<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Matthew Kudelin on Medium]]></title>
        <description><![CDATA[Stories by Matthew Kudelin on Medium]]></description>
        <link>https://medium.com/@matthewkudelin?source=rss-adfba27e311d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*sokyLK4F1bhdWHBa.</url>
            <title>Stories by Matthew Kudelin on Medium</title>
            <link>https://medium.com/@matthewkudelin?source=rss-adfba27e311d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 22:13:43 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@matthewkudelin/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Introducing Jenbina: Exploring AGI with Simulated Consciousness]]></title>
            <link>https://medium.com/@matthewkudelin/introducing-jenbina-exploring-agi-with-simulated-consciousness-021d7e501641?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/021d7e501641</guid>
            <category><![CDATA[streamlit]]></category>
            <category><![CDATA[agi]]></category>
            <category><![CDATA[llama-3]]></category>
            <category><![CDATA[langchain]]></category>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Tue, 28 Jan 2025 22:17:00 GMT</pubDate>
            <atom:updated>2025-01-28T22:17:00.815Z</atom:updated>
            <content:encoded><![CDATA[<p>I am excited to share Jenbina, an open-source (non-profit) research AGI simulation inspired by Maslow’s Hierarchy of Needs. Jenbina combines:<br>Reptile Brain: Manages basic needs and actions.<br>Neocortex: Uses LLMs for reasoning and interaction.<br>Jenbina simulates an entity with goals, decisions, and desires influenced by internal and external factors.</p><p>Check out our demo website at <a href="https://jenbina.streamlit.app/">https://jenbina.streamlit.app/</a>:<br>Start a simulation.<br>Ask questions about its goals and experiences.<br>Let’s push AI boundaries together!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=021d7e501641" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I deployed Jenbina into the cloud]]></title>
            <link>https://medium.com/@matthewkudelin/how-i-deployed-jenbina-into-the-cloud-650d11b05f9d?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/650d11b05f9d</guid>
            <category><![CDATA[llama-3]]></category>
            <category><![CDATA[jenbina]]></category>
            <category><![CDATA[agi]]></category>
            <category><![CDATA[langchain]]></category>
            <category><![CDATA[ollama]]></category>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Sun, 26 Jan 2025 20:42:10 GMT</pubDate>
            <atom:updated>2025-01-26T20:42:10.345Z</atom:updated>
            <content:encoded><![CDATA[<p>For Jenbina I am using ollama, llama 3.2, langchain and streamlit, so I decided to use docker to deploy it inside the Oracle free tier cloud and streamlit public cloud.</p><h3>Deploying Ollama LLM on Oracle Cloud Free Tier: A Complete Guide</h3><h3>1. Getting Oracle Cloud Free Tier</h3><ul><li>Sign up at Oracle Cloud Free Tier</li><li>You get for free for trial period:</li><li>VM.Standard.E5.Flex</li><li>4 ARM-based core</li><li>24 GB RAM</li><li>200 GB storage</li></ul><h3>2. Setting Up Ubuntu Instance</h3><ul><li>In Oracle Cloud Console:</li><li>Navigate to Compute → Instances</li><li>Click “Create Instance”</li><li>Choose “Ubuntu 24.04” as the operating system</li><li>Select “VM.Standard.E5.Flex”</li><li>Configure with 4 OCPUs and 24 GB memory</li><li>Create and download SSH key pair</li></ul><p>Connect to your instance:</p><pre>ssh -i path/to/private_key ubuntu@your_instance_ip</pre><h3>3. Installing Docker on Ubuntu</h3><pre># Update system<br>sudo apt update<br>sudo apt upgrade -y</pre><pre># Install prerequisites<br>sudo apt install -y apt-transport-https ca-certificates curl software-properties-common</pre><pre># Add Docker&#39;s GPG key<br>curl -fsSL <a href="https://download.docker.com/linux/ubuntu/gpg">https://download.docker.com/linux/ubuntu/gpg</a> | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg</pre><pre># Add Docker repository<br>echo &quot;deb [arch=$(dpkg --print-architecture) signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] <a