The latest articles on DEV Community by DezF (@lthreec). https://dev.to/lthreec https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3834174%2Fb31ae486-d6dd-4317-9a71-87ffe73b001b.png https://dev.to/lthreec en DezF Thu, 19 Mar 2026 19:13:35 +0000 https://dev.to/lthreec/pyruns-a-local-first-web-ui-for-running-and-organizing-python-experiments-2l44 https://dev.to/lthreec/pyruns-a-local-first-web-ui-for-running-and-organizing-python-experiments-2l44 <p>If you often do things like:</p> <ul> <li>tweak a few parameters and rerun the same script over and over</li> <li>keep shell commands in terminal history and lose track of what actually worked</li> <li>run multiple experiments at once and end up with mixed, unreadable logs</li> <li>want lightweight experiment management without pushing your workflow into a heavyweight cloud platform</li> </ul> <p>then <strong>Pyruns</strong> might be worth a look.</p> <h2> What it is </h2> <p><strong>Pyruns</strong> is a local-first experiment workspace for Python scripts and shell tasks.</p> <p>It is built around a simple idea:</p> <blockquote> <p>experiment tooling should stay close to the way people already work</p> </blockquote> <p>That means:</p> <ul> <li>keep using your own scripts</li> <li>keep using your own terminal and conda environment</li> <li>keep everything local</li> <li>keep task configs, logs, notes, and run history on disk in a predictable workspace</li> </ul> <h2> What it does </h2> <ul> <li>visual parameter editing for Python script workflows</li> <li>batch task generation in form mode</li> <li>YAML-based single-task editing</li> <li>shell task management with host-terminal semantics</li> <li>task search, filtering, rerun, delete, pin, and notes</li> <li>terminal-style live logs in the browser</li> <li>CSV export for recorded metrics</li> <li>one isolated config/log directory per task</li> </ul> <h2> The main entrypoints </h2> <p>The primary path is script-first:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight shell"><code>pyr train.py pyr train.py config.yaml </code></pre> </div> <p>This is the recommended first experience for most users.</p> <p>There is also a shell-first path:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight shell"><code>pyr </code></pre> </div> <p>That opens a shell workspace for the current directory.<br> It is useful when what you want to manage is a set of commands rather than a single Python script.</p> <h2> Why it feels different </h2> <p>Pyruns is not trying to replace your workflow with a platform.</p> <p>It feels more like a practical local workbench for people who already have:</p> <ul> <li>Python scripts</li> <li>YAML configs</li> <li>shell commands</li> <li>local logs</li> <li>repeated experiment runs</li> </ul> <p>and simply want those pieces to become easier to generate, run, inspect, and revisit.</p> <h2> Current stack </h2> <ul> <li>React frontend</li> <li>FastAPI backend</li> <li>Home / Generator / Manager / Monitor UI flow</li> </ul> <h2> Where it fits well </h2> <ul> <li> <code>argparse</code>-based training scripts</li> <li> <code>pyruns.load()</code> / YAML-driven workflows</li> <li>local ML or automation experiments</li> <li>shell-heavy command workflows</li> <li>users who want local visibility without cloud lock-in</li> </ul> <h2> Links </h2> <ul> <li>GitHub: <a href="https://github.com/LthreeC/pyruns" rel="noopener noreferrer">https://github.com/LthreeC/pyruns</a> </li> <li>Docs: <a href="https://lthreec.github.io/pyruns/" rel="noopener noreferrer">https://lthreec.github.io/pyruns/</a> </li> </ul> <h2> Install </h2> <div class="highlight js-code-highlight"> <pre class="highlight shell"><code>pip <span class="nb">install </span>pyruns </code></pre> </div> <h2> Start </h2> <div class="highlight js-code-highlight"> <pre class="highlight shell"><code>pyr train.py </code></pre> </div> programming python machinelearning webdev