<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:video="http://www.google.com/schemas/sitemap-video/1.1">
  <url>
    <loc>https://novatubenews.live/en/watch/57cPmzwCqd4</loc>
    <lastmod>2025-04-02T15:36:52.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/57cPmzwCqd4/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>LLM-Driven Robotics: Training the Inspire RH56DFQ Hand to Crawl</video:title>
      <video:description>In this session, we push the limits of the Inspire RH56DFQ robotic hands by attempting to program a crawling motion using Large Language Models (LLMs). By leveraging the &quot;vibe coding&quot; approach with tools like Cursor and Claude 3.7 Sonnet, we explore how AI can assist in generating complex movement logic that would traditionally require manual animation or rigid, manual coding. The project involves testing various gestures and refining the hand&apos;s articulation to simulate a crawling gait. We also share the custom GitHub package developed during this process, illustrating the synergy between modern AI coding assistants and hardware robotics. This experiment highlights the potential for future robotic control systems to be driven by natural language and intelligent agents.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/57cPmzwCqd4</video:player_loc>
      <video:publication_date>2025-04-02T15:36:52.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
  <url>
    <loc>https://novatubenews.live/en/watch/MeHWIXLV3Zo</loc>
    <lastmod>2025-03-31T19:40:55.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/MeHWIXLV3Zo/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>Vibe Coding Robot Hands with Cursor: A New Approach to Control</video:title>
      <video:description>In this video, we explore the concept of &apos;vibe coding&apos; as we use Cursor and the 3.7 Sonnet language model to program the Inspire RH56 series robotic hands. This innovative approach allows us to generate control code quickly and intuitively, without getting bogged down in low-level details. We share our experience and the resulting Python package that we&apos;ve made available on GitHub. The video starts with an introduction to vibe coding and why it&apos;s a game-changer for robotics development. We then demonstrate how we use Cursor and the language model to generate code for various hand movements, from simple finger flexes to more complex gestures. We also discuss the challenges we faced and how we overcame them, including debugging and refining the generated code. The final package is a comprehensive library that makes it easy to control the Inspire hands. Whether you&apos;re a seasoned roboticist or a beginner, this video offers a fresh perspective on programming robots. The approach shown here can save you time and effort, allowing you to focus on the creative aspects of your projects. Check the GitHub link in the description to get the code and start vibe coding your own robots.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/MeHWIXLV3Zo</video:player_loc>
      <video:publication_date>2025-03-31T19:40:55.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
  <url>
    <loc>https://novatubenews.live/en/watch/kQ3yrLMe1ac</loc>
    <lastmod>2025-03-08T22:13:01.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/kQ3yrLMe1ac/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>Discovering Paradise: The 23 Most Stunning Islands Across the United States</video:title>
      <video:description>While the United States is vast, spanning fifty states and five territories, some of its most breathtaking landscapes are found off the mainland. These islands offer a remarkable level of geographic diversity that mirrors the variety of the country&apos;s interior. From rugged coastlines to tropical escapes, each location presents a distinct personality shaped by its unique flora, fauna, and environmental characteristics. Beyond just being scenic retreats, these islands are steeped in rich local culture and history. Travelers can explore the specific traditions and historical narratives that define each community, making every visit a learning experience. Whether you are looking for arresting natural beauty or a deep dive into maritime heritage, the wide range of options ensures there is a destination suited for every type of explorer. This comprehensive look at the nation&apos;s most beautiful islands showcases the hidden corners and famous archipelagos that make the U.S. coastlines so extraordinary.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/kQ3yrLMe1ac</video:player_loc>
      <video:publication_date>2025-03-08T22:13:01.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
  <url>
    <loc>https://novatubenews.live/en/watch/WKF__cJTxvg</loc>
    <lastmod>2025-02-16T01:08:55.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/WKF__cJTxvg/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>The Future of Development: Mastering LLM Agents in 2025</video:title>
      <video:description>Stay ahead of the curve with our guide to using LLM agents for software development in 2025. This video covers essential tips and tricks for leveraging modern agents like OpenHands to build applications and automate complex tasks. Whether you are using a web-based UI or a local agentic framework, these strategies will help you achieve more reliable and efficient results.We explore best practices for guiding agents through the development lifecycle, from initial architecture to final debugging. As these tools become more prevalent, understanding how to effectively collaborate with AI agents is a critical skill for any developer. Resources for OpenHands and other community links are provided in the description to help you get started.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/WKF__cJTxvg</video:player_loc>
