<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Kevin Lu</title><description>Kevin Lu — co-founder of Sweep AI (YC S23), previously an ML engineer at Roblox. I write about large language models, fine-tuning, inference, and developer tools.</description><link>https://yuhai.lu/</link><item><title>Building sub-100ms autocompletion for JetBrains IDEs</title><link>https://yuhai.lu/blog/next-edit-jetbrains/</link><guid isPermaLink="true">https://yuhai.lu/blog/next-edit-jetbrains/</guid><description>We&apos;re building the best autocomplete for JetBrains</description><pubDate>Mon, 15 Sep 2025 00:00:00 GMT</pubDate></item><item><title>Open sourcing a 1.5B parameter Next-Edit Autocomplete Model</title><link>https://yuhai.lu/blog/oss-next-edit/</link><guid isPermaLink="true">https://yuhai.lu/blog/oss-next-edit/</guid><description>We&apos;re open sourcing Sweep Next-Edit, a locally runnable SOTA LLM for next-edit autocompletion.</description><pubDate>Tue, 13 Jan 2026 00:00:00 GMT</pubDate></item><item><title>Open sourcing Sweep Next-Edit v2 and our inference runtime</title><link>https://yuhai.lu/blog/sweep-next-edit-v2/</link><guid isPermaLink="true">https://yuhai.lu/blog/sweep-next-edit-v2/</guid><description>We&apos;ve open sourced our new 7B next-edit autocomplete model and the TensorRT-LLM runtime we use to serve it.</description><pubDate>Sun, 21 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Teaching LLMs to spell with token healing</title><link>https://yuhai.lu/blog/token-healing-autocomplete/</link><guid isPermaLink="true">https://yuhai.lu/blog/token-healing-autocomplete/</guid><description>Using token healing to make our autocomplete spell correctly in under 1ms.</description><pubDate>Mon, 22 Sep 2025 00:00:00 GMT</pubDate></item></channel></rss>