<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>In-Context Learning — FoundationIndex</title>
    <link>https://foundationindex.ustunozgur.com/s/in-context-learning</link>
    <description>The ability of a foundation model to pick up a new task, format, or reasoning style solely from examples and instructions inside the prompt—without changing any of its saved weights.</description>
    <language>en</language>
    <atom:link href="https://foundationindex.ustunozgur.com/s/in-context-learning/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>In-Context Learning · Paper · 2026-02-13</title>
      <link>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-doc-to-lora-2026</link>
      <guid>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-doc-to-lora-2026</guid>
      <pubDate>Fri, 13 Feb 2026 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Charakorn et al. at Sakana AI introduced Doc-to-LoRA, training a hypernetwork to convert unseen long prompts into LoRA adapters in a single forward pass—internalizing context without KV-cache memory overhead and maintaining zero-shot recall at over 4× a model&amp;apos;s native context window.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2602.15902&quot;&gt;Doc-to-LoRA: Learning to Instantly Internalize Contexts&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>In-Context Learning · Paper · 2025-12-31</title>
      <link>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-nested-learning-2025</link>
      <guid>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-nested-learning-2025</guid>
      <pubDate>Wed, 31 Dec 2025 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Behrouz et al. at Google Research published Nested Learning: The Illusion of Deep Learning Architectures, unifying neural architectures and gradient optimizers as nested multi-frequency context-compression problems in which in-context learning emerges naturally and motivates self-modifying continual memory modules (Hope).&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2512.24695&quot;&gt;Nested Learning: The Illusion of Deep Learning Architectures&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>In-Context Learning · Paper · 2024-04-17</title>
      <link>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-many-shot-2024</link>
      <guid>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-many-shot-2024</guid>
      <pubDate>Wed, 17 Apr 2024 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Agarwal et al. at Google DeepMind published Many-Shot In-Context Learning, showing that expanding prompts from dozens to hundreds or thousands of demonstrations inside long context windows (such as Gemini 1.5 Pro) yields steep log-linear gains that rival or surpass supervised fine-tuning on non-natural and reasoning tasks.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2404.11018&quot;&gt;Many-Shot In-Context Learning&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>In-Context Learning · Paper · 2022-09-24</title>
      <link>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-induction-heads-2022</link>
      <guid>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-induction-heads-2022</guid>
      <pubDate>Sat, 24 Sep 2022 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Olsson et al. at Anthropic published In-context Learning and Induction Heads, linking the sharp emergence of in-context learning during pre-training to two-layer Attention circuits that match earlier token pairs ([A][B] ... [A] -&amp;gt; [B]) and generalize to abstract pattern completion.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2209.11895&quot;&gt;In-context Learning and Induction Heads&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>In-Context Learning · Paper · 2022-01-28</title>
      <link>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-chain-of-thought-2022</link>
      <guid>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-chain-of-thought-2022</guid>
      <pubDate>Fri, 28 Jan 2022 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Wei et al. published Chain-of-Thought Prompting Elicits Reasoning in Large Language Models, showing that including intermediate step-by-step derivations inside few-shot prompts unlocks multi-step arithmetic, commonsense, and symbolic reasoning at scale—laying the prompting foundation for later test-time reasoning models.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2201.11903&quot;&gt;Chain-of-Thought Prompting Elicits Reasoning in Large Language Models&lt;/a&gt;&lt;/p&gt;</description>
    </item>
    <item>
      <title>In-Context Learning · Paper · 2020-05-28</title>
      <link>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-gpt3-few-shot-2020</link>
      <guid>https://foundationindex.ustunozgur.com/s/in-context-learning#icl-gpt3-few-shot-2020</guid>
      <pubDate>Thu, 28 May 2020 00:00:00 GMT</pubDate>
      <description>&lt;p&gt;Brown et al. at OpenAI published Language Models are Few-Shot Learners, demonstrating that scaling a 175B-parameter autoregressive Transformer (GPT-3) unlocks zero-, one-, and few-shot task adaptation directly from prompt demonstrations without updating model weights.&lt;/p&gt;&lt;p&gt;Sources: &lt;a href=&quot;https://arxiv.org/abs/2005.14165&quot;&gt;Language Models are Few-Shot Learners&lt;/a&gt;&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