      <video:publication_date>2025-02-16T01:08:55.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
  <url>
    <loc>https://novatubenews.live/en/watch/VyseRArtl5E</loc>
    <lastmod>2024-10-13T16:54:00.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/VyseRArtl5E/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>Developer Update: Greetings from the Ranch</video:title>
      <video:description>Welcome to a quick update and personal greeting from the ranch. While we usually focus on deep technical tutorials and robotics development, this video provides a moment to connect with the community and share the environment behind the projects. It&apos;s a brief look at the setting where the research and coding for the Unitree G1 and other AI systems take place.Stay tuned for more technical content coming soon. You can support the channel&apos;s ongoing research through memberships or by following our social media channels on Twitter, Instagram, and Twitch. We appreciate the continued support from the community as we explore the frontiers of robotics and artificial intelligence.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/VyseRArtl5E</video:player_loc>
      <video:publication_date>2024-10-13T16:54:00.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
  <url>
    <loc>https://novatubenews.live/en/watch/pCX_3p40Efc</loc>
    <lastmod>2024-03-06T19:01:15.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/pCX_3p40Efc/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>Constructing High-Quality Datasets for LLM Fine-Tuning</video:title>
      <video:description>This video provides a detailed walkthrough of the process involved in building a custom dataset for QLoRA fine-tuning. Specifically, it focuses on leveraging a massive collection of public Reddit comments to create a specialized training set, which was subsequently used to develop the Walls1337bot adapter for Llama 2. The tutorial covers the technical hurdles of processing large-scale data and formatting it correctly for language model training. By examining the WSB dataset and the 5000-step training process, viewers gain insight into the pipeline required to fine-tune a model to adopt a specific persona or domain-specific knowledge base.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/pCX_3p40Efc</video:player_loc>
      <video:publication_date>2024-03-06T19:01:15.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
  <url>
    <loc>https://novatubenews.live/en/watch/ChfEO8l-fas</loc>
    <lastmod>2024-02-14T18:23:24.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/ChfEO8l-fas/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>Inside the Black Box: Visualizing Neural Network Training and Inference</video:title>
      <video:description>Have you ever wondered what actually happens inside a neural network as it learns? In this video, we utilize custom visualization tools to peek into the internals of a model during both the training and inference phases. By observing how weights and activations change in real-time, we can gain a much deeper understanding of the learning process and the mathematics behind AI. This tutorial covers the implementation of visualization code and discusses the significance of the patterns observed within the layers. These insights are invaluable for debugging models and optimizing neural architecture. All source code is provided on GitHub, allowing you to run these visualizations on your own networks and explore the hidden dynamics of machine learning algorithms.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/ChfEO8l-fas</video:player_loc>
      <video:publication_date>2024-02-14T18:23:24.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
  <url>
    <loc>https://novatubenews.live/en/watch/mm9IHqgCbZc</loc>
    <lastmod>2024-02-07T19:22:33.000Z</lastmod>
    <changefreq>daily</changefreq>
    <priority>0.8</priority>
    <video:video>
      <video:thumbnail_loc>https://i.ytimg.com/vi/mm9IHqgCbZc/maxresdefault.jpg</video:thumbnail_loc>
      <video:title>Off-Grid Networking: Starlink and Ubiquiti Wireless Bridges</video:title>
      <video:description>Building a robust network in an &apos;internet desert&apos; requires creative engineering. In this video, we detail how we established a high-speed connection using Starlink and a series of Ubiquiti NanoStation 5AC Locos. We walk through the setup of a wireless bridge to network buildings separated by over 100 meters, ensuring reliable connectivity across the entire property. This guide covers the technical specifications and hardware choices necessary for anyone looking to implement advanced networking solutions in rural or off-grid locations.</video:description>
      <video:player_loc allow_embed="yes">https://www.youtube.com/embed/mm9IHqgCbZc</video:player_loc>
      <video:publication_date>2024-02-07T19:22:33.000Z</video:publication_date>
      <video:family_friendly>yes</video:family_friendly>
      <video:live>no</video:live>
    </video:video>
  </url>
</urlset>
