https://elgun.si/blog/Guides and research notes on building and evaluating AI systems.enhttps://elgun.si/blog/fine-tuning-a-small-language-model-with-lora/https://elgun.si/blog/fine-tuning-a-small-language-model-with-lora/A practical starting point for supervised fine-tuning: define the task, prepare data, train an adapter, and evaluate the result.Fri, 25 Sep 2026 00:00:00 GMThttps://elgun.si/blog/a-leakage-safe-evaluation-checklist/https://elgun.si/blog/a-leakage-safe-evaluation-checklist/How to keep information from the test set out of preprocessing, model selection, and reported results.Thu, 24 Sep 2026 00:00:00 GMThttps://elgun.si/blog/when-a-model-meets-a-new-dataset/https://elgun.si/blog/when-a-model-meets-a-new-dataset/A practical way to think about external validation, distribution shift, and the limits of internal scores.Wed, 23 Sep 2026 00:00:00 GMThttps://elgun.si/blog/what-trustworthy-mental-health-ai-requires/https://elgun.si/blog/what-trustworthy-mental-health-ai-requires/Reliability, causal sensitivity, and interpretation matter alongside predictive performance.Thu, 20 Aug 2026 00:00:00 GMT