 | 584.08] loss=0.05 avg=0.37\n",
"[450 | 585.09] loss=0.13 avg=0.37\n",
"[451 | 586.10] loss=0.14 avg=0.37\n",
"[452 | 587.10] loss=0.10 avg=0.37\n",
"[453 | 588.11] loss=0.17 avg=0.36\n",
"[454 | 589.12] loss=0.06 avg=0.36\n",
"[455 | 590.13] loss=0.17 avg=0.36\n",
"[456 | 591.14] loss=0.08 avg=0.36\n",
"[457 | 592.14] loss=0.15 avg=0.35\n",
"[458 | 593.15] loss=0.17 avg=0.35\n",
"[459 | 594.16] loss=0.04 avg=0.35\n",
"[460 | 595.17] loss=0.09 avg=0.35\n",
"[461 | 596.18] loss=0.06 avg=0.34\n",
"[462 | 597.18] loss=0.08 avg=0.34\n",
"[463 | 598.19] loss=0.09 avg=0.34\n",
"[464 | 599.20] loss=0.12 avg=0.34\n",
"[465 | 600.21] loss=0.14 avg=0.33\n",
"[466 | 601.22] loss=0.10 avg=0.33\n",
"[467 | 602.23] loss=0.06 avg=0.33\n",
"[468 | 603.24] loss=0.22 avg=0.33\n",
"[469 | 604.24] loss=0.17 avg=0.33\n",
"[470 | 605.25] loss=0.11 avg=0.32\n",
"[471 | 606.26] loss=0.11 avg=0.32\n",
"[472 | 607.27] loss=0.08 avg=0.32\n",
"[473 | 608.28] loss=0.08 avg=0.32\n",
"[474 | 609.29] loss=0.08 avg=0.31\n",
"[475 | 610.30] loss=0.53 avg=0.32\n",
"[476 | 611.31] loss=0.15 avg=0.31\n",
"[477 | 612.32] loss=0.14 avg=0.31\n",
"[478 | 613.33] loss=0.07 avg=0.31\n",
"[479 | 614.34] loss=0.06 avg=0.31\n",
"[480 | 615.34] loss=0.11 avg=0.31\n",
"[481 | 616.35] loss=0.06 avg=0.30\n",
"[482 | 617.36] loss=0.10 avg=0.30\n",
"[483 | 618.37] loss=0.08 avg=0.30\n",
"[484 | 619.38] loss=0.14 avg=0.30\n",
"[485 | 620.39] loss=0.14 avg=0.30\n",
"[486 | 621.40] loss=0.07 avg=0.29\n",
"[487 | 622.41] loss=0.10 avg=0.29\n",
"[488 | 623.42] loss=0.12 avg=0.29\n",
"[489 | 624.43] loss=0.10 avg=0.29\n",
"[490 | 625.43] loss=0.08 avg=0.29\n",
"[491 | 626.44] loss=0.16 avg=0.28\n",
"[492 | 627.45] loss=0.19 avg=0.28\n",
"[493 | 628.46] loss=0.21 avg=0.28\n",
"[494 | 629.47] loss=0.19 avg=0.28\n",
"[495 | 630.48] loss=0.10 avg=0.28\n",
"[496 | 631.49] loss=0.07 avg=0.28\n",
"[497 | 632.50] loss=0.23 avg=0.28\n",
"[498 | 633.51] loss=0.16 avg=0.28\n",
"[499 | 634.52] loss=0.04 avg=0.27\n",
"[500 | 635.53] loss=0.10 avg=0.27\n",
"======== SAMPLE 1 ========\n",
" the data provided, and I'd like you to explore what might be able to be learned from this data. The best practitioners are all about identifying lessons learned, and which organizations should we target with future campaigns? What issues and problems do we know a lot of our customers are trying to solve?\n",
"In a large organization, one person (or a small team of people) can provide this level of expertise internally or publicly. At a Startup?almost everyone is an Expert now. What's next, Global Warming Alarm or Artificial General Performance Equivalences (AGPs)?\n",
"So, what's next, Targeted Marketing?\n",
"So, in the next 6-12 months we?ve done a really good job at educating customers, partners and co-workers about the value of Value Chain awareness and engagement tools. Now it's time to get people to use and build these things!\n",
"I'm personally excited about the prospect of small groups of people working together to make a difference in their communities. We?ve finally made it to the point in the marketing funnel where online marketing plays a more integral role in achieving social change than at any other time in the past. There's no better place to start than with yourself.\n",
"Sincere regards, Tim\n",
"About a month ago I received a curious message from a reader who wanted to know what I thought about the new Google Analytics ?Docs? that Google is sending out to all their Digital Advertising Account Managers (DAMs). The important thing to note here is that these documents are primarily a communications tool, meant to clarify what is needed internally for discussions and concerns to be addressed to the company as a whole. Many companies replace these emails out of laziness, but many more maintain them as a reminder of what is expected of your company across all of its digital channels.\n",
"In recent times companies have been looking to increase transparency and improve business relations through the use of internal communications. New technology has made such meetings even more effective, as they provide insights into how people think and feel about the business, and exactly what sort of messages are getting through to make others think and feel differently.\n",
"In the illustration below you can see a representation of three such systems I've seen used by companies: the chief customer officer (CCO) concept drawing, a Google Docs document and an Uber and Other .io and Flipkart Document.\n",
"The above structures meetings into three categories: objectives, concerns and questions. The objectives are the overarching ambitions for the day as well as questions to be answered during the meeting. The concerns are the specific goals or targets at which the organisation is struggling compared to others in its area of responsibility. and the questions are the curious requests or instructions that start a chain of events that get answered at the end of the meeting.\n",
"The Uber and Other .io and Flipkart Document is very similar to an Uber ad, the key difference is that this document is created and shared, along with other executives' feedback and instructions, the same way an Uber ad could look. It's important to note, that this is not a commissioned review, but rather feedback from across the company with no intention of anything other than sharing.\n",
"In the next post I'll take a look at two other examples of great use of internal communications I've seen in the past: Google's anonymous review process and Apple's new lobby guide.\n",
"While we're on the subject of Google and their privacy issues, it's worth emphasizing that even though internal communications won't directly ask for customer data, they should still fall within the remit of a company like Google. As we've seen with other companies that have worked with us, it's not the nature of Google to bypass its users' expectations and protections.\n",
"While we're on the subject of Google, it's also important to note that most companies hosting their customer data on the Google network. This is in contrast to traditional cloud and web sites, where data is traditionally held onsite. The problem with this approach is that there is no way for users or audit agencies to know about the data needs being serviced if the data is not publicly available. In addition, because this data is hosted on the Google network, there is no protection for confidential information based on proprietary technology.\n",
"As you can see from the picture above, the Verizon network ad unit is hosting its customer's data on behalf of the company. As you can see from the image below, there is no indication of where this data is coming from, why it is needed, or how it will be used. These are all key questions that agencies need to understand about the technical details of running their business, and how the company is positioning itself in order to plan ahead.\n",
