Hacker News: martianvoidhttps://news.ycombinator.com/user?id=martianvoidHacker News RSShttps://hnrss.org/hnrss v2.1.1Tue, 28 Jul 2026 18:19:43 +0000<![CDATA[New comment by martianvoid in "ARC-AGI Leaderboard"]]>It's actually crazy to see the difference between opus 5 and the next best model on ARC AGI 3 when you actually look at the ARC AGI problems

]]>Sat, 25 Jul 2026 06:59:42 +0000https://news.ycombinator.com/item?id=49045169martianvoidhttps://news.ycombinator.com/item?id=49045169https://news.ycombinator.com/item?id=49045169<![CDATA[NeurIPS Mech Interp Workshop is getting more AI slop]]>Article URL: https://twitter.com/NeelNanda5/status/2077117795066871908

Comments URL: https://news.ycombinator.com/item?id=48918463

Points: 3

# Comments: 0

]]>Wed, 15 Jul 2026 09:55:22 +0000https://twitter.com/NeelNanda5/status/2077117795066871908martianvoidhttps://news.ycombinator.com/item?id=48918463https://news.ycombinator.com/item?id=48918463<![CDATA[New comment by martianvoid in "Harness Engineering for Self-Improvement"]]>I am very much skeptical on the joint optimization of the harness and the model weights, I mean we can argue that claude code itself was a joint optimization of the claude, they released the first claude model and then the claude code harness then improved claudes performance on the harness then improved the harness after the model improved, but this was a process managed by humans all the time, it would be interesting to see claude improving it's own weights or harness in the future (or it might already be happening at anthropic right now)

I just didn't find the SIA paper to be testing this out rigorously or either provide sufficient evidence that it works

]]>Tue, 07 Jul 2026 08:09:34 +0000https://news.ycombinator.com/item?id=48814931martianvoidhttps://news.ycombinator.com/item?id=48814931https://news.ycombinator.com/item?id=48814931<![CDATA[You can never replace your understanding]]>Article URL: https://www.anthropic.com/research/claude-code-expertise

Comments URL: https://news.ycombinator.com/item?id=48682270

Points: 4

# Comments: 0

]]>Fri, 26 Jun 2026 04:19:12 +0000https://www.anthropic.com/research/claude-code-expertisemartianvoidhttps://news.ycombinator.com/item?id=48682270https://news.ycombinator.com/item?id=48682270<![CDATA[The Art of War]]>Article URL: https://en.wikipedia.org/wiki/The_Art_of_War

Comments URL: https://news.ycombinator.com/item?id=48682223

Points: 3

# Comments: 0

]]>Fri, 26 Jun 2026 04:09:05 +0000https://en.wikipedia.org/wiki/The_Art_of_Warmartianvoidhttps://news.ycombinator.com/item?id=48682223https://news.ycombinator.com/item?id=48682223<![CDATA[Show HN: Multi Agent Protocol for AI Scientist by Hexo Labs]]>Article URL: https://github.com/hexo-ai/socrates

Comments URL: https://news.ycombinator.com/item?id=48671952

Points: 5

# Comments: 0

]]>Thu, 25 Jun 2026 11:35:51 +0000https://github.com/hexo-ai/socratesmartianvoidhttps://news.ycombinator.com/item?id=48671952https://news.ycombinator.com/item?id=48671952<![CDATA[The Process Was the Point]]>Article URL: https://darshanmakwana412.github.io/2026/06/the-process-was-the-point/

Comments URL: https://news.ycombinator.com/item?id=48447754

Points: 2

# Comments: 0

]]>Mon, 08 Jun 2026 16:47:43 +0000https://darshanmakwana412.github.io/2026/06/the-process-was-the-point/martianvoidhttps://news.ycombinator.com/item?id=48447754https://news.ycombinator.com/item?id=48447754<![CDATA[AI Agent that at inference time updates it's harness and model weights]]>Article URL: https://github.com/hexo-ai/sia

Comments URL: https://news.ycombinator.com/item?id=48344807

Points: 5

# Comments: 0

]]>Sun, 31 May 2026 11:13:43 +0000https://github.com/hexo-ai/siamartianvoidhttps://news.ycombinator.com/item?id=48344807https://news.ycombinator.com/item?id=48344807<![CDATA[Tsplat – Run Gaussian splatting in your terminal]]>Article URL: https://github.com/darshanmakwana412/tsplat

Comments URL: https://news.ycombinator.com/item?id=48307851

Points: 37

# Comments: 8

]]>Thu, 28 May 2026 12:15:55 +0000https://github.com/darshanmakwana412/tsplatmartianvoidhttps://news.ycombinator.com/item?id=48307851https://news.ycombinator.com/item?id=48307851<![CDATA[Gaussian Splatting for Dummies]]>Article URL: https://darshanmakwana412.github.io/2026/04/gaussian-splatting/

Comments URL: https://news.ycombinator.com/item?id=48196371

Points: 1

# Comments: 0

]]>Tue, 19 May 2026 17:28:26 +0000https://darshanmakwana412.github.io/2026/04/gaussian-splatting/martianvoidhttps://news.ycombinator.com/item?id=48196371https://news.ycombinator.com/item?id=48196371<![CDATA[Gaussian Splatting for Dummies]]>Article URL: https://darshanmakwana412.github.io/2026/04/gaussian-splatting/

Comments URL: https://news.ycombinator.com/item?id=48166599

Points: 2

# Comments: 0

]]>Sun, 17 May 2026 06:48:43 +0000https://darshanmakwana412.github.io/2026/04/gaussian-splatting/martianvoidhttps://news.ycombinator.com/item?id=48166599https://news.ycombinator.com/item?id=48166599<![CDATA[New comment by martianvoid in "Show HN: Stage CLI – An easier way of reading your AI generated changes locally"]]>This is just git with extra steps...

]]>Thu, 07 May 2026 21:03:35 +0000https://news.ycombinator.com/item?id=48054973martianvoidhttps://news.ycombinator.com/item?id=48054973https://news.ycombinator.com/item?id=48054973<![CDATA[New comment by martianvoid in "LAWS: A new transform operation turning LLM inference into cheap cache lookups"]]>Is there any codebase associated with this that we can check and tinker with? or is this just a promo thing?

]]>Thu, 07 May 2026 06:18:16 +0000https://news.ycombinator.com/item?id=48046055martianvoidhttps://news.ycombinator.com/item?id=48046055https://news.ycombinator.com/item?id=48046055