<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Vigneshwararaj Sathianarayanan — Journal</title><link>https://vigneshwararaj.raidder.xyz/</link><description>Build notes on my projects.</description><item><title>Web search without an API key</title><link>https://vigneshwararaj.raidder.xyz/journal/2026-09-22-web-search-without-an-api-key/</link><guid>https://vigneshwararaj.raidder.xyz/journal/2026-09-22-web-search-without-an-api-key/</guid><pubDate>Tue, 22 Sep 2026 00:00:00 GMT</pubDate><description>Bing's API is gone and Brave's free tier ended, so my assistant now searches through a SearXNG instance running in Docker on my own machine.</description></item><item><title>Teaching a 3B model to call tools</title><link>https://vigneshwararaj.raidder.xyz/journal/2026-09-21-tool-calling-phi-vs-qwen/</link><guid>https://vigneshwararaj.raidder.xyz/journal/2026-09-21-tool-calling-phi-vs-qwen/</guid><pubDate>Mon, 21 Sep 2026 00:00:00 GMT</pubDate><description>Phi-3.5 has no built-in tool calling, so I had to fake it with prompts. Qwen2.5-3B has it natively, and the difference was 0 out of 6 against 6 out of 6.</description></item><item><title>A chat model in 4 GB of VRAM</title><link>https://vigneshwararaj.raidder.xyz/journal/2026-08-29-a-local-model-on-4gb/</link><guid>https://vigneshwararaj.raidder.xyz/journal/2026-08-29-a-local-model-on-4gb/</guid><pubDate>Sat, 29 Aug 2026 00:00:00 GMT</pubDate><description>Step one of my local assistant. A 3.8B model, 4-bit quantisation, and a terminal chat loop that streams its answer.</description></item></channel></rss>