<?xml version="1.0" encoding="UTF-8"?><urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:news="http://www.google.com/schemas/sitemap-news/0.9" xmlns:xhtml="http://www.w3.org/1999/xhtml" xmlns:image="http://www.google.com/schemas/sitemap-image/1.1" xmlns:video="http://www.google.com/schemas/sitemap-video/1.1"><url><loc>https://rfriedmann.de/</loc></url><url><loc>https://rfriedmann.de/blog/ai-that-isnt-an-llm/</loc></url><url><loc>https://rfriedmann.de/blog/axo-learning-without-backprop/</loc></url><url><loc>https://rfriedmann.de/blog/bumping-a-dependency-you-cant-read/</loc></url><url><loc>https://rfriedmann.de/blog/continuous-batching/</loc></url><url><loc>https://rfriedmann.de/blog/gguf-decode-beats-llama-cpp/</loc></url><url><loc>https://rfriedmann.de/blog/how-a-big-model-fits-on-one-card/</loc></url><url><loc>https://rfriedmann.de/blog/how-a-model-learns/</loc></url><url><loc>https://rfriedmann.de/blog/how-an-agent-uses-tools-and-memory/</loc></url><url><loc>https://rfriedmann.de/blog/how-imp-works-model-to-token/</loc></url><url><loc>https://rfriedmann.de/blog/how-models-see-images/</loc></url><url><loc>https://rfriedmann.de/blog/how-to-prompt/</loc></url><url><loc>https://rfriedmann.de/blog/kv-cache-and-long-context/</loc></url><url><loc>https://rfriedmann.de/blog/letting-an-agent-loose/</loc></url><url><loc>https://rfriedmann.de/blog/llms-in-the-terminal/</loc></url><url><loc>https://rfriedmann.de/blog/llms-in-your-infrastructure/</loc></url><url><loc>https://rfriedmann.de/blog/mixture-of-experts-explained/</loc></url><url><loc>https://rfriedmann.de/blog/nvfp4-at-the-bit-level/</loc></url><url><loc>https://rfriedmann.de/blog/online-softmax-and-flashattention/</loc></url><url><loc>https://rfriedmann.de/blog/optimizing-kernels-consumer-blackwell/</loc></url><url><loc>https://rfriedmann.de/blog/prefill-vs-decode-roofline/</loc></url><url><loc>https://rfriedmann.de/blog/pretraining-finetuning-rlhf/</loc></url><url><loc>https://rfriedmann.de/blog/prompt-injection/</loc></url><url><loc>https://rfriedmann.de/blog/reading-model-specs/</loc></url><url><loc>https://rfriedmann.de/blog/self-hosting-an-llm-for-your-team/</loc></url><url><loc>https://rfriedmann.de/blog/serving-30b-models-rtx-5090/</loc></url><url><loc>https://rfriedmann.de/blog/speculative-decoding/</loc></url><url><loc>https://rfriedmann.de/blog/the-context-window/</loc></url><url><loc>https://rfriedmann.de/blog/what-is-a-gpu/</loc></url><url><loc>https://rfriedmann.de/blog/what-is-an-ai-agent/</loc></url><url><loc>https://rfriedmann.de/blog/what-is-an-llm/</loc></url><url><loc>https://rfriedmann.de/blog/what-the-5090-lacks-vs-datacenter/</loc></url><url><loc>https://rfriedmann.de/blog/when-not-to-use-ai/</loc></url><url><loc>https://rfriedmann.de/blog/when-the-model-is-not-a-transformer/</loc></url><url><loc>https://rfriedmann.de/blog/why-different-answers/</loc></url><url><loc>https://rfriedmann.de/blog/why-models-cant-spell/</loc></url><url><loc>https://rfriedmann.de/blog/why-models-make-things-up/</loc></url><url><loc>https://rfriedmann.de/blog/why-run-ai-locally/</loc></url><url><loc>https://rfriedmann.de/blog/writing-cuda-with-an-agent/</loc></url><url><loc>https://rfriedmann.de/blog/your-own-documents/</loc></url><url><loc>https://rfriedmann.de/learn/</loc></url><url><loc>https://rfriedmann.de/log/</loc></url><url><loc>https://rfriedmann.de/news/</loc></url></urlset>