<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: ketothekingdom</title><link>https://news.ycombinator.com/user?id=ketothekingdom</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 19 Oct 2026 03:58:04 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=ketothekingdom" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by ketothekingdom in "You could have invented transformers from memory"]]></title><description><![CDATA[
<p>Definitely! It's more of an intuition but doesn't explain what vector gets put in the weights and the analysis doesn't involve the learning dynamics at all. I link some papers at the end that do think about how memory relates to the dataset. But the architecture having this storing+lookup property probably is important in helping transformers remember so much information.</p>
]]></description><pubDate>Sat, 17 Oct 2026 17:04:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=50132862</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=50132862</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50132862</guid></item><item><title><![CDATA[New comment by ketothekingdom in "You could have invented transformers from memory"]]></title><description><![CDATA[
<p>Thank you! It's been sitting in my head for so long. I'm glad I finally got to share it with people</p>
]]></description><pubDate>Sat, 17 Oct 2026 03:36:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=50127675</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=50127675</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50127675</guid></item><item><title><![CDATA[New comment by ketothekingdom in "You could have invented transformers from memory"]]></title><description><![CDATA[
<p>That's such a neat idea! Now I want to try it out too</p>
]]></description><pubDate>Sat, 17 Oct 2026 03:35:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=50127671</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=50127671</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50127671</guid></item><item><title><![CDATA[New comment by ketothekingdom in "You could have invented transformers from memory"]]></title><description><![CDATA[
<p>Hey author here! I have a short summary here if people find it helpful: <a href="https://x.com/heyyalexwang/status/2111158818730397932?s=46" rel="nofollow">https://x.com/heyyalexwang/status/2111158818730397932?s=46</a></p>
]]></description><pubDate>Sat, 17 Oct 2026 00:46:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=50126675</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=50126675</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50126675</guid></item><item><title><![CDATA[Put a price on breakthroughs]]></title><description><![CDATA[
<p>Article URL: <a href="https://alexwang.ai/posts/put-a-price-on-breakthroughs/">https://alexwang.ai/posts/put-a-price-on-breakthroughs/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50028982">https://news.ycombinator.com/item?id=50028982</a></p>
<p>Points: 36</p>
<p># Comments: 10</p>
]]></description><pubDate>Sat, 10 Oct 2026 02:37:47 +0000</pubDate><link>https://alexwang.ai/posts/put-a-price-on-breakthroughs/</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=50028982</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50028982</guid></item><item><title><![CDATA[The Model Is the Company]]></title><description><![CDATA[
<p>Article URL: <a href="https://alexwang.ai/posts/the-model-is-the-company/">https://alexwang.ai/posts/the-model-is-the-company/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47365133">https://news.ycombinator.com/item?id=47365133</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Fri, 13 Mar 2026 14:40:35 +0000</pubDate><link>https://alexwang.ai/posts/the-model-is-the-company/</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=47365133</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47365133</guid></item><item><title><![CDATA[New comment by ketothekingdom in "Neural audio codecs: how to get audio into LLMs"]]></title><description><![CDATA[
<p>Cartesia is doing constant time models for audio: <a href="https://cartesia.ai/" rel="nofollow">https://cartesia.ai/</a></p>
]]></description><pubDate>Wed, 22 Oct 2025 01:31:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=45663963</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=45663963</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45663963</guid></item><item><title><![CDATA[Test-time regression: a unifying framework for designing sequence models]]></title><description><![CDATA[
<p>Article URL: <a href="https://arxiv.org/abs/2501.12352">https://arxiv.org/abs/2501.12352</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=42794658">https://news.ycombinator.com/item?id=42794658</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 22 Jan 2025 16:41:00 +0000</pubDate><link>https://arxiv.org/abs/2501.12352</link><dc:creator>ketothekingdom</dc:creator><comments>https://news.ycombinator.com/item?id=42794658</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42794658</guid></item></channel></rss>