<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Phi — FoundationIndex</title>
    <link>https://foundationindex.ustunozgur.com/s/phi</link>
    <description>Microsoft Research&apos;s family of open-weight Small Language Models (SLMs) that proved compact 1.3B–15B parameter models trained on curated &apos;textbook-quality&apos; synthetic data can rival models an order of magnitude larger.</description>
    <language>en</language>
    <atom:link href="https://foundationindex.ustunozgur.com/s/phi/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Phi · Release · 2025-04-30</title>
      <link>https://foundationindex.ustunozgur.com/s/phi#phi-4-reasoning-suite-2025</link>
      <guid>https://foundationindex.ustunozgur.com/s/phi#phi-4-reasoning-suite-2025</guid>
      <pubDate>Wed, 30 Apr 2025 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Microsoft open-weighted Phi-4-reasoning and Phi-4-reasoning-plus (14B) alongside Phi-4-mini-reasoning (3.8B), pairing supervised reasoning traces with reinforcement learning for compact test-time compute scaling.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2504.21318&quot;&gt;Phi-4-reasoning Technical Report&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>Phi · Release · 2025-03-03</title>
      <link>https://foundationindex.ustunozgur.com/s/phi#phi-4-multimodal-mini-2025</link>
      <guid>https://foundationindex.ustunozgur.com/s/phi#phi-4-multimodal-mini-2025</guid>
      <pubDate>Mon, 03 Mar 2025 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Microsoft released Phi-4-Multimodal (5.6B) and Phi-4-Mini (3.8B), introducing a Mixture-of-LoRAs architecture that attaches modality-specific LoRA adapters for vision and speech/audio onto a frozen text backbone without degrading base language accuracy.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2503.01743&quot;&gt;Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>Phi · Release · 2024-12-12</title>
      <link>https://foundationindex.ustunozgur.com/s/phi#phi-4-14b-2024</link>
      <guid>https://foundationindex.ustunozgur.com/s/phi#phi-4-14b-2024</guid>
      <pubDate>Thu, 12 Dec 2024 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Microsoft Research released Phi-4 (14B) under an MIT license, introducing 50+ multi-agent Synthetic Data &amp;amp; Self-Play generation pipelines alongside pivotal-token DPO (which weights preference optimization on the specific tokens that flip a reasoning chain from incorrect to correct).&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2412.08905&quot;&gt;Phi-4 Technical Report&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>Phi · Release · 2024-04-23</title>
      <link>https://foundationindex.ustunozgur.com/s/phi#phi-3-family-2024</link>
      <guid>https://foundationindex.ustunozgur.com/s/phi#phi-3-family-2024</guid>
      <pubDate>Tue, 23 Apr 2024 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Microsoft launched the open-weight Phi-3 series—starting with Phi-3-mini (3.8B, trained on 3.3T tokens in 4K and 128K context variants) followed by Small (7B), Medium (14B), and the August 2024 Phi-3.5 MoE and Vision checkpoints—designed for 4-bit Quantization on phones and laptops.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2404.14219&quot;&gt;Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>Phi · Release · 2023-12-12</title>
      <link>https://foundationindex.ustunozgur.com/s/phi#phi-2-27b-2023</link>
      <guid>https://foundationindex.ustunozgur.com/s/phi#phi-2-27b-2023</guid>
      <pubDate>Tue, 12 Dec 2023 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Microsoft Research released Phi-2 (2.7B parameters), scaling its synthetic textbook and educational web data recipe to 1.4 trillion tokens and matching or surpassing 13B–70B open models of late 2023 on common-sense and math benchmarks.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://www.microsoft.com/en-us/research/blog/phi-2-the-surprising-power-of-small-language-models/&quot;&gt;Phi-2: The surprising power of small language models&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>Phi · Paper · 2023-06-20</title>
      <link>https://foundationindex.ustunozgur.com/s/phi#phi-1-textbooks-2023</link>
      <guid>https://foundationindex.ustunozgur.com/s/phi#phi-1-textbooks-2023</guid>
      <pubDate>Tue, 20 Jun 2023 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Microsoft Research published Textbooks Are All You Need, training the 1.3B-parameter Phi-1 on ~7B tokens of filtered code and synthetically generated Python textbooks and exercises to outperform much larger models on HumanEval.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2306.11644&quot;&gt;Textbooks Are All You Need&lt;/a&gt;&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
