<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Automatone - AI &amp; Dev Tools Blog</title>
    <description>Practical guides on AI tools, developer workflows, and automation.</description>
    <link>https://automatone.win</link>
    <atom:link href="https://automatone.win/feed.xml" rel="self" type="application/rss+xml" />
    <language>en</language>
    <lastBuildDate>Wed, 17 Jun 2026 08:59:15 GMT</lastBuildDate>
    <item>
      <title><![CDATA[uv vs pip vs Poetry: Python Package Managers in 2026]]></title>
      <description><![CDATA[uv, pip, and Poetry each solve a different problem in 2026. Here is the benchmark-backed case for defaulting to uv on new projects, keeping pip as the universal baseline, and reaching for Poetry when you publish libraries — plus a skimmable decision guide and migration notes.]]></description>
      <link>https://automatone.win/blog/uv-vs-pip-vs-poetry-2026</link>
      <guid isPermaLink="true">https://automatone.win/blog/uv-vs-pip-vs-poetry-2026</guid>
      <pubDate>Wed, 17 Jun 2026 04:02:26 GMT</pubDate>
      <category>comparisons</category>
    </item>
    <item>
      <title><![CDATA[How to Self-Host n8n with Docker Compose on a VPS]]></title>
      <description><![CDATA[A 2026 production guide to self-hosting n8n on a VPS with Docker Compose: official images, PostgreSQL, external task runners, automatic HTTPS via Caddy, and a backup routine that protects your encryption key. Updated for the n8n 2.0 hardening release and the March 2026 RCE disclosures.]]></description>
      <link>https://automatone.win/blog/self-host-n8n-docker-compose-vps</link>
      <guid isPermaLink="true">https://automatone.win/blog/self-host-n8n-docker-compose-vps</guid>
      <pubDate>Wed, 17 Jun 2026 04:01:06 GMT</pubDate>
      <category>guides</category>
    </item>
    <item>
      <title><![CDATA[Fix "CUDA out of memory" Errors in PyTorch]]></title>
      <description><![CDATA[CUDA out of memory in PyTorch often fires while nvidia-smi still shows free VRAM, because the caching allocator needs a contiguous block, not just free space. This guide gives a copy-paste triage path from fastest fixes to fragmentation tuning, plus how to tell a real leak apart from fragmentation.]]></description>
      <link>https://automatone.win/blog/fix-cuda-out-of-memory-pytorch</link>
      <guid isPermaLink="true">https://automatone.win/blog/fix-cuda-out-of-memory-pytorch</guid>
      <pubDate>Wed, 17 Jun 2026 03:58:43 GMT</pubDate>
      <category>troubleshooting</category>
    </item>
    <item>
      <title><![CDATA[How Speculative Decoding Speeds Up LLM Inference]]></title>
      <description><![CDATA[Speculative decoding makes LLM inference 2-3x faster by letting a small draft model guess ahead and a large model verify the guesses in one parallel pass. A rejection-sampling step keeps the output mathematically identical to the slow path. Here's how it works, why it's lossless, and where it stops helping.]]></description>
      <link>https://automatone.win/blog/how-speculative-decoding-speeds-up-llm-inference</link>
      <guid isPermaLink="true">https://automatone.win/blog/how-speculative-decoding-speeds-up-llm-inference</guid>
      <pubDate>Wed, 17 Jun 2026 03:56:52 GMT</pubDate>
      <category>ai-ml</category>
    </item>
  </channel>
</rss>