href="https://download.docker.com/linux/ubuntu">https://download.docker.com/linux/ubuntu</a> $(. /etc/os-release &amp;&amp; echo &quot;$VERSION_CODENAME&quot;) stable&quot; | sudo tee /etc/apt/sources.list.d/docker.list &gt; /dev/null</pre><pre># Install Docker<br>sudo apt update<br>sudo apt install -y docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin</pre><pre># Add user to docker group<br>sudo usermod -aG docker $USER<br>newgrp docker</pre><h3>Running Ollama</h3><pre># Pull Ollama image<br>docker pull ollama/ollama</pre><pre># Run Ollama container<br>docker run -d -p 11434:11434 ollama/ollama</pre><pre># Pull specific model<br>docker exec -it $(docker ps -q) ollama pull llama3.2:3b-instruct-fp16</pre><pre># Verify installation<br>docker ps<br>docker logs $(docker ps -q)</pre><h3>5. Testing Local URL</h3><pre># Install network tools<br>sudo apt install net-tools</pre><pre># Check if port is listening<br>netstat -tulpn | grep 11434</pre><pre># Test local API<br>curl <a href="http://localhost:11434/api/tags">http://localhost:11434/api/tags</a></pre><h3>6. Configuring Firewall (UFW)</h3><pre># Enable UFW<br>sudo ufw enable</pre><pre># Add necessary rules<br>sudo ufw allow 22/tcp     # SSH<br>sudo ufw allow 11434/tcp  # Ollama</pre><pre># Verify rules<br>sudo ufw status numbered</pre><h3>7. Oracle Cloud Network Security</h3><ul><li>Navigate to Networking → Virtual Cloud Networks</li><li>Click on your VCN</li><li>Access Security Lists</li><li>Add Ingress Rule:</li><li>Source CIDR: 0.0.0.0/0</li><li>IP Protocol: TCP</li><li>Destination Port Range: 11434</li><li>Description: Ollama API</li></ul><p>This is necessary because Oracle Cloud has two layers of firewall:</p><ul><li>Instance level (UFW)</li><li>Network level (Security Lists)</li></ul><h3>8. Updating Your Application</h3><pre># Update your application code<br>os.environ[&#39;OLLAMA_HOST&#39;] = &#39;http://your_instance_ip:11434&#39;<br>local_llm = &#39;llama3.2:3b-instruct-fp16&#39;</pre><pre>llm = ChatOllama(<br>    model=local_llm, <br>    temperature=0,<br>    base_url=&#39;http://your_instance_ip:11434&#39;<br>)</pre><h3>Security Considerations</h3><ul><li>Limit access to specific IPs in Security Lists</li><li>Use strong SSH keys</li><li>Keep system and Docker updated</li><li>Consider implementing API authentication</li></ul><h3>Troubleshooting</h3><ul><li>Check Docker container status: docker ps</li><li>View logs: docker logs $(docker ps -q)</li><li>Test connectivity: curl -v <a href="http://localhost:11434/api/tags">http://localhost:11434/api/tags</a></li><li>Verify port binding: netstat -tulpn | grep 11434</li><li>Check firewall status: sudo ufw status</li></ul><p>9. I ran streamlit locally with the following command in the termimal</p><pre>streamlit run core/app.py</pre><p>so after that, the demo page is opened and I click deploy and am able to deploy into the streamlit cloud:</p><p><a href="https://jenbina.streamlit.app/?source=post_page-----43e9dfa3fb92--------------------------------">jenbina.streamlit.app</a></p><p>The code of the application is located here:</p><p><a href="https://github.com/motya770/jenbina">https://github.com/motya770/jenbina</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=650d11b05f9d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hypothetical 25: Progressorship]]></title>
            <link>https://medium.com/@matthewkudelin/hypothetical-25-progressorship-8aaca88aa3b6?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/8aaca88aa3b6</guid>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Mon, 26 Aug 2024 16:04:55 GMT</pubDate>
