<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Microsoft on English AI Terms Dictionary</title><link>https://terms-en.ai-term-hub.com/en/tags/microsoft/</link><description>Recent content in Microsoft on English AI Terms Dictionary</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Sat, 18 Jul 2026 11:44:44 +0000</lastBuildDate><atom:link href="https://terms-en.ai-term-hub.com/en/tags/microsoft/index.xml" rel="self" type="application/rss+xml"/><item><title>Phi3</title><link>https://terms-en.ai-term-hub.com/en/terms/phi3/</link><pubDate>Sat, 18 Jul 2026 10:10:59 +0000</pubDate><guid>https://terms-en.ai-term-hub.com/en/terms/phi3/</guid><description>&lt;h2 id="definition">Definition&lt;/h2>
&lt;p>Phi-3 is a series of small language models (SLMs) released by Microsoft, designed to deliver high performance comparable to larger models while requiring significantly less computational resources. These models are trained on high-quality, filtered synthetic data and real text, focusing on reasoning, mathematics, and coding capabilities. Phi-3 supports various context lengths and is optimized for deployment on edge devices, making it suitable for on-device AI applications without relying on heavy cloud infrastructure.&lt;/p></description></item><item><title>Phi</title><link>https://terms-en.ai-term-hub.com/en/terms/phi/</link><pubDate>Sat, 18 Jul 2026 10:10:45 +0000</pubDate><guid>https://terms-en.ai-term-hub.com/en/terms/phi/</guid><description>&lt;h2 id="definition">Definition&lt;/h2>
&lt;p>Phi, short for &amp;lsquo;Foundation models based on Teaching-Learning Paradigm&amp;rsquo;, is a family of compact large language models created by Microsoft. Unlike traditional LLMs trained on massive web corpora, Phi models are trained on high-quality synthetic text derived from trusted sources. They demonstrate exceptional reasoning capabilities relative to their size, making them suitable for resource-constrained environments and applications requiring precise logical inference without the overhead of larger models.&lt;/p>
&lt;h3 id="summary">Summary&lt;/h3>
&lt;p>A series of small but highly efficient large language models developed by Microsoft Research, focusing on knowledge density and reasoning.&lt;/p></description></item><item><title>Dataset:Ms Marco</title><link>https://terms-en.ai-term-hub.com/en/terms/datasetms_marco/</link><pubDate>Sat, 18 Jul 2026 09:53:44 +0000</pubDate><guid>https://terms-en.ai-term-hub.com/en/terms/datasetms_marco/</guid><description>&lt;h2 id="definition">Definition&lt;/h2>
&lt;p>MS MARCO (Microsoft Machine Reading Comprehension) is a widely used dataset in natural language processing, particularly for information retrieval and question answering. It consists of anonymized search queries from Bing and corresponding relevant passages from web documents. Researchers use it to train models to rank documents based on relevance to a query or to extract direct answers, serving as a foundational benchmark for modern dense retrieval and passage ranking models.&lt;/p></description></item></channel></rss>