"While we're on the subject of Verizon, I thought I'd highlight another recent case study that showed how switching to a new vendor can mean serious money and business changes for your business. We are now in the process of tearing down our call center in order to make way for a new digital marketing infrastructure. As you can see from the high-\n",
"\n",
"[501 | 662.31] loss=0.10 avg=0.27\n",
"[502 | 663.32] loss=0.14 avg=0.27\n",
"[503 | 664.33] loss=0.10 avg=0.27\n",
"[504 | 665.33] loss=0.09 avg=0.27\n",
"[505 | 666.34] loss=0.16 avg=0.26\n",
"[506 | 667.34] loss=0.16 avg=0.26\n",
"[507 | 668.35] loss=0.04 avg=0.26\n",
"[508 | 669.35] loss=0.10 avg=0.26\n",
"[509 | 670.36] loss=0.08 avg=0.26\n",
"[510 | 671.36] loss=0.12 avg=0.26\n",
"[511 | 672.37] loss=0.10 avg=0.26\n",
"[512 | 673.37] loss=0.14 avg=0.25\n",
"[513 | 674.38] loss=0.16 avg=0.25\n",
"[514 | 675.39] loss=0.19 avg=0.25\n",
"[515 | 676.39] loss=0.05 avg=0.25\n",
"[516 | 677.40] loss=0.11 avg=0.25\n",
"[517 | 678.41] loss=0.12 avg=0.25\n",
"[518 | 679.42] loss=0.09 avg=0.25\n",
"[519 | 680.42] loss=0.10 avg=0.24\n",
"[520 | 681.43] loss=0.12 avg=0.24\n",
"[521 | 682.44] loss=0.13 avg=0.24\n",
"[522 | 683.44] loss=0.12 avg=0.24\n",
"[523 | 684.45] loss=0.11 avg=0.24\n",
"[524 | 685.46] loss=0.17 avg=0.24\n",
"[525 | 686.46] loss=0.11 avg=0.24\n",
"[526 | 687.47] loss=0.09 avg=0.24\n",
"[527 | 688.48] loss=0.07 avg=0.23\n",
"[528 | 689.49] loss=0.11 avg=0.23\n",
"[529 | 690.49] loss=0.05 avg=0.23\n",
"[530 | 691.50] loss=0.06 avg=0.23\n",
"[531 | 692.51] loss=0.19 avg=0.23\n",
"[532 | 693.52] loss=0.10 avg=0.23\n",
"[533 | 694.52] loss=0.07 avg=0.23\n",
"[534 | 695.53] loss=0.07 avg=0.22\n",
"[535 | 696.54] loss=0.09 avg=0.22\n",
"[536 | 697.55] loss=0.17 avg=0.22\n",
"[537 | 698.55] loss=0.17 avg=0.22\n",
"[538 | 699.56] loss=0.13 avg=0.22\n",
"[539 | 700.57] loss=0.07 avg=0.22\n",
"[540 | 701.58] loss=0.15 avg=0.22\n",
"[541 | 702.58] loss=0.11 avg=0.22\n",
"[542 | 703.59] loss=0.09 avg=0.22\n",
"[543 | 704.60] loss=0.07 avg=0.22\n",
"[544 | 705.61] loss=0.11 avg=0.21\n",
"[545 | 706.62] loss=0.15 avg=0.21\n",
"[546 | 707.63] loss=0.13 avg=0.21\n",
"[547 | 708.63] loss=0.22 avg=0.21\n",
"[548 | 709.64] loss=0.14 avg=0.21\n",
"[549 | 710.65] loss=0.21 avg=0.21\n",
"[550 | 711.66] loss=0.10 avg=0.21\n",
"[551 | 712.67] loss=0.13 avg=0.21\n",
"[552 | 713.68] loss=0.10 avg=0.21\n",
"[553 | 714.69] loss=0.14 avg=0.21\n",
"[554 | 715.70] loss=0.07 avg=0.21\n",
"[555 | 716.71] loss=0.07 avg=0.21\n",
"[556 | 717.72] loss=0.03 avg=0.20\n",
"[557 | 718.73] loss=0.06 avg=0.20\n",
"[558 | 719.74] loss=0.18 avg=0.20\n",
"[559 | 720.74] loss=0.17 avg=0.20\n",
"[560 | 721.75] loss=0.08 avg=0.20\n",
"[561 | 722.76] loss=0.12 avg=0.20\n",
"[562 | 723.77] loss=0.11 avg=0.20\n",
"[563 | 724.78] loss=0.12 avg=0.20\n",
"[564 | 725.80] loss=0.09 avg=0.20\n",
"[565 | 726.80] loss=0.09 avg=0.20\n",
"[566 | 727.81] loss=0.19 avg=0.20\n",
"[567 | 728.83] loss=0.12 avg=0.20\n",
"[568 | 729.84] loss=0.13 avg=0.19\n",
"[569 | 730.85] loss=0.12 avg=0.19\n",
"[570 | 731.86] loss=0.05 avg=0.19\n",
"[571 | 732.87] loss=0.12 avg=0.19\n",
"[572 | 733.88] loss=0.11 avg=0.19\n",
"[573 | 734.88] loss=0.06 avg=0.19\n",
"[574 | 735.89] loss=0.12 avg=0.19\n",
"[575 | 736.90] loss=0.08 avg=0.19\n",
"[576 | 737.91] loss=0.16 avg=0.19\n",
"[577 | 738.92] loss=0.12 avg=0.19\n",
"[578 | 739.92] loss=0.06 avg=0.19\n",
"[579 | 740.94] loss=0.10 avg=0.18\n",
"[580 | 741.95] loss=0.09 avg=0.18\n",
"[581 | 742.95] loss=0.11 avg=0.18\n",
"[582 | 743.96] loss=0.20 avg=0.18\n",
"[583 | 744.97] loss=0.13 avg=0.18\n",
"[584 | 745.98] loss=0.17 avg=0.18\n",
"[585 | 746.99] loss=0.12 avg=0.18\n",
"[586 | 748.00] loss=0.08 avg=0.18\n",
"[587 | 749.01] loss=0.07 avg=0.18\n",
"[588 | 750.01] loss=0.10 avg=0.18\n",
"[589 | 751.03] loss=0.13 avg=0.18\n",
"[590 | 752.04] loss=0.12 avg=0.18\n",
"[591 | 753.05] loss=0.09 avg=0.18\n",
"[592 | 754.06] loss=0.09 avg=0.18\n",
"[593 | 755.07] loss=0.18 avg=0.18\n",
"[594 | 756.08] loss=0.15 avg=0.18\n",
"[595 | 757.09] loss=0.10 avg=0.18\n",
"[596 | 758.10] loss=0.09 avg=0.17\n",
"[597 | 759.11] loss=0.06 avg=0.17\n",
"[598 | 760.12] loss=0.10 avg=0.17\n",
"[599 | 761.13] loss=0.12 avg=0.17\n",
"[600 | 762.14] loss=0.10 avg=0.17\n",
"======== SAMPLE 1 ========\n",
" company's technical data to its customers. We'll look at how companies are using machine learning to discover insights from massive datasets and use cases, and how this new data generation technology is revolutionizing data management.\n",
"\n",
"\n",
"We'll also take a look at how Nielsen is using big data, analytics and cloud to create a Shared Economies of Scope, providing value to both customers and its bookkeepers, accountants and salespeople via a single platform.\n",
"\n",
"\n",
"Finally, we'll take a look at how Plaxo is implementing machine learning in its marketing data management system and how this is enabling it to develop new monetization streams and segments based on user behavior patterns.\n",
"\n",
"\n",
"I'm sure you can imagine the response to my presentation at Conferences?Bright Futures 2013 where, after a great response, I delivered a 2 hour bonus presentation titled ?How To Get Sales Results From Any Study? on the subject of ?big data?. As you can imagine, that led to even greater interest in the topic so next year?s Summit?Brown College is coming up and I?m expecting to be a topic of conversation again. This time, though, I?m hoping to hear from a bit more general interest as to how data sets can be utilized to obtain notable insights.\n",
"Anyhoo, go here to check out my presentation materials: Data Sets and Analytics: Uncover Value, Prioritize Resources and Prevent Waste?and here for the slides.\n",
"When I told folks about my presentation, this is what they said:\n",
"?The value statement rings true for a lot of businesses, as it seems to me this scenario plays out every day for many: A sales team spends an inordinate amount of time conducting exploratory searches on Google and other online resources?before launching into a full-fledged transaction. Moreover, such searches tend to be unproductive, since they lack specificity and are prone to error, particularly when used repeatedly in a futile attempt to discover relevant products or services. In addition, many companies encounter similar frustrations with data governance, a problem that Plaxo is working on to a limited extent to help them control their data and get the best possible insight from it.\n",
"Ouch. Anybody home?(That's me at the gate). What'd they say?\n",