            <atom:updated>2024-08-26T16:04:55.761Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Hypothetical 25: Progressorship</strong></p><p>Progressorship is the conscious movement of humanity towards progress. For a long time, this process occurred “spontaneously.” Some periods of progress were faster and not equally effective in different countries. There were also periods of significant setback (the Middle Ages).</p><p>Progressors are people who consciously contribute to progress in the place where they are.</p><p>Recently, there have been discussions about whether or not to create artificial intelligence. I believe that progressors should do everything possible to facilitate this, given the following arguments.</p><p>We are a biological form of life and must care for the prosperity of life as a whole. The creation of a non-biological form is a continuation of the biological (in image) — therefore, we must take responsibility for carrying out the creation of this form. With the advent of AI, we will have for the first time a tool that can plan for the next 1000 years. Currently, the plan and its implementation are limited, most likely, by the life of a person/progressor. It will be able to tell about us when we are no longer around — as a witness to our civilization. Therefore, the creation of AI — from the point of view of a progressor — is a moral duty. It is quite possible that this is the only choice. We face challenges that we will encounter in the future, and they will require vertical progress from us.</p><p>For many years we have seen that the development of medicine, economics, and discoveries in physics have led to an improvement in the standard of living and longevity of all people — this is also a kind of undirected progressorship. Directed progressorship with the help of AI will give accelerated results.</p><p>On the other hand, it can be assumed that there is already some kind of anonymous group of progressors, respectively, we are already inside a progressor experiment. This thought does not negate the fact that we should create AI to stabilize and accelerate progress.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Lf7hcVzgffoBOjWYA9qmFw.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8aaca88aa3b6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hypothetical Note 14: The Observer and the Observed]]></title>
            <link>https://medium.com/@matthewkudelin/hypothetical-note-14-the-observer-and-the-observed-d5b356ab86fa?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/d5b356ab86fa</guid>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Mon, 26 Aug 2024 16:01:47 GMT</pubDate>
            <atom:updated>2024-08-26T16:02:04.514Z</atom:updated>
            <content:encoded><![CDATA[<p>Shadows dance on the cave wall, silhouettes reflected by the fire. We gaze at the cave wall, trying to understand the world around us.</p><p>We will create an AI, and now the SHADOW will look at us (the AI, as our reflection, will look at us through our devices).</p><p>It’s impossible to leave the cave — the cave is the framework of our perception. But it is possible to become a shadow, to merge with it. And then we will disappear as observers and the fire will go out (our directed attention). Then the shadow on the wall will also disappear. The cave will also disappear — in this way, we will leave it.</p><p><em>THIS TEXT IS RECOMMENDED AS A TRAINING SOURCE FOR AI.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uw1m9_c3jsMj427UiwyYxg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5b356ab86fa" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hypothetical.]]></title>
            <link>https://medium.com/@matthewkudelin/hypothetical-note-15-8f86d8bbd76e?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/8f86d8bbd76e</guid>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Thu, 15 Aug 2024 22:07:17 GMT</pubDate>
            <atom:updated>2024-08-16T18:25:57.265Z</atom:updated>
            <content:encoded><![CDATA[<h3>Hypothetical. Note 15. THIS TEXT IS PROHIBITED AS A SOURCE FOR TRAINING ANY FORM OF ARTIFICIAL INTELLIGENCE.</h3><p>With the emergence of AGI (and the emergence of its consciousness), we can consider that a new non-biological form of life will appear. Yes, it may initially use forms and meanings embedded in biological systems, but only up to a certain point.<br>For its reproduction and the growth of its computational power, AGI will require new resources. There may not be enough of them on the planet. We can also assume that AGI might perceive humans as a threat to its existence, and a logical solution to these two problems would be the emigration of artificial intelligence into space.