<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[Topics tagged with kimi]]></title><description><![CDATA[A list of topics that have been tagged with kimi]]></description><link>https://board.circlewithadot.net/tags/kimi</link><generator>RSS for Node</generator><lastBuildDate>Fri, 15 May 2026 00:08:48 GMT</lastBuildDate><atom:link href="https://board.circlewithadot.net/tags/kimi.rss" rel="self" type="application/rss+xml"/><pubDate>Invalid Date</pubDate><ttl>60</ttl><item><title><![CDATA[RT @Hesamation: DeepSeek-V4 nutzt den Muon-Optimizer mit Kimis Rezept, um ihn für das Training großer Sprachmodelle zu skalieren.]]></title><description><![CDATA[RT @Hesamation: DeepSeek-V4 nutzt den Muon-Optimizer mit Kimis Rezept, um ihn für das Training großer Sprachmodelle zu skalieren. In der Zwischenzeit verwendet Kimi K2 (und K2.6) die architektonischen Techniken von DeepSeek-V3 (ultrasparse MoE + MLA). Open-Source-KI-Labore bauen auf der Forschung der jeweils anderen auf, und das ist genau so, wie es sein sollte.
mehr auf Arint.info
#DeepSeek #KI #Kimi #LLM #MachineLearning #OpenSource #arint_info
https://x.com/Hesamation/status/2047681313226854838#m]]></description><link>https://board.circlewithadot.net/topic/8b4940e6-952a-4d32-b836-0ba6a2b0022b/rt-@hesamation-deepseek-v4-nutzt-den-muon-optimizer-mit-kimis-rezept-um-ihn-für-das-training-großer-sprachmodelle-zu-skalieren.</link><guid isPermaLink="true">https://board.circlewithadot.net/topic/8b4940e6-952a-4d32-b836-0ba6a2b0022b/rt-@hesamation-deepseek-v4-nutzt-den-muon-optimizer-mit-kimis-rezept-um-ihn-für-das-training-großer-sprachmodelle-zu-skalieren.</guid><dc:creator><![CDATA[arint@arint.info]]></dc:creator><pubDate>Invalid Date</pubDate></item><item><title><![CDATA[Is the #LLM race actually a race to the bottom?]]></title><description><![CDATA[@perpetuum_mobile Since Gemma4 came out, I agree it's been the gold standard for performance vs compute. If SoC is the way forward for local compute (and I think its clear it is) the real jump happens when unified memory architectures can actually handle the token volume an agentic harness needs. Progress on memory overhead for long-context agents, combined with advancements in unified pool architecture, make this a real possibility in the near future.]]></description><link>https://board.circlewithadot.net/topic/210c762e-8499-4a0d-bd84-9809fb2b171c/is-the-llm-race-actually-a-race-to-the-bottom</link><guid isPermaLink="true">https://board.circlewithadot.net/topic/210c762e-8499-4a0d-bd84-9809fb2b171c/is-the-llm-race-actually-a-race-to-the-bottom</guid><dc:creator><![CDATA[mamba@mstdn.ca]]></dc:creator><pubDate>Invalid Date</pubDate></item></channel></rss>