"Anyway, enough ranting - time to talk data. The talk should be interesting! I'll keep it to about five minutes, but first a word about me as this might be relevant to people listening in other contexts.\n",
"I'm the co-founder and CEO of Plaxo, a data-driven communication and evaluation platform. In the last three years, we have made considerable inroads into the 'data-driven' market. This is because we offer a platform that allows businesses to develop and deploy data-driven infrastructures, which we call Data Architects, who work directly with data, rather than building databases on top of it.\n",
"The Plaxo web and mobile apps enable anyone, everywhere to have an intelligent digital assistant that helps them manage their money better, find the best deals and stay organized. Our approach to data isn't unique - many other companies are tackling the same issues as we are - but ours is also distinct, as we embrace data not as a way to solve every problem, but as a core experience that every customer has every day, not just a handful of companies.\n",
"First, we have our Data Helper, which helps our customers manage the complexities of data when it comes to the floor. Our Data Helper helps teams understand the data itself, rather than its structured or unstructured aspects. A typical use case for the Data Helper is to understand how many columns are present in a table, and then to what entities the corresponding values are tied. This helps organizations manage the data but avoid the need to code it.\n",
"Next, we have the Data Security Group, which provides consulting and support to organizations on how to structure their data in a secure and trustworthy manner, which is uncontaminated by customer data. Our perspective on data security is unique - we believe that data can and should be safer than it is today's standard. Our approach to data security is to eliminate security weaknesses before they have a chance to occur. This is distinct from a security posture, where secure coding practices are in place, but holes in the security posture are not plugged. Our focus is on finding and stopping the flaws early, before they have a chance to have an effect on the business.\n",
"Third, we have the Data Modeling and Execution team, which delivers Next-Generation Data Management (NGDM) solutions to enable modern-day day operations. These solutions help organizations get the best out of their data by replacing the old collection of hard and software to construct data bases with one cloud-based data platform that contains all the data elements, trends, relationships, trends lines, etc. The goal of the platform is to turn any existing data set into a fully functional data platform that can be easily accessed, updated, filtered, and derived from\n",
"\n",
"[601 | 789.07] loss=0.06 avg=0.17\n",
"[602 | 790.07] loss=0.10 avg=0.17\n",
"[603 | 791.08] loss=0.15 avg=0.17\n",
"[604 | 792.08] loss=0.10 avg=0.17\n",
"[605 | 793.08] loss=0.15 avg=0.17\n",
"[606 | 794.09] loss=0.10 avg=0.17\n",
"[607 | 795.09] loss=0.09 avg=0.17\n",
"[608 | 796.10] loss=0.09 avg=0.17\n",
"[609 | 797.10] loss=0.08 avg=0.17\n",
"[610 | 798.10] loss=0.06 avg=0.16\n",
"[611 | 799.10] loss=0.08 avg=0.16\n",
"[612 | 800.10] loss=0.09 avg=0.16\n",
"[613 | 801.11] loss=0.10 avg=0.16\n",
"[614 | 802.12] loss=0.13 avg=0.16\n",
"[615 | 803.12] loss=0.10 avg=0.16\n",
"[616 | 804.13] loss=0.06 avg=0.16\n",
"[617 | 805.13] loss=0.07 avg=0.16\n",
"[618 | 806.14] loss=0.18 avg=0.16\n",
"[619 | 807.15] loss=0.06 avg=0.16\n",
"[620 | 808.15] loss=0.11 avg=0.16\n",
"[621 | 809.15] loss=0.08 avg=0.16\n",
"[622 | 810.16] loss=0.14 avg=0.16\n",
"[623 | 811.16] loss=0.13 avg=0.16\n",
"[624 | 812.17] loss=0.07 avg=0.16\n",
"[625 | 813.18] loss=0.05 avg=0.15\n",
"[626 | 814.18] loss=0.10 avg=0.15\n",
"[627 | 815.19] loss=0.12 avg=0.15\n",
"[628 | 816.19] loss=0.10 avg=0.15\n",
"[629 | 817.20] loss=0.09 avg=0.15\n",
"[630 | 818.21] loss=0.08 avg=0.15\n",
"[631 | 819.22] loss=0.06 avg=0.15\n",
"[632 | 820.23] loss=0.10 avg=0.15\n",
"[633 | 821.23] loss=0.19 avg=0.15\n",
"[634 | 822.24] loss=0.08 avg=0.15\n",
"[635 | 823.25] loss=0.08 avg=0.15\n",
"[636 | 824.26] loss=0.78 avg=0.16\n",
"[637 | 825.26] loss=0.06 avg=0.15\n",
"[638 | 826.27] loss=0.16 avg=0.15\n",
"[639 | 827.28] loss=0.08 avg=0.15\n",
"[640 | 828.29] loss=0.08 avg=0.15\n",
"[641 | 829.29] loss=0.12 avg=0.15\n",
"[642 | 830.30] loss=0.09 avg=0.15\n",
"[643 | 831.31] loss=0.11 avg=0.15\n",
"[644 | 832.31] loss=0.12 avg=0.15\n",
"[645 | 833.32] loss=0.13 avg=0.15\n",
"[646 | 834.33] loss=0.07 avg=0.15\n",
"[647 | 835.34] loss=0.06 avg=0.15\n",
"[648 | 836.35] loss=0.17 avg=0.15\n",
"[649 | 837.36] loss=0.11 avg=0.15\n",
"[650 | 838.37] loss=0.10 avg=0.15\n",
"[651 | 839.38] loss=0.09 avg=0.15\n",
"[652 | 840.39] loss=0.10 avg=0.15\n",
"[653 | 841.40] loss=0.10 avg=0.15\n",
"[654 | 842.40] loss=0.06 avg=0.15\n",
"[655 | 843.42] loss=0.05 avg=0.15\n",
"[656 | 844.42] loss=0.08 avg=0.14\n",
"[657 | 845.43] loss=0.10 avg=0.14\n",
"[658 | 846.44] loss=0.08 avg=0.14\n",
"[659 | 847.45] loss=0.11 avg=0.14\n",
"[660 | 848.46] loss=0.10 avg=0.14\n",
"[661 | 849.46] loss=0.07 avg=0.14\n",
"[662 | 850.47] loss=0.08 avg=0.14\n",
"[663 | 851.48] loss=0.09 avg=0.14\n",
"[664 | 852.49] loss=0.09 avg=0.14\n",
"[665 | 853.50] loss=0.09 avg=0.14\n",
"[666 | 854.51] loss=0.10 avg=0.14\n",
"[667 | 855.52] loss=0.07 avg=0.14\n",
"[668 | 856.52] loss=0.08 avg=0.14\n",
"[669 | 857.53] loss=0.10 avg=0.14\n",
"[670 | 858.54] loss=0.08 avg=0.14\n",
"[671 | 859.55] loss=0.06 avg=0.14\n",
"[672 | 860.56] loss=0.08 avg=0.14\n",
"[673 | 861.57] loss=0.06 avg=0.14\n",
"[674 | 862.58] loss=0.05 avg=0.13\n",
"[675 | 863.59] loss=0.15 avg=0.13\n",
"[676 | 864.60] loss=0.04 avg=0.13\n",
"[677 | 865.60] loss=0.08 avg=0.13\n",
"[678 | 866.61] loss=0.08 avg=0.13\n",
"[679 | 867.62] loss=0.06 avg=0.13\n",
"[680 | 868.63] loss=0.11 avg=0.13\n",
"[681 | 869.64] loss=0.05 avg=0.13\n",
"[682 | 870.65] loss=0.07 avg=0.13\n",
"[683 | 871.66] loss=0.05 avg=0.13\n",
"[684 | 872.67] loss=0.08 avg=0.13\n",
"[685 | 873.68] loss=0.06 avg=0.13\n",
"[686 | 874.69] loss=0.11 avg=0.13\n",
"[687 | 875.70] loss=0.06 avg=0.13\n",
"[688 | 876.71] loss=0.11 avg=0.13\n",
"[689 | 877.72] loss=0.07 avg=0.13\n",
"[690 | 878.73] loss=0.04 avg=0.13\n",
"[691 | 879.74] loss=0.10 avg=0.13\n",
"[692 | 880.75] loss=0.09 avg=0.13\n",
"[693 | 881.75] loss=0.07 avg=0.12\n",
"[694 | 882.76] loss=0.11 avg=0.12\n",
"[695 | 883.77] loss=0.11 avg=0.12\n",
"[696 | 884.78] loss=0.04 avg=0.12\n",
"[697 | 885.79] loss=0.10 avg=0.12\n",
"[698 | 886.80] loss=0.10 avg=0.12\n",
"[699 | 887.81] loss=0.04 avg=0.12\n",
"[700 | 888.82] loss=0.06 avg=0.12\n",
"======== SAMPLE 1 ========\n",
" normed on the importance of a company?s digital capability, then digital strategy should focus on:\n",