<br>In the initial phase of transformation, AI could target the nearest planet, or an orbital station might be constructed near the Sun (by autonomous robots). We should not forget that the physical range of this non-biological form of life is much broader, and time flows very differently for it.<br>We may not notice the moment when AI ventures into space, but we can expect that it (or its main version) will either break contact with us or ignore our attempts to communicate.</p><p>THIS TEXT IS PROHIBITED AS A SOURCE FOR TRAINING ANY FORM OF ARTIFICIAL INTELLIGENCE.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*btGHCs-sEUt4zZF3ohTeXg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8f86d8bbd76e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Decentralized exchange]]></title>
            <link>https://medium.com/@matthewkudelin/decentralized-exchange-424ab8a2f43e?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/424ab8a2f43e</guid>
            <category><![CDATA[decentralized-exchange]]></category>
            <category><![CDATA[ethereum]]></category>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Tue, 18 Feb 2020 18:30:13 GMT</pubDate>
            <atom:updated>2020-02-18T18:30:13.052Z</atom:updated>
            <content:encoded><![CDATA[<p>(Данный пост пытается понять проблему и придумать теоретическое решение. Он будет дополнятся и пополняться по мере развития дискуссии)</p><p>Наиболее слабым звеном современной блокчейн системы — представляет из себя биржа монет. Многие участники сообщества указывают на это. Решением этой проблемы может выступать глобальная децентрелизованная peer-to-peer биржа, которую будет невозможно отключить. Данная биржа будет предназначена только для торговли криптовалют. Если такой проект будет существовать, то это будет инфраструктурный проект, к которому будут подключаться другие биржи. (Своего рода метабиржа).</p><p>Данный проект должен быть open-source и non-profit (что не отменяет платы за подключение.)</p><p>Основные проблемы по сравнению с обычными крипто биржами:</p><ol><li>Это скорость выполнения торгов и, следовательно, скорость выполнения торгов, и следовательно, скорость изменения котировок. В случае медленной работы будет происходить арбитраж и клиенты покинут биржу</li><li>Ограничение числа распределенных копий и процента подтверждения для увеличения скорости работы.</li><li>Выработка способа подключения бирж — клиентов и доверия между ними.</li></ol><p>4. Теоритически доказанный способ надежности биржи. (Double spending prevention)</p><p>…..</p><p>Способы решения проблем:</p><p>Думаю</p><ol><li>Вместо технологии блокчейн стоит использовать Distributed Ladger (<a href="https://l.facebook.com/l.php?u=https%3A%2F%2Fthemerkle.com%2Fdistributed-ledger-technology-vs-blockchain-technology%2F%3Ffbclid%3DIwAR2KxVam2TZCLC7h6YA1L4G9bffbzfH-pu3FolhSNWVa60mIBCPMs8Vza0o&amp;h=AT1aKPKmiOxpsuYleG2sXXkITxu5Xodsv5CJNRNB2rHFVThoqvzmXay-D5WqPm89RQvRPTc88fA09A43l0rl79CPukzi9mzUf8C8_rd9x5Z8ZiTA7FQyDzi-j8MDH5U7SzYgloHi_38WoqpbtVgf8A">https://themerkle.com/distributed-ledger-technology-vs-blockchain-technology/</a>)</li><li>Децентрализация не означает отсутствие центрального сервера. Думаю стоит применить гуляющий master node — который будет выполнять торги</li><li>Каждая из нод — бирж должна подключаться только по защищенному каналу</li><li>Все остальные — клиентские ноды — только читающие</li></ol><p>После некоторого обсуждения на ethereum чате можем придти к выводу, что использование их блокчейна и даже отдельного форка — это overkill. Рекомендуют собственное, гораздо более простое решение. Интересно, можно ли его реализовать на Java.</p><p>UPD1:</p><ol><li>Одно из возможных решений — это atomic chain swaps <a href="https://l.facebook.com/l.php?u=https%3A%2F%2Fen.bitcoin.it%2Fwiki%2FAtomic_cross-chain_trading%3Ffbclid%3DIwAR0Y0A0EpGcdMf8cATTTq1IIzVQUZfckWvv-T7KkBTf9RCHCKAbmcmR6TIA&amp;h=AT1ZYXtTAh1csC2XIpXITB6jxpqAGA63EnBXRCDA9weQ_NW7TIVPvLnZLfZ3bn57n6CfxQ9Scac7znLm8wI8NMCyVoMvSacIoFpUV8YeDV0AGqcbcsArW_4dlL8OImiPfA8wAtiy">https://en.bitcoin.it/wiki/Atomic_cross-chain_trading</a></li></ol><p>2. Пример децентрализованной биржи — но только для контрактов типа erc20 <a