<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: limoce</title><link>https://news.ycombinator.com/user?id=limoce</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sun, 24 May 2026 22:18:25 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=limoce" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[China Blocks Meta's $2B Acquisition of AI Firm Manus]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.bloomberg.com/news/articles/2026-04-27/china-blocks-meta-s-2-billion-acquisition-of-ai-startup-manus">https://www.bloomberg.com/news/articles/2026-04-27/china-blocks-meta-s-2-billion-acquisition-of-ai-startup-manus</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47919231">https://news.ycombinator.com/item?id=47919231</a></p>
<p>Points: 19</p>
<p># Comments: 9</p>
]]></description><pubDate>Mon, 27 Apr 2026 08:54:20 +0000</pubDate><link>https://www.bloomberg.com/news/articles/2026-04-27/china-blocks-meta-s-2-billion-acquisition-of-ai-startup-manus</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47919231</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47919231</guid></item><item><title><![CDATA[Claude Code Source Map]]></title><description><![CDATA[
<p>Article URL: <a href="https://github.com/ChinaSiro/claude-code-sourcemap">https://github.com/ChinaSiro/claude-code-sourcemap</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47585850">https://news.ycombinator.com/item?id=47585850</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 31 Mar 2026 11:36:10 +0000</pubDate><link>https://github.com/ChinaSiro/claude-code-sourcemap</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47585850</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47585850</guid></item><item><title><![CDATA[NeurIPS retracts overly broad sanctions list]]></title><description><![CDATA[
<p>Article URL: <a href="https://twitter.com/NeurIPSConf/status/2037438893457289364">https://twitter.com/NeurIPSConf/status/2037438893457289364</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47540475">https://news.ycombinator.com/item?id=47540475</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Fri, 27 Mar 2026 09:08:45 +0000</pubDate><link>https://twitter.com/NeurIPSConf/status/2037438893457289364</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47540475</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47540475</guid></item><item><title><![CDATA[New comment by limoce in "$500 GPU outperforms Claude Sonnet on coding benchmarks"]]></title><description><![CDATA[
<p>The title should be "Adaptive Test-time Learning and Autonomous Specialization".</p>
]]></description><pubDate>Fri, 27 Mar 2026 01:47:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=47538189</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47538189</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47538189</guid></item><item><title><![CDATA[NeurIPS Statement on Sanctions]]></title><description><![CDATA[
<p>Article URL: <a href="https://twitter.com/NeurIPSConf/status/2037066494983426374">https://twitter.com/NeurIPSConf/status/2037066494983426374</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47528093">https://news.ycombinator.com/item?id=47528093</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Thu, 26 Mar 2026 08:57:22 +0000</pubDate><link>https://twitter.com/NeurIPSConf/status/2037066494983426374</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47528093</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47528093</guid></item><item><title><![CDATA[Declining to review for NeurIPS due to policy on sanctioned institutions]]></title><description><![CDATA[
<p>Article URL: <a href="https://pastebin.com/LEmwfMZT">https://pastebin.com/LEmwfMZT</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47527496">https://news.ycombinator.com/item?id=47527496</a></p>
<p>Points: 5</p>
<p># Comments: 0</p>
]]></description><pubDate>Thu, 26 Mar 2026 07:14:43 +0000</pubDate><link>https://pastebin.com/LEmwfMZT</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47527496</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47527496</guid></item><item><title><![CDATA[PetClaw: A Desktop Pet Companion That Works 24/7]]></title><description><![CDATA[
<p>Article URL: <a href="https://petclaw.ai/">https://petclaw.ai/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47501321">https://news.ycombinator.com/item?id=47501321</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 24 Mar 2026 11:52:56 +0000</pubDate><link>https://petclaw.ai/</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47501321</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47501321</guid></item><item><title><![CDATA[Classical Chinese Jailbreak Prompt Optimization via Bio-Inspired Search]]></title><description><![CDATA[
<p>Article URL: <a href="https://arxiv.org/abs/2602.22983">https://arxiv.org/abs/2602.22983</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47487292">https://news.ycombinator.com/item?id=47487292</a></p>
<p>Points: 3</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 23 Mar 2026 09:55:19 +0000</pubDate><link>https://arxiv.org/abs/2602.22983</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47487292</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47487292</guid></item><item><title><![CDATA[Cursor's 'Composer 2' model is apparently just Kimi K2.5 with RL fine-tuning]]></title><description><![CDATA[
<p>Article URL: <a href="https://old.reddit.com/r/singularity/comments/1ryrs2w/cursors_composer_2_model_is_apparently_just_kimi/">https://old.reddit.com/r/singularity/comments/1ryrs2w/cursors_composer_2_model_is_apparently_just_kimi/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47452903">https://news.ycombinator.com/item?id=47452903</a></p>
<p>Points: 1</p>
<p># Comments: 1</p>
]]></description><pubDate>Fri, 20 Mar 2026 10:57:06 +0000</pubDate><link>https://old.reddit.com/r/singularity/comments/1ryrs2w/cursors_composer_2_model_is_apparently_just_kimi/</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47452903</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47452903</guid></item><item><title><![CDATA[Add NPU support for AICore for Gemma4 model]]></title><description><![CDATA[
