https://muradbaghirli.com/tags/nvidia-triton/Recent content in NVIDIA Triton on Murad BaghirliHugo -- gohugo.ioen© 2026 Murad BaghirliSat, 31 Jan 2026 00:00:00 +0000https://muradbaghirli.com/projects/sentence-boundary-detection/Sat, 31 Jan 2026 00:00:00 +0000https://muradbaghirli.com/projects/sentence-boundary-detection/<p><em>Status: <strong>Complete</strong> · Nov 2025 – Jan 2026 · <a href="https://github.com/muradbaghirli/SentenceBoundaryDetection" target="_blank" rel="noreferrer">GitHub</a></em></p> <style> .dark .dark-img { display: block; } .dark .light-img { display: none; } .dark-img { display: none; } .light-img { display: block; } </style> <h2 class="relative group">TL;DR <div id="tldr" class="anchor"></div> <span class="absolute top-0 w-6 transition-opacity opacity-0 -start-6 not-prose group-hover:opacity-100 select-none"> <a class="text-primary-300 dark:text-neutral-700 !no-underline" href="#tldr" aria-label="Anchor">#</a> </span> </h2> <p>I built a sentence boundary detector specialized for <strong>Azerbaijani legal text</strong>, not general-purpose prose. It is a random forest classifier that decides, for every candidate terminator character (<code>.</code>, <code>!</code>, <code>?</code>, <code>;</code>, quotes, newlines…), whether a sentence actually ends there. Each candidate is described by a 19-character context window plus 9 engineered linguistic and punctuation features. The model reaches <strong>0.958 F1</strong> on boundary detection.</p>