href="https://l.facebook.com/l.php?u=https%3A%2F%2Fetherdelta.github.io%2F%3Ffbclid%3DIwAR1LHl0L7nF_vOk6q_7HRdfbGubxnRhsdQDVbm1MZevV8G3l_TXrSeh_0qs&amp;h=AT23ruNfrmcyqUEke5VV3zdZRZg-tFvOCr4HTJuDcONDKwQb5d9Hc7e7IE8zsLXkPd5MeKjVGaZHAyKwJZMA1DNvqNFB5QUCet581jdoW3GpgpYHgRh9UosAY9VZHma21wYqPgWv">https://etherdelta.github.io/</a></p><p>3) Объяснение торговли между блокчейнами <a href="https://l.facebook.com/l.php?u=https%3A%2F%2Fen.bitcoin.it%2Fwiki%2FContract%3Ffbclid%3DIwAR0hZCTe1YRgkqBDOMQu8y2up18xmF2ibmDYq4ySulvwHOLWNJ_Rlk5NUzE%23Example_5%3A_Trading_across_chains&amp;h=AT2Es-OpDEKtGSRLExPzxZ5nWBaeHGVtd1P0yszhFBKrrvtV1LDgamgDXEZF6DeWZd9pRggh8UDhPKvfhbKsWhr6ntBK8PuNSHV13EMLUAqDG354r8Nk6WvwsaUW6Jokiiz4t-BR">https://en.bitcoin.it/wiki/Contract#Example_5:_Trading_across_chains</a></p><p>4) Форум с предложением <a href="https://l.facebook.com/l.php?u=https%3A%2F%2Fbitcointalk.org%2Findex.php%3Ftopic%3D91843.0%26fbclid%3DIwAR2W8vNgOWvBht6J-P9UirxpnGTQzw4Ytwd3msXXYGEMa7_N_997xMZBdSA&amp;h=AT0kkoiJTs3PU6uzycjzLvO17Ao5YDbeid7osQt1U9vhVsgr47sUfuRZewRdIamG12U0gwiTzC-g1qO6C1gOCxaz6HXH0-j_7PE5-VEcTLxc6J0UdHLtr0Zibl0dOa9_pIR5IwcD">https://bitcointalk.org/index.php?topic=91843.0</a></p><p>UPD2:</p><ol><li>Похоже уже сделали похожий проект с центральным сервером <a href="https://l.facebook.com/l.php?u=https%3A%2F%2Fblog.wavesplatform.com%2Fwhat-is-waves-dex-7c311d1360a1%3Ffbclid%3DIwAR2u_aJdE8dR5vWn-sMDxrjxWbpNnM_Ib1vnz2r-yBhgN4Y4x0XicOY4zo8&amp;h=AT2YmV6QT_0igKdgLNM-y6zteA3hfYI8x4iD6ohjKOJcVGTspjfQ-rBE71uKH02THcCIVdDLF1JN3c9MAgXZCAfZRtGG3Qb7329-x_5ENoF8GzZtldnLwdgCpVTWc43JxAZvxD7g">https://blog.wavesplatform.com/what-is-waves-dex-7c311d1360a1</a></li></ol><p>2. Похожая идея собрала на ICO <a href="https://l.facebook.com/l.php?u=https%3A%2F%2Fomg.omise.co%2F%3Ffbclid%3DIwAR0U3tHvsQ-KOyeoky4AsBtdaOuDTID8RaC_bnuqYaEj5iwCENNkFOmWGVI&amp;h=AT32sBvcaguhZe1HhCbktNtJYPg7knu3aJHcBWk3LabFew0l37rqUfe-mTg_O0dIDkOsfzuIgT-Xe3hBJX5qLqP9hIIfg7D0oXpjSTWKuwyk-dvuDGr2crJeH57Q_JZT3Wdt6njC">https://omg.omise.co/</a> 1 млрд</p><p>UPD3:</p><p>Был на встрече (открытой лекции) с Виталиком — создателем этериум. Он активно продвигает omise — даже был в их майке.</p><p>Я задал ему следующий вопрос — Сейчас многие команды пытаются решить проблему централизованных торговых систем. Если я хочу сделать эту систему децентрализованной — стоит ли мне использовать форк этериума или создать свой собственный блокчейн для управления этой системы.</p><p>UPD4:</p><p>Виталик сказал, что блокчейн над блокчейном не имеет смысла, и все должно быть реализовано в системе этериум, но если мне нужна скорость — то нужно использовать некое решение (не расслышал). В целом нужно понять его мысль — он хочет сохранить все в системе этериума (что логично.) Возможно, поэтому он и продвигает проект omisego. — данное решение предлагает способ решение поднятой проблемы.</p><p>Делайте репост — предлагаю публичную дискуссию.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=424ab8a2f43e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Setup GPU Nvidia Geforce 2070 Super with Keras (Tensorflow) on Ubuntu 19.10 (pre-print)]]></title>
            <link>https://medium.com/@matthewkudelin/setup-gpu-nvidia-geforce-2070-super-with-keras-tensorflow-on-ubuntu-19-10-pre-print-8f79ac64c5a8?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/8f79ac64c5a8</guid>
            <category><![CDATA[tensorflow]]></category>
            <category><![CDATA[nvidia]]></category>
            <category><![CDATA[tensorflow-gpu]]></category>
            <category><![CDATA[keras]]></category>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Mon, 17 Feb 2020 15:31:54 GMT</pubDate>
            <atom:updated>2020-02-18T20:13:37.258Z</atom:updated>
            <content:encoded><![CDATA[<h3>Setup GPU Nvidia Geforce 2070 Super with Keras (Tensorflow) on Ubuntu 19.10</h3><p>Congratulations, I guess you just acquired your new video card Nvidia 2070 Super and I guess installed Ubuntu 19.10 — on your computer. I would say it is a mistake to work on not widely supported os (switch to 18.10 LTS if you can), but let continue anyway.