<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>PA\RMAN</title><description>Scatered notis sette ich doun.”</description><link>https://www.pnxlr.eu.org/</link><item><title>FastestDetV2: a Light-weight Object Detector</title><link>https://www.pnxlr.eu.org/posts/fastestdetv2-a-light-weight-object-detector/</link><guid isPermaLink="true">https://www.pnxlr.eu.org/posts/fastestdetv2-a-light-weight-object-detector/</guid><description>Even faster and stronger than FastestDet.</description><pubDate>Tue, 25 Aug 2026 00:00:00 GMT</pubDate></item><item><title>Linear Attention Algorithms: LA, RetNet, Mamba2, and Gated DeltaNet</title><link>https://www.pnxlr.eu.org/posts/linear-attention-algorithms-la-retnet-mamba2-and-gated-deltanet/</link><guid isPermaLink="true">https://www.pnxlr.eu.org/posts/linear-attention-algorithms-la-retnet-mamba2-and-gated-deltanet/</guid><description>A review of linear attention methods, covering LA, RetNet, Mamba-2, and Gated DeltaNet.</description><pubDate>Fri, 14 Aug 2026 00:00:00 GMT</pubDate></item><item><title>Understanding Prefill, Decode, and the KV Cache in LLMs</title><link>https://www.pnxlr.eu.org/posts/understanding-prefill-decode-and-the-kv-cache-in-llms/</link><guid isPermaLink="true">https://www.pnxlr.eu.org/posts/understanding-prefill-decode-and-the-kv-cache-in-llms/</guid><description>Hands-on example to understand the generation pipeline of an LLM.</description><pubDate>Fri, 14 Aug 2026 00:00:00 GMT</pubDate></item><item><title>Rediscover the Structure of an LLM</title><link>https://www.pnxlr.eu.org/posts/rediscover-the-structure-of-an-llm/</link><guid isPermaLink="true">https://www.pnxlr.eu.org/posts/rediscover-the-structure-of-an-llm/</guid><description>A clearer way to understand the structure of a decoder-only LLM.</description><pubDate>Thu, 13 Aug 2026 00:00:00 GMT</pubDate></item></channel></rss>