<p>Article URL: <a href="https://github.com/google-ai-edge/LiteRT-LM/pull/1573">https://github.com/google-ai-edge/LiteRT-LM/pull/1573</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47308640">https://news.ycombinator.com/item?id=47308640</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 09 Mar 2026 13:14:45 +0000</pubDate><link>https://github.com/google-ai-edge/LiteRT-LM/pull/1573</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47308640</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47308640</guid></item><item><title><![CDATA[SteptronOss: Lightweight, AI-native training framework for large language models]]></title><description><![CDATA[
<p>Article URL: <a href="https://github.com/stepfun-ai/SteptronOss">https://github.com/stepfun-ai/SteptronOss</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47231390">https://news.ycombinator.com/item?id=47231390</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Tue, 03 Mar 2026 12:30:21 +0000</pubDate><link>https://github.com/stepfun-ai/SteptronOss</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=47231390</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47231390</guid></item><item><title><![CDATA[New comment by limoce in "GLM-5 was trained entirely on Huawei chips"]]></title><description><![CDATA[
<p>> This site is an independent informational resource and is not officially affiliated with Zhipu AI.</p>
]]></description><pubDate>Thu, 12 Feb 2026 07:00:04 +0000</pubDate><link>https://news.ycombinator.com/item?id=46985649</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46985649</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46985649</guid></item><item><title><![CDATA[New comment by limoce in "Ship Types, Not Docs"]]></title><description><![CDATA[
<p>What about using LLMs to automatically update docs?</p>
]]></description><pubDate>Wed, 11 Feb 2026 02:11:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=46969914</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46969914</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46969914</guid></item><item><title><![CDATA[Adding Support for Qwen3.5]]></title><description><![CDATA[
<p>Article URL: <a href="https://github.com/huggingface/transformers/pull/43830">https://github.com/huggingface/transformers/pull/43830</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46942237">https://news.ycombinator.com/item?id=46942237</a></p>
<p>Points: 5</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 09 Feb 2026 06:38:24 +0000</pubDate><link>https://github.com/huggingface/transformers/pull/43830</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46942237</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46942237</guid></item><item><title><![CDATA[Reduce Complexity of Matmul to O(N^2logN) by Asymptotically Optimal Quantum Algo]]></title><description><![CDATA[
<p>Article URL: <a href="https://arxiv.org/abs/2602.05541">https://arxiv.org/abs/2602.05541</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46913950">https://news.ycombinator.com/item?id=46913950</a></p>
<p>Points: 4</p>
<p># Comments: 1</p>
]]></description><pubDate>Fri, 06 Feb 2026 15:24:04 +0000</pubDate><link>https://arxiv.org/abs/2602.05541</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46913950</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46913950</guid></item><item><title><![CDATA[Learning from context is harder than we thought]]></title><description><![CDATA[
<p>Article URL: <a href="https://hy.tencent.com/research/100025?langVersion=en">https://hy.tencent.com/research/100025?langVersion=en</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46870514">https://news.ycombinator.com/item?id=46870514</a></p>
<p>Points: 245</p>
<p># Comments: 136</p>
]]></description><pubDate>Tue, 03 Feb 2026 13:07:37 +0000</pubDate><link>https://hy.tencent.com/research/100025?langVersion=en</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46870514</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46870514</guid></item><item><title><![CDATA[Residual Context Diffusion]]></title><description><![CDATA[
<p>Article URL: <a href="https://yuezhouhu.github.io/projects/residual-context-diffusion/index.html">https://yuezhouhu.github.io/projects/residual-context-diffusion/index.html</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46853078">https://news.ycombinator.com/item?id=46853078</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 02 Feb 2026 06:30:25 +0000</pubDate><link>https://yuezhouhu.github.io/projects/residual-context-diffusion/index.html</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46853078</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46853078</guid></item><item><title><![CDATA[Step 3.5 Flash]]></title><description><![CDATA[
<p>Article URL: <a href="https://static.stepfun.com/blog/step-3.5-flash/">https://static.stepfun.com/blog/step-3.5-flash/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46851806">https://news.ycombinator.com/item?id=46851806</a></p>
<p>Points: 7</p>
<p># Comments: 1</p>
]]></description><pubDate>Mon, 02 Feb 2026 02:42:32 +0000</pubDate><link>https://static.stepfun.com/blog/step-3.5-flash/</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46851806</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46851806</guid></item><item><title><![CDATA[LongCat-Flash-Lite 100B A3B Technical Report [pdf]]]></title><description><![CDATA[
<p>Article URL: <a href="https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/blob/main/tech_report.pdf">https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/blob/main/tech_report.pdf</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46809216">https://news.ycombinator.com/item?id=46809216</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Thu, 29 Jan 2026 12:23:17 +0000</pubDate><link>https://huggingface.co/meituan-longcat/LongCat-Flash-Lite/blob/main/tech_report.pdf</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46809216</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46809216</guid></item><item><title><![CDATA[New comment by limoce in "Which programming languages are most token-efficient?"]]></title><description><![CDATA[
<p>I think separating thinking tokens from "representing" tokens might be a better approach, like what those thinking models does</p>
]]></description><pubDate>Mon, 12 Jan 2026 03:13:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=46583593</link><dc:creator>limoce</dc:creator><comments>https://news.ycombinator.com/item?id=46583593</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46583593</guid></item></channel></rss>