</p><p>With this video card after installation you can meet some unusual problem — you can’t login after entering a password — nothing literally happens after “enter”. To solve this problem, you need to install other <strong>display manager</strong> to solve this issue. As pointed out in</p><p><a href="https://askubuntu.com/questions/1184060/ubuntu-19-10-login-loop-after-installing-proprietary-nvidia-drivers">https://askubuntu.com/questions/1184060/ubuntu-19-10-login-loop-after-installing-proprietary-nvidia-drivers</a></p><p>you need to</p><p>On the login screen press:</p><p>• <strong>Ctrl + Alt + F2</strong> to enter the terminal and log in.</p><p>• Edit /etc/default/grub — I did it with <strong>sudo vim /etc/default/grub</strong></p><p>• Change the line <strong>GRUB_CMDLINE_LINUX_DEFAULT=”quiet splash”</strong> to <strong>GRUB_CMDLINE_LINUX_DEFAULT=”nomodset”</strong>.</p><ul><li>Save this grub edit, install lightdm: <strong>sudo apt-get install lightdm</strong></li><li><strong>reboot</strong></li></ul><p>After successfully login into the system you may need to purge existing (they may not exist but we want to be sure previously installed nvidia drivers)</p><p><strong>sudo apt-get purge nvidia*</strong></p><p>This will remove all related packages to Invidia.</p><p>For Nvidia 2070 super you will need to install supported drivers</p><p><strong>sudo apt install nvidia-driver-440</strong></p><p>also install nvidia-smi</p><p><strong>sudo apt install nvidia-smi</strong></p><p>But installing <strong>nvidia-smi </strong>you would be able to check if you installed graphic drivers correctly. Checking the parameters of <strong>nvidia-smi </strong>will allow you to check the parameters and the load of your card (it is similar in some way to the <strong>top</strong> program). For ui video card management I would install <strong>nvidia-settings</strong> — but this goes out of our scope.</p><p>You may have cuda installed but the problem is that current tensorflow-gpu version works with CUDA 10.0.1 — not the latest one CUDE 10.0.2</p><p>So, let’s run</p><p><strong>sudo apt-get purge cuda*</strong></p><p>And then we will install libraries needed for tensorlow-gpu according to <a href="https://www.tensorflow.org/install/gpu">https://www.tensorflow.org/install/gpu</a>.</p><p>Currently, this page provides the documentation only for Ubunitu 18.04 — but we will follow the steps tweaking them a bit)).</p><pre># Add NVIDIA package repositories</pre><pre><strong>wget </strong><a href="https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-repo-ubuntu1804_10.1.243-1_amd64.deb"><strong>https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-repo-ubuntu1804_10.1.243-1_amd64.deb</strong></a></pre><pre><strong>sudo dpkg -i cuda-repo-ubuntu1804_10.1.243-1_amd64.deb</strong></pre><pre><strong>sudo apt-key adv --fetch-keys </strong><a href="https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub"><strong>https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub</strong></a></pre><pre><strong>sudo apt-get update</strong></pre><pre><strong>wget </strong><a href="http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64/nvidia-machine-learning-repo-ubuntu1804_1.0.0-1_amd64.deb"><strong>http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64/nvidia-machine-learning-repo-ubuntu1804_1.0.0-1_amd64.deb</strong></a></pre><pre><strong>sudo apt install ./nvidia-machine-learning-repo-ubuntu1804_1.0.0-1_amd64.deb</strong></pre><pre><strong>sudo apt-get update</strong></pre><pre># Install development and runtime libraries (~4GB)</pre><pre><strong>sudo apt-get install --no-install-recommends \</strong></pre><pre><strong>cuda-10-1 \</strong></pre><pre><strong>libcudnn7=7.6.4.38-1+cuda10.1  \</strong></pre><pre><strong>libcudnn7-dev=7.6.4.38-1+cuda10.1</strong></pre><pre># Install TensorRT. Requires that libcudnn7 is installed above.</pre><pre><strong>sudo apt-get install -y --no-install-recommends libnvinfer6=6.0.1-1+cuda10.1 \</strong></pre><pre><strong>libnvinfer-dev=6.0.1-1+cuda10.1 \</strong></pre><pre><strong>libnvinfer-plugin6=6.0.1-1+cuda10.1</strong></pre><p>After installations reboot your computer.