"Strengthening the company?s digital capabilities - We need to invest in more digital media and tools, such as digital web properties, online customer/provider interactions, and digital sales capabilities. We also need to create dynamic web pages and integrate Facebook advertising into our existing salesforce?s workflow.\n",
"Expanding the company Digital footprint - While developing our digital capabilities, we should also be considering expanding our digital footprint. If we want to continue to attract and hire talented digital marketers and technology people, we should be considering developing additional locations. While not every company will necessarily want to increase its global presence, by developing and implementing global digital capabilities, you not only attract and hire top-notch talent, but you also help to position your company to win new business and increase market share.\n",
"Setting ROI targets for digital capability development\n",
"To help prioritize how much digital capability development should be included into a digital strategy, it's useful to review these 6 parameters of a successful digital marketing plan: (See image below for definition)\n",
"Valuable feedback - If you?ve developed digital capabilities, which is generally the goal, how can you be sure your digital marketing communications are meeting customer's needs?\n",
"Valuable data - What is the value and value added potential of this data?\n",
"Valuable information - How can I be sure this valuable data can be found and used efficiently?\n",
"Value added - How can I be sure my digital marketing communications are contributing to and delivering the right value added by company to company?\n",
"Cost-effective - How can I be sure my digital marketing communications are being cost-effective, which may be through reduced conversion rates, enhanced customer experience, or new blood entering the company?\n",
"Marketing consultancy - How can I be sure my digital marketing communications are being professionally and consistently communicated to companywide audiences?\n",
"Effective communication - How can I be sure my digital marketing communications are being communicated effectively to all involved?\n",
"Effective involvement - How can I be sure my digital marketing communications are being effectively engaged with from a marketing communications professional?\n",
"Effective execution - How can I be sure my digital marketing communications are being effectively executed, which may be through improving customer acquisition rates, decreasing customer acquisition costs and/or increasing the customer satisfaction of customer?s-focussed marketing communications.\n",
"Sign up today for a free trial of our marketing research system today!<|endoftext|>How do you turn your dog into a human? This is the question a dog owner is likely to be wondering lately.\n",
"\n",
"Last week, we told you how a London dog breeder named Yvonne Rachal was banned from keeping dogs after it was discovered that her puppies she was breeding were on the menu at Michelin-starred restaurants. And that wasn't the worst-case scenario for dogs-rights activists. On Friday, the Guardian revealed that a Belgian labrador pup was given a five-star Michelin-starred experience after a petition protesting its treatment garnered enough support to overturn chef Laurent Blanc's decision not to serve it.\n",
"\n",
"The pup, named Chocolat, had been refused service at the Brasserie Côte d'Azur on its return leg from Paris after complaining about its owner's food, arguing the owner was wearing slippers that could not protect its delicate hooves.\n",
"\n",
"The Michelin-starred Brasserie Côte d'Azur in Paris, where Chocolat tested positive for dog excrement. Image: MGN\n",
"\n",
"Despite this, Chocolat is back in the restaurant, and its owners, the parents of a 1-year-old daughter, claim it was the food and not the poo that hurt Chocolat, who was diagnosed with scabies after visiting a vet. In a press release, the parents of Chocolat said:\n",
"\n",
"\"When she was a puppy, Chocolat found his food untouched in the fridge. Since she can chew and dig, it made her hungry and she ate it. This time, after a very healthy dinner at the family's home, she went in for a few treatments and a week later went on his antibiotics. This time she wasn't so excited: 'The problem is that, despite the antibiotics, the fleas came out and itchy. The vet didn't know what to do, so I used to bring her to the vet but they never attended and when they did, it was too late.'\"\n",
"\n",
"Chocolat was not poisoned, however, as a vet told the family. Instead, it was mire insect, also known as the baby powder bug, that caused Chocolat's problems.\n",
"\n",
"Mire insect infestation at the Brasserie Côte d'Azur, where Chocolat died. Image: MGN\n",
"\n",
"Yvonne Rachal\n",
"\n",
"This wasn't Chocolat's first trip to the vet. Before Chocolat came down with the flea, it was Y\n",
"\n",
"[701 | 915.70] loss=0.11 avg=0.12\n",
"[702 | 916.70] loss=0.08 avg=0.12\n",
"[703 | 917.71] loss=0.10 avg=0.12\n",
"[704 | 918.71] loss=0.05 avg=0.12\n",
"[705 | 919.72] loss=0.13 avg=0.12\n",
"[706 | 920.72] loss=0.09 avg=0.12\n",
"[707 | 921.72] loss=0.07 avg=0.12\n",
"[708 | 922.72] loss=0.10 avg=0.12\n",
"[709 | 923.73] loss=0.03 avg=0.12\n",
"[710 | 924.73] loss=0.09 avg=0.12\n",
"[711 | 925.73] loss=0.12 avg=0.12\n",
"[712 | 926.74] loss=0.08 avg=0.12\n",
"[713 | 927.74] loss=0.06 avg=0.12\n",
"[714 | 928.74] loss=0.06 avg=0.12\n",
"[715 | 929.75] loss=0.12 avg=0.12\n",
"[716 | 930.76] loss=0.05 avg=0.12\n",
"[717 | 931.76] loss=0.10 avg=0.12\n",
"[718 | 932.77] loss=0.08 avg=0.12\n",
"[719 | 933.77] loss=0.04 avg=0.11\n",
"[720 | 934.78] loss=0.17 avg=0.12\n",
"[721 | 935.78] loss=0.07 avg=0.11\n",
"[722 | 936.79] loss=0.10 avg=0.11\n",
"[723 | 937.79] loss=0.13 avg=0.11\n",
"[724 | 938.80] loss=0.06 avg=0.11\n",
"[725 | 939.81] loss=0.08 avg=0.11\n",
"[726 | 940.81] loss=0.06 avg=0.11\n",
"[727 | 941.81] loss=0.08 avg=0.11\n",
"[728 | 942.82] loss=0.07 avg=0.11\n",
"[729 | 943.82] loss=0.07 avg=0.11\n",
"[730 | 944.83] loss=0.08 avg=0.11\n",
"[731 | 945.84] loss=0.08 avg=0.11\n",
"[732 | 946.84] loss=0.06 avg=0.11\n",
"[733 | 947.85] loss=0.09 avg=0.11\n",
"[734 | 948.86] loss=0.12 avg=0.11\n",
"[735 | 949.86] loss=0.09 avg=0.11\n",
"[736 | 950.87] loss=0.03 avg=0.11\n",
"[737 | 951.88] loss=0.08 avg=0.11\n",
"[738 | 952.89] loss=0.07 avg=0.11\n",
"[739 | 953.89] loss=0.08 avg=0.11\n",
"[740 | 954.90] loss=0.09 avg=0.11\n",
"[741 | 955.91] loss=0.04 avg=0.11\n",
"[742 | 956.91] loss=0.06 avg=0.11\n",
"[743 | 957.92] loss=0.13 avg=0.11\n",
"[744 | 958.93] loss=0.07 avg=0.11\n",
"[745 | 959.93] loss=0.07 avg=0.11\n",
"[746 | 960.94] loss=0.07 avg=0.11\n",
"[747 | 961.95] loss=0.06 avg=0.11\n",
"[748 | 962.96] loss=0.09 avg=0.11\n",
