<?xml version="1.0" encoding="UTF-8"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Fine-tuning &amp; Training — The Latent Daily</title><description>Fine-tuning, distillation and training runs.</description><link>https://thelatentdaily.com/</link><language>en-us</language><atom:link href="https://thelatentdaily.com/rss/fine-tuning-training.xml" rel="self" type="application/rss+xml"/><item><title>Build, buy or fine-tune? How companies choose their AI approach</title><link>https://thelatentdaily.com/articles/build-buy-or-fine-tune-enterprise-ai/</link><guid isPermaLink="true">https://thelatentdaily.com/articles/build-buy-or-fine-tune-enterprise-ai/</guid><description>Off-the-shelf assistants, API integrations, retrieval-augmented generation, fine-tuning or your own model: the trade-offs in cost, control and time to value.</description><pubDate>Sun, 18 Oct 2026 05:53:00 GMT</pubDate><category>Business &amp; Enterprise</category><category>enterprise</category><category>rag</category><category>fine-tuning</category><category>strategy</category><author>dario_admin@resolvia-crm.tech</author></item><item><title>Fine-tuning an LLM with LoRA: a practical overview</title><link>https://thelatentdaily.com/guides/fine-tuning-llm-with-lora-practical-overview/</link><guid isPermaLink="true">https://thelatentdaily.com/guides/fine-tuning-llm-with-lora-practical-overview/</guid><description>When fine-tuning helps, how LoRA and QLoRA make it affordable, how to prepare data, train, evaluate and deploy an adapter — and when not to fine-tune at all.</description><pubDate>Sun, 18 Oct 2026 04:49:00 GMT</pubDate><category>Fine-tuning &amp; Training</category><category>fine-tuning</category><category>lora</category><category>qlora</category><category>open-weights</category><author>dario_admin@resolvia-crm.tech</author></item><item><title>Knowledge distillation explained: how small AI models learn from big ones</title><link>https://thelatentdaily.com/articles/knowledge-distillation-explained/</link><guid isPermaLink="true">https://thelatentdaily.com/articles/knowledge-distillation-explained/</guid><description>Distillation trains a compact &quot;student&quot; model to imitate a large &quot;teacher&quot;. How it works, why it powers today&apos;s small models, and its legal and quality limits.</description><pubDate>Sun, 18 Oct 2026 04:48:00 GMT</pubDate><category>Fine-tuning &amp; Training</category><category>distillation</category><category>small-models</category><category>training</category><category>explainer</category><author>dario_admin@resolvia-crm.tech</author></item><item><title>Paper of the day: Direct Preference Optimization (DPO), explained</title><link>https://thelatentdaily.com/research/direct-preference-optimization-dpo-paper-explained/</link><guid isPermaLink="true">https://thelatentdaily.com/research/direct-preference-optimization-dpo-paper-explained/</guid><description>DPO showed language models can be aligned to human preferences without a separate reward model or reinforcement learning. How it works and why it spread so fast.</description><pubDate>Sun, 18 Oct 2026 02:54:00 GMT</pubDate><category>Paper of the Day</category><category>research</category><category>dpo</category><category>alignment</category><category>rlhf</category><author>dario_admin@resolvia-crm.tech</author></item></channel></rss>