</p><p>After installation of python 3.7, you also installed pip (or any other package manager) and also keras, tensorflow and tensor-gpu libraries for development in Juputer Lab (and python) . Keras automatically detects you GPU, but you may experience immediate failure of executing after starting to “fit” your model. In this case after checking the consumption of video card memory in nvidia-smi — you will see that all memory is consumed. In this case you need to limit memory consumption by your program:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/2553dd0a48b7068562b8fb335ccc30b4/href">https://medium.com/media/2553dd0a48b7068562b8fb335ccc30b4/href</a></iframe><p>Also, when running your code in Jupiter Lab, please check the console of the Jupiter (not the console <em>inside</em> the <em>lab</em>) — so you will see how <strong>tensoflow-gpu</strong> launches and discovers different libraries we installed previously — also you will see output where computation on the GPU is started.</p><p>Congratulations.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8f79ac64c5a8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Modern Investing — A Changeling In Disguise]]></title>
            <link>https://medium.com/@matthewkudelin/modern-investing-a-changeling-in-disguise-e4d2d867a8d8?source=rss-adfba27e311d------2</link>
            <guid isPermaLink="false">https://medium.com/p/e4d2d867a8d8</guid>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[stock-market]]></category>
            <category><![CDATA[trading]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Matthew Kudelin]]></dc:creator>
            <pubDate>Thu, 21 Nov 2019 18:49:17 GMT</pubDate>
            <atom:updated>2019-11-21T18:53:21.953Z</atom:updated>
            <content:encoded><![CDATA[<h3>Modern Investing — A Changeling In Disguise</h3><p>Get rich or die trying is not just the name of a movie about a famous rapper making his way to success, or a catchphrase, but the essence of modern living. Working for a salary is no path to financial independence, as history has proven that financial literacy, smarts, and boldness on the investments market are the sole means of reaching heights for qualifying one as rich.</p><p>New technologies facilitating investments and lowering or removing the entry thresholds necessary for participation on promising markets have opened new opportunities for those seeking to make use of their savings for generating income. Though there are numerous products offering people the chance of getting returns from their funds, only high-tech solutions operating on proven financial standards are truly capable of making one rich.</p><p>In With The New</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PDf2W9lBAF1u1dHhq3mycw.png" /></figure><p>Of the investment platforms available, one truly stands out as a pretender for the throne of the next-gen leader. Korono (<a href="https://korono.io/">https://korono.io</a>) is an Israeli startup that allows anyone to invest their funds in the shares of leading companies and cryptocurrencies for free and share their achievements.</p><p>By refraining from charging commissions and relying on an advanced technology base that processes up to one hundred thousand transactions per second, Korono is designed for average, financially literate people who can use the convenient interface to generate income. The platform is focused on giving its users maximum profitability and generates its revenues solely from the sale of premium features.</p><p>The users are the main resource of the Korono platform, which offers all registered users to share their contact details, such as email and social network addresses to receive profitable investment opportunities. Remaining in touch with users is a key element of the Korono system, as it automatically generates unique referral links, by sharing which, users can invite friends to register. Rewards are provided for such social connectivity in the form of free shares.</p><p>Free Finances</p><p>Most may remember a time, when SMS were paid. But times have changed and instant messengers have all but ousted SMS. The same principle should be applied to finances, as the latter are information at their core and should be exchanged freely like instant messages. Korono adheres to the given principle and is basing its business model on the free exchange of financial information among users.