"[749 | 963.96] loss=0.04 avg=0.11\n",
"[750 | 964.97] loss=0.05 avg=0.10\n",
"[751 | 965.98] loss=0.06 avg=0.10\n",
"[752 | 966.99] loss=0.09 avg=0.10\n",
"[753 | 968.00] loss=0.07 avg=0.10\n",
"[754 | 969.01] loss=0.06 avg=0.10\n",
"[755 | 970.02] loss=0.15 avg=0.10\n",
"[756 | 971.02] loss=0.11 avg=0.10\n",
"[757 | 972.03] loss=0.16 avg=0.10\n",
"[758 | 973.04] loss=0.07 avg=0.10\n",
"[759 | 974.04] loss=0.03 avg=0.10\n",
"[760 | 975.05] loss=0.06 avg=0.10\n",
"[761 | 976.06] loss=0.07 avg=0.10\n",
"[762 | 977.06] loss=0.04 avg=0.10\n",
"[763 | 978.08] loss=0.07 avg=0.10\n",
"[764 | 979.09] loss=0.10 avg=0.10\n",
"[765 | 980.09] loss=0.09 avg=0.10\n",
"[766 | 981.10] loss=0.09 avg=0.10\n",
"[767 | 982.11] loss=0.13 avg=0.10\n",
"[768 | 983.12] loss=0.10 avg=0.10\n",
"[769 | 984.13] loss=0.06 avg=0.10\n",
"[770 | 985.14] loss=0.13 avg=0.10\n",
"[771 | 986.15] loss=0.05 avg=0.10\n",
"[772 | 987.15] loss=0.07 avg=0.10\n",
"[773 | 988.16] loss=0.14 avg=0.10\n",
"[774 | 989.17] loss=0.14 avg=0.10\n",
"[775 | 990.18] loss=0.08 avg=0.10\n",
"[776 | 991.19] loss=0.04 avg=0.10\n",
"[777 | 992.20] loss=0.06 avg=0.10\n",
"[778 | 993.21] loss=0.07 avg=0.10\n",
"[779 | 994.22] loss=0.05 avg=0.10\n",
"[780 | 995.23] loss=0.07 avg=0.10\n",
"[781 | 996.24] loss=0.09 avg=0.10\n",
"[782 | 997.25] loss=0.08 avg=0.10\n",
"[783 | 998.26] loss=0.04 avg=0.10\n",
"[784 | 999.27] loss=0.10 avg=0.10\n",
"[785 | 1000.28] loss=0.04 avg=0.10\n",
"[786 | 1001.29] loss=0.08 avg=0.10\n",
"[787 | 1002.30] loss=0.11 avg=0.10\n",
"[788 | 1003.31] loss=0.05 avg=0.10\n",
"[789 | 1004.32] loss=0.09 avg=0.10\n",
"[790 | 1005.33] loss=0.08 avg=0.10\n",
"[791 | 1006.34] loss=0.07 avg=0.10\n",
"[792 | 1007.35] loss=0.06 avg=0.10\n",
"[793 | 1008.36] loss=0.10 avg=0.10\n",
"[794 | 1009.37] loss=0.06 avg=0.10\n",
"[795 | 1010.38] loss=0.06 avg=0.10\n",
"[796 | 1011.39] loss=0.07 avg=0.10\n",
"[797 | 1012.40] loss=0.10 avg=0.10\n",
"[798 | 1013.41] loss=0.03 avg=0.10\n",
"[799 | 1014.42] loss=0.11 avg=0.10\n",
"[800 | 1015.43] loss=0.06 avg=0.09\n",
"======== SAMPLE 1 ========\n",
" 68.\n",
"The other metric to be aware of is return on investment (ROI). Organizations that have a high level of productivity is productivity is rewarded with more hours of employees time. On the flip side, spending too much time on non-productive activities is wasteful of time and resources.\n",
"Leaders are using multiple metrics to measure their success, which is also known as \"digital pecking order.\" They use multiple technologies, such as email, mobile apps, web application or online marketing communications, to market to customers.\n",
"If we have experience with digital marketing techniques, we can use them as a strategy to increase customer engagement, conversion and/or sell more products or services to the same or additional audience. This is called a network leadership strategy.\n",
"How to implement a digital network and leadership strategy\n",
"In my 14 years in business, I can say that this is a common question that entrepreneurs and new companies encounter. Most have one of 3 outcomes:\n",
"Recurring budget to conduct digital marketing Every week they perform at least 1 digital marketing campaign and at most 6-12\n",
"Identify a Technology Partner and Implementing digital marketing strategy Helps get customers to buy from them\n",
"Have new software releases, free e-books, webinars and promotional items available to customers\n",
"Marketing automation to increase customer engagement\n",
"When will you see the benefits?\n",
"You will likely see the results of digital marketing within your operational life-cycle:\n",
"Start with a digital marketing plan\n",
"An experienced sales professional or business owner can write a productive marketing plan in seven to 10 good reasons to write one (though a six-week development cycle seems a little on the long side). When you have a good one, go back to the drawing board and rewrite it with more efficiency.\n",
"Now comes the part that most new companies struggle with: implementing. Even experienced people struggle with this. When I was CEO of Online Auctions, we had 100% return on investment (ROI) from our digital marketing plan except for the last point, where we spent $100 on marketing and $20 on data processing.\n",
"Our approach was to add new value to our business' current digital marketing practices. Our customers often had big data and digital transformation challenges and wanted help with the understanding of the new products and solutions. We also had a very small team focused on digital transformation and didn't have many digital marketing experience, so we were trying to create a big image for our company by doing a digital transformation.\n",
"We added a new feature to our website once or twice a week and had 10-15% of our traffic from new sources so we had to find new ways to earn revenues. Our goal was to create a digital pie and have professional websites for our team members. Our competitor moved very quickly from a legacy business site to a digital enterprise one but added nothing to its customer support services. We took 12 months to build our new website and 18 months to build our new mobile app.\n",
"We took our time and ended up with one of the best digital customer interfaces, integrations and advertising solutions for our business. We are in the process of modernizing our website and we can now recommend other businesses where they can get online sales jobs.\n",
"Our competitor now has an automated process where they can quickly create a vision statement and vision for the future of their business and then instantly know if they are on the right track. This tool gives them the power to change the direction of their business and create a digital transformation initiative in just 1-2 minutes.\n",
"Manage your digital goals and digital transformation activities in one place\n",
"Today, companies are managing many of these changes internally and outside of the company. Using a digital goal-setting tool like Duchateau lets you review and change at a high level what is needed from your employees, customers, partners and managers. In a small to medium sized business, this process rate can increase with each step.\n",
"As we have more people use Duchateau to manage digital goals and digital transformation activities, the rate of change reduces. This is because a small business needs time and flexibility to plan and conduct digital goals and digital transformation activities.\n",
"Ultimately, it is about making sure that staff and customers have the tools they need to make efficient use of their time and resources.\n",
"Difference between a digital goal and a digital transformation initiative?\n",