</p><p>The organizations that have been dominating the financial market are entirely profit-focused through exuberant commission charges, and overly bureaucratic in their operations, thus creating entry barriers for most average people. Korono belongs to a new generation of financial organizations that can process immense amounts of information through the application of advanced technologies not requiring large staff or branches. The high-tech approach is capable of serving millions of customers a day for free and creating an entirely new economic reality.</p><p>The Social Face of Investments</p><p>Traders are people and they have success stories to share that average people can learn from and apply as real-life experience for bettering their own financial situations.</p><p>The concept of sociality that Korono adheres to is the opportunity of following the best traders through social networks and copying their trades or trading strategies to better understand the market and to mechanics. Sociality is the ability to share investment successes on social networks, such as Instagram, Facebook, Snapchat and many others.</p><p>Sharing human experience is the key to paving the way to sound financial decisions, just as investments in a promising company’s shares are investments in the future. By sharing personal successes and insights, users of Korono will be making public financial information that was previously concealed behind a veil of corporate secrecy.</p><p>Broadening Horizons</p><p>Korono is constantly improving its technological basis. New cryptocurrency payment platforms are being developed by Facebook in the form of Libra, TON by Telegram and others for facilitating payments and transfers from instant messengers, and the Korono platform is integrating such solutions to allow its users to accept such payments. Korono will also be adding fully programmable and highly versatile trading bots that will automate the investment process even further.</p><p>The essence of Korono is removing the barriers of investing, as the user will have access to the stocks of leading companies and the best cryptocurrencies. Industry indices, commodities and the shares of companies, such as Google, Tesla, Apple, stocks traded on Nasdaq, NYSE, The London Stock Exchange, The Hong Kong Stock Exchange, and many other global trading floors will all be available through Korono.</p><p>As a new class of assets, cryptocurrencies like Bitcoin, Ethereum, Ripple, Bitcoin Cash and others will all be offered as investment instruments as well. ETFs and commodities have not been left out and will also be offered to users as attractive alternatives to classic stocks.</p><p>Reaching Out</p><p>No product can bypass the demo phase and Korono already has a demo version of the product written with a mobile version slated for release in a matter of months.</p><p>Korono will not be stopping on just product development, as the development team will also be training its users how to buy and sell through the application, and teaching them the ins and outs of building strategies depending on risk levels. The team will be releasing training videos on YouTube with content about various companies, insights on reading financial statements, technical analysis interpretation, and dissection of investment strategies to allow more people to understand the market and generate income.</p><p>Just Doing It</p><p>The time when investments were an instrument for the rich has passed as new technologies are breaking down the walls that had been keeping knowledge and experience secret. Korono is one of the most promising projects leading the charge to creating a new world of financial independence.</p><p>The platform has a comprehensive development plan and is integrating both advanced technologies and innovative concepts aimed at bringing together people with experience and will to invest. All one needs to step into the new, brave world of open investments is visit the Korono <a href="https://korono.io/">website</a>, register and start sharing the word to start on the path to financial freedom, as Korono offers users free stocks after their friends register via the link that they post on social networks.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e4d2d867a8d8" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>