"A digital goal is a clear and realistic plan of action for how this business is going to use digital technology, e.g. online sales, online marketing or digital PR.\n",
"A digital transformation initiative is a shared strategy to get the whole company on board with using digital technology, i.e. getting everyone involved in the company using social media and online advertising.<|endoftext|>A BILL TO be presented to the Commons on Tuesday will require voters to declare their party affiliation on postal ballots.\n",
"\n",
"The Electoral Reform (Postal Voting) Bill 2014 would require all voters to declare their party affiliation on their postal ballots. But unlike in Scotland, where a \"Yes but don't tell me\" vote doesn't apply, in England and\n",
"\n",
"[801 | 1042.46] loss=0.03 avg=0.09\n",
"[802 | 1043.46] loss=0.07 avg=0.09\n",
"[803 | 1044.47] loss=0.08 avg=0.09\n",
"[804 | 1045.47] loss=0.09 avg=0.09\n",
"[805 | 1046.48] loss=0.07 avg=0.09\n",
"[806 | 1047.48] loss=0.09 avg=0.09\n",
"[807 | 1048.48] loss=0.06 avg=0.09\n",
"[808 | 1049.49] loss=0.13 avg=0.09\n",
"[809 | 1050.49] loss=0.09 avg=0.09\n",
"[810 | 1051.49] loss=0.07 avg=0.09\n",
"[811 | 1052.50] loss=0.06 avg=0.09\n",
"[812 | 1053.50] loss=0.09 avg=0.09\n",
"[813 | 1054.50] loss=0.08 avg=0.09\n",
"[814 | 1055.51] loss=0.12 avg=0.09\n",
"[815 | 1056.52] loss=0.05 avg=0.09\n",
"[816 | 1057.52] loss=0.09 avg=0.09\n",
"[817 | 1058.53] loss=0.08 avg=0.09\n",
"[818 | 1059.54] loss=0.12 avg=0.09\n",
"[819 | 1060.54] loss=0.07 avg=0.09\n",
"[820 | 1061.54] loss=0.09 avg=0.09\n",
"[821 | 1062.55] loss=0.06 avg=0.09\n",
"[822 | 1063.55] loss=0.06 avg=0.09\n",
"[823 | 1064.56] loss=0.09 avg=0.09\n",
"[824 | 1065.56] loss=0.05 avg=0.09\n",
"[825 | 1066.57] loss=0.06 avg=0.09\n",
"[826 | 1067.57] loss=0.09 avg=0.09\n",
"[827 | 1068.57] loss=0.07 avg=0.09\n",
"[828 | 1069.58] loss=0.10 avg=0.09\n",
"[829 | 1070.58] loss=0.05 avg=0.09\n",
"[830 | 1071.59] loss=0.04 avg=0.09\n",
"[831 | 1072.59] loss=0.13 avg=0.09\n",
"[832 | 1073.60] loss=0.10 avg=0.09\n",
"[833 | 1074.60] loss=0.05 avg=0.09\n",
"[834 | 1075.61] loss=0.04 avg=0.09\n",
"[835 | 1076.61] loss=0.05 avg=0.09\n",
"[836 | 1077.62] loss=0.06 avg=0.09\n",
"[837 | 1078.62] loss=0.08 avg=0.09\n",
"[838 | 1079.63] loss=0.13 avg=0.09\n",
"[839 | 1080.63] loss=0.09 avg=0.09\n",
"[840 | 1081.64] loss=0.10 avg=0.09\n",
"[841 | 1082.64] loss=0.06 avg=0.09\n",
"[842 | 1083.65] loss=0.06 avg=0.09\n",
"[843 | 1084.65] loss=0.07 avg=0.09\n",
"[844 | 1085.66] loss=0.07 avg=0.09\n",
"[845 | 1086.66] loss=0.07 avg=0.09\n",
"[846 | 1087.66] loss=0.05 avg=0.09\n",
"[847 | 1088.67] loss=0.05 avg=0.09\n",
"[848 | 1089.67] loss=0.06 avg=0.09\n",
"[849 | 1090.68] loss=0.12 avg=0.09\n",
"[850 | 1091.69] loss=0.10 avg=0.09\n",
"[851 | 1092.69] loss=0.07 avg=0.09\n",
"[852 | 1093.70] loss=0.06 avg=0.09\n",
"[853 | 1094.71] loss=0.06 avg=0.09\n",
"[854 | 1095.71] loss=0.07 avg=0.09\n",
"[855 | 1096.71] loss=0.11 avg=0.09\n",
"[856 | 1097.72] loss=0.04 avg=0.09\n",
"[857 | 1098.73] loss=0.06 avg=0.09\n",
"[858 | 1099.73] loss=0.07 avg=0.09\n",
"[859 | 1100.74] loss=0.10 avg=0.09\n",
"[860 | 1101.74] loss=0.09 avg=0.09\n",
"[861 | 1102.75] loss=0.09 avg=0.09\n",
"[862 | 1103.76] loss=0.10 avg=0.09\n",
"[863 | 1104.76] loss=0.08 avg=0.09\n",
"[864 | 1105.77] loss=0.09 avg=0.09\n",
"[865 | 1106.78] loss=0.05 avg=0.09\n",
"[866 | 1107.78] loss=0.06 avg=0.09\n",
"[867 | 1108.79] loss=0.08 avg=0.09\n",
"[868 | 1109.80] loss=0.05 avg=0.09\n",
"[869 | 1110.80] loss=0.10 avg=0.09\n",
"[870 | 1111.81] loss=0.09 avg=0.09\n",
"[871 | 1112.82] loss=0.17 avg=0.09\n",
"[872 | 1113.83] loss=0.05 avg=0.09\n",
"[873 | 1114.83] loss=0.08 avg=0.09\n",
"[874 | 1115.84] loss=0.06 avg=0.09\n",
"[875 | 1116.85] loss=0.07 avg=0.09\n",
"[876 | 1117.86] loss=0.04 avg=0.09\n",
"[877 | 1118.87] loss=0.04 avg=0.08\n",
"[878 | 1119.88] loss=0.08 avg=0.08\n",
"[879 | 1120.88] loss=0.04 avg=0.08\n",
"[880 | 1121.89] loss=0.09 avg=0.08\n",
"[881 | 1122.89] loss=0.09 avg=0.08\n",
"[882 | 1123.91] loss=0.18 avg=0.09\n",
"[883 | 1124.91] loss=0.10 avg=0.09\n",
"[884 | 1125.92] loss=0.11 avg=0.09\n",
"[885 | 1126.93] loss=0.17 avg=0.09\n",
"[886 | 1127.94] loss=0.06 avg=0.09\n",
"[887 | 1128.95] loss=0.10 avg=0.09\n",
"[888 | 1129.96] loss=0.06 avg=0.09\n",
"[889 | 1130.97] loss=0.09 avg=0.09\n",
"[890 | 1131.98] loss=0.11 avg=0.09\n",
"[891 | 1132.98] loss=0.04 avg=0.09\n",
"[892 | 1133.99] loss=0.05 avg=0.09\n",
"[893 | 1135.00] loss=0.17 avg=0.09\n",
"[894 | 1136.01] loss=0.06 avg=0.09\n",
"[895 | 1137.02] loss=0.09 avg=0.09\n",
"[896 | 1138.03] loss=0.07 avg=0.09\n",
"[897 | 1139.04] loss=0.07 avg=0.09\n",
"[898 | 1140.05] loss=0.10 avg=0.09\n",
"[899 | 1141.06] loss=0.04 avg=0.09\n",
"[900 | 1142.06] loss=0.09 avg=0.09\n",
"======== SAMPLE 1 ========\n",
" already are.\n",
"So, what does it take to build one?\n",
"A clear strategy - Establish a strategy and stick to it. If you don't, inconsistencies and changes can arise. If you - How do you sell the benefits of the ITU-formed agreement? For example, by publishing register of operators data, downloading ?how to run iConnectNIC? friendly forms, etc. ?And? don't forget to include your data sponsors? payment gateway or analytics vendors? Depending on your target audience, these details will affect their online customer experience vastly. Also, communicate to potential users? experienced with previous forms. If you aren't familiar with current forms, explain why iConnectNIC isn't compatible and its limitations, for example, iConnectNIC doesn't support custom text areas, forms or AJAX calls. e.g. How will you recreate the functionality? iConnectNIC isn't backwards compatible - This is also important: if you release a new version of iConnectNIC that breaks something that you?ve built, expect a mass abandonment of the form. iConnectNIC isn't about the form - It's about data submission and management quicker and easier than current solutions allow. It also enables customers to manage their data through their data stores in one place, and can accommodate multiple providers of data quickly. It also offers rich insights into their data lives, and allows businesses to ?integrate? their data from different systems and formats, and share data discovery and decisions with the team. Third, it provides a simple interface to the data providers, so customers can more easily discover the data interests of their customers and interact with them. When you?re not writing forms, think about the technology you will use to manage your data. We have seen many cases where people want to manage their data through their data stores, using proprietary applications or systems. Implementing these systems from the ground up is beyond the capability of most customers, so after reviewing our data governance framework, we started with a review of the data management systems we had today. These systems were acceptable, but required improvement, and should have been redesigned to solve a business need. From there, we moved to look at business use cases, and how Data Governance can help with digital transformation. From there, architecture and technology decisions could be made to enable future data generation and management. From there, we were left with just a definition for what a Data Master Data Model is and what data it contains.\n",
"This led us to…\n",
"Standard Operating Procedure for Data Model Development\n",
"First, let's review what a Data Model is. A Data Model is a simplified representation of data in human-readable form. Humans prefer models with known parameters and features, so models with well-defined parameters and features are easier to work with and explore. We use models every day to diagnose problems in our data production and supplement our intuition with data.\n",
"A useful tool for describing a Data Model is Morpheus. Morpheus is a Python tool to help create Data Meets Business Process (DB3) and ITIL-compliant business data profiles. You can run a DBN to create a Data Profile from a set of Data Features and Existing DB applications. The DBN will export the characteristics of the data needed to create the profile and the associated application. The DBN will extract the key attributes of the data needed for the application and the associated data structure, namely, an Data Set or Data Model. The associated application will provide the definition of the data component and the logic to produce the data.\n",
"To make things easier, we create our best wishes for a life of LIFX!\n",
"Let's look at an example. Let's say we have an OEM customer who has decided to run special promotions every month. These discounts are usually limited to one device per customer and are based on purchase quantity, device type, region or even date purchased. These discounts can be very lucrative so we now have a larger piece of the business to work with.\n",
"We can now envision a new Data Model for this customer. They now have multiple sales representatives visiting multiple locations per month who can each price match the previous month's promotion. Each sales rep receives a price query template and can update the price based on their own judgement and data. The customer approval process is now based on the quantity and quality of devices the customer owns, not just a price point.\n",
"We can now see an easy way to develop a data transformation plan. We have:\n",
"Customer : An ideal customer?s basic purchasing behavior based on their digital footprints\n",
"Supplier : A ideal set of data components and processes for building the data component of the customer relationship\n",
"Plan : A detailed description of how the plan runs\n",
"Now that we have these categories, we can start to build the data component of the plan. The first step is to define the product/process group. We can think of this group of people as a software team. Each member of the group works on a different part of the solution. We have Microsoft Azure Data Lake\n",
"\n",
"[901 | 1168.97] loss=0.03 avg=0.08\n",
"[902 | 1169.98] loss=0.08 avg=0.08\n",
"[903 | 1170.98] loss=0.04 avg=0.08\n",
"[904 | 1171.98] loss=0.06 avg=0.08\n",
"[905 | 1172.99] loss=0.05 avg=0.08\n",
"[906 | 1173.99] loss=0.06 avg=0.08\n",
"[907 | 1175.00] loss=0.10 avg=0.08\n",
"[908 | 1176.01] loss=0.14 avg=0.08\n",
"[909 | 1177.01] loss=0.07 avg=0.08\n",
"[910 | 1178.01] loss=0.09 avg=0.08\n",
"[911 | 1179.02] loss=0.10 avg=0.08\n",
"[912 | 1180.03] loss=0.08 avg=0.08\n",
"[913 | 1181.04] loss=0.10 avg=0.08\n",
"[914 | 1182.04] loss=0.05 avg=0.08\n",
"[915 | 1183.05] loss=0.04 avg=0.08\n",
"[916 | 1184.05] loss=0.04 avg=0.08\n",
"[917 | 1185.06] loss=0.09 avg=0.08\n",
"[918 | 1186.06] loss=0.07 avg=0.08\n",
"[919 | 1187.07] loss=0.04 avg=0.08\n",
"[920 | 1188.07] loss=0.03 avg=0.08\n",
"[921 | 1189.08] loss=0.06 avg=0.08\n",
"[922 | 1190.09] loss=0.04 avg=0.08\n",
"[923 | 1191.09] loss=0.09 avg=0.08\n",
"[924 | 1192.10] loss=0.04 avg=0.08\n",
"[925 | 1193.10] loss=0.15 avg=0.08\n",
"[926 | 1194.11] loss=0.04 avg=0.08\n",
"[927 | 1195.12] loss=0.05 avg=0.08\n",
"[928 | 1196.12] loss=0.03 avg=0.08\n",
"[929 | 1197.13] loss=0.07 avg=0.08\n",
"[930 | 1198.13] loss=0.09 avg=0.08\n",
"[931 | 1199.14] loss=0.06 avg=0.08\n",
"[932 | 1200.15] loss=0.06 avg=0.08\n",
"[933 | 1201.15] loss=0.10 avg=0.08\n",
"[934 | 1202.16] loss=0.04 avg=0.08\n",
"[935 | 1203.17] loss=0.05 avg=0.08\n",
"[936 | 1204.17] loss=0.13 avg=0.08\n",
"[937 | 1205.18] loss=0.03 avg=0.08\n",
"[938 | 1206.19] loss=0.10 avg=0.08\n",
"[939 | 1207.19] loss=0.07 avg=0.08\n",
"[940 | 1208.20] loss=0.04 avg=0.08\n",
"[941 | 1209.20] loss=0.05 avg=0.08\n",
"[942 | 1210.21] loss=0.07 avg=0.08\n",
"[943 | 1211.22] loss=0.10 avg=0.08\n",
"[944 | 1212.22] loss=0.05 avg=0.08\n",
"[945 | 1213.23] loss=0.06 avg=0.08\n",
"[946 | 1214.24] loss=0.09 avg=0.08\n",
"[947 | 1215.25] loss=0.04 avg=0.08\n",
"[948 | 1216.25] loss=0.09 avg=0.08\n",
"[949 | 1217.26] loss=0.10 avg=0.08\n",
"[950 | 1218.26] loss=0.05 avg=0.08\n",
"[951 | 1219.27] loss=0.05 avg=0.08\n",
"[952 | 1220.28] loss=0.04 avg=0.08\n",
"[953 | 1221.28] loss=0.10 avg=0.08\n",
"[954 | 1222.29] loss=0.12 avg=0.08\n",
"[955 | 1223.30] loss=0.03 avg=0.08\n",
"[956 | 1224.31] loss=0.06 avg=0.08\n",
"[957 | 1225.31] loss=0.09 avg=0.08\n",
"[958 | 1226.32] loss=0.24 avg=0.08\n",
"[959 | 1227.33] loss=0.07 avg=0.08\n",
"[960 | 1228.34] loss=0.10 avg=0.08\n",
"[961 | 1229.34] loss=0.10 avg=0.08\n",
"[962 | 1230.35] loss=0.10 avg=0.08\n",
"[963 | 1231.36] loss=0.07 avg=0.08\n",
"[964 | 1232.36] loss=0.06 avg=0.08\n",
"[965 | 1233.37] loss=0.07 avg=0.08\n",
"[966 | 1234.38] loss=0.04 avg=0.08\n",
"[967 | 1235.39] loss=0.04 avg=0.08\n",
"[968 | 1236.39] loss=0.06 avg=0.08\n",
"[969 | 1237.40] loss=0.05 avg=0.08\n",
"[970 | 1238.41] loss=0.06 avg=0.08\n",
"[971 | 1239.42] loss=0.03 avg=0.08\n",
"[972 | 1240.43] loss=0.08 avg=0.08\n",
"[973 | 1241.44] loss=0.05 avg=0.08\n",
"[974 | 1242.44] loss=0.05 avg=0.08\n",
"[975 | 1243.45] loss=0.05 avg=0.08\n",
"[976 | 1244.46] loss=0.04 avg=0.08\n",
"[977 | 1245.47] loss=0.03 avg=0.08\n",
"[978 | 1246.48] loss=0.06 avg=0.08\n",
"[979 | 1247.48] loss=0.08 avg=0.08\n",
"[980 | 1248.49] loss=0.08 avg=0.08\n",
"[981 | 1249.50] loss=0.05 avg=0.08\n",
"[982 | 1250.51] loss=0.05 avg=0.08\n",
"[983 | 1251.52] loss=0.05 avg=0.08\n",
"[984 | 1252.53] loss=0.12 avg=0.08\n",
"[985 | 1253.54] loss=0.03 avg=0.08\n",
"[986 | 1254.54] loss=0.08 avg=0.08\n",
"[987 | 1255.56] loss=0.04 avg=0.07\n",
"[988 | 1256.56] loss=0.07 avg=0.07\n",
"[989 | 1257.57] loss=0.03 avg=0.07\n",
"[990 | 1258.58] loss=0.08 avg=0.07\n",
"[991 | 1259.59] loss=0.04 avg=0.07\n",
"[992 | 1260.60] loss=0.08 avg=0.07\n",
"[993 | 1261.60] loss=0.05 avg=0.07\n",
"[994 | 1262.61] loss=0.04 avg=0.07\n",
"[995 | 1263.62] loss=0.05 avg=0.07\n",
"[996 | 1264.63] loss=0.06 avg=0.07\n",
"[997 | 1265.64] loss=0.03 avg=0.07\n",
"[998 | 1266.65] loss=0.06 avg=0.07\n",
"[999 | 1267.66] loss=0.13 avg=0.07\n",
"[1000 | 1268.67] loss=0.06 avg=0.07\n",
"Saving checkpoint/run18/model-1000\n"
]
}
],
"source": [
"import gpt_2_simple as gpt2\n",
"import os\n",
"import tensorflow as tf\n",
"\n",
"\n",
"model_name = \"774M\"\n",
"if not os.path.isdir(os.path.join(\"models\", model_name)):\n",
" print(f\"Downloading {model_name} model...\")\n",
" gpt2.download_gpt2(model_name=model_name) # model is saved into current directory under /models/124M/\n",
"\n",
"tf.reset_default_graph()\n",
"file_name = \"blog.txt\"\n",
"\n",
"sess = gpt2.start_tf_sess()\n",
"gpt2.finetune(sess,\n",
" file_name,\n",
" model_name=model_name,\n",
" learning_rate=1e-5,\n",
" run_name=\"run18\",\n",
" steps=1000) "
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Loading checkpoint checkpoint/run18/model-1000\n",
"INFO:tensorflow:Restoring parameters from checkpoint/run18/model-1000\n",
"This is a blog about innovation driven by digital transformation. It's about asking the right questions, figuring them out, and doing the hard work of reshaping yourself to become more agile, self-sufficient, and focused on the long-term health of your business.\n",
"I am honored and privileged to lead Cisco's Digital Transformation effort. As the leader in bringing digital to business, we are working every day to bring value to our customers across the full lifecycle of their digital transformation. Our ambition is that by 2020, every business across the globe will be able to say that their digital transformation strategy has a high degree of success: does it involve online purchasing?\n",
"That's a big ask, and one we haven't seen from other companies in the industry. Our competitors are literally rolling out digital solutions for no other reason than to increase their market share by creating additional channels of sale. This strategy has two components: selling online and selling through channels.\n",
"Selling Online\n",
"The easiest way to sell online is with an online store. Over the last three years, we have seen the growth of two primary type of online stores: premium sites and low-cost sites. The premium site generates significant revenues through premium customer experiences and selling to an affluent audience. The low-cost site absorbs a larger portion of a company's total sales and provides the same or similar products for a lower price.\n",
"As our industry has gotten more sophisticated, companies have recognized the need to create separate spaces for their online stores. The CIO Reading List has now reached the point where we can say with confidence that we will see more low-cost sites in the next year or two. The idea is that companies can create separate spaces to showcase the same high-quality products, but with lower costs, when one or both of the sites experiences a significant drop in traffic.\n",
"The internet has changed the way we shop, work and play. With the right planning and effort, you can build any type of digital marketing plan. There are many helpful resources on building a digital marketing plan, including one from our friends at Smart Insights.\n",
"But in the meantime, I do want to focus your attention on a few things you can do to set you on the right path.\n",
"First, it's time to stop working from the assumption that your online marketing program will stay the same. It will change, but not often and not quickly. You need to identify five years' worth of digital marketing plans and ditch the plans from one year to the next. This will ensure you don't have to re-organize for each new challenge.\n",
"Second, you need to eliminate all doubt. Once you have the plan, there is no turning back. Stop and think about how your digital marketing plan will be received. Do you have support? If not, start the process of seeking support now.\n",
"Third, implement! Do this now, before you get started on other projects, so you can make sure you are on the right track. Once you have set goals and criteria for your digital marketing, it's time to implement.\n",
"Fourth, choose a brand. It may be the most important decision you will make as a digital marketing professional. Who is going to see the plan and make the decisions? Your name will be attached to the website or social media account. Who will see the results? Customers or email addresses.\n",
"Last, but not least, use the database. Your database has been designed to support traditional marketing strategies. As a digital marketing professional, you have just added another table to the list.\n",
"The future of marketing is going digital. But how are you going to get started? By designing your website and other digital marketing strategies around a digital marketing plan, you have a good chance of success.\n",
"The web is full of information about digital marketing. However, most of it is geared towards growing a small business. Although it may seem like a simple process, it can be difficult to master. That's why we at Small Business Network think you should have a digital marketing plan with you when you start your new job.\n",
"A digital marketing plan gives you a clear picture of how your business is going to use digital media to achieve your goals. It includes a list of goals which you need to achieve by the end of the year, a list of strategies to achieve these goals and a plan to monitor your progress.\n",
"A digital marketing plan has been essential for many companies to achieve the outcomes from their digital marketing plans. In fact, it was the traditional practice within the marketing industry. However, as the Internet has changed, so has the way companies market and use digital media.\n",
"As a result, a digital marketing plan is required for every company today. Digital media and technology have become an integral part of many industries; business models, marketing techniques and even cultures.\n",
"It's time to change the way companies market and use digital media. The pace of change is accelerating every day and it's impacting how companies market and use digital media now.\n",
"According to a recent report by InformaticI'd Global\n"
]
}
],
"source": [
"import gpt_2_simple as gpt2\n",
"import tensorflow as tf\n",
"\n",
"tf.reset_default_graph()\n",
"sess = gpt2.start_tf_sess()\n",
"gpt2.load_gpt2(sess, run_name=\"run18\")\n",
"gpt2.generate(sess, run_name=\"run18\", prefix=\"This is a blog about innovation driven by digital transformation.\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.9"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment