<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: lossolo</title><link>https://news.ycombinator.com/user?id=lossolo</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 21 Sep 2026 13:12:27 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=lossolo" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by lossolo in "The Hugging Face Hack Wasn't What It Was Cracked Up to Be"]]></title><description><![CDATA[
<p>After a recent interview[1] with Noam Brown (OpenAI), in which he said they had specifically trained agents for cooperation before this hack, the hack doesn't seem as impressive anymore.<p>They didn't even bother to control the post training rollouts, so the training data got contaminated and was included in the training of other agents. Connect these two dots and you have the Hugging Face hack. And at the beginning, when these incidents were first reported, it was portrayed as if all of this (the communication between agents etc.) was emergent behaviour.<p>1. <a href="https://www.youtube.com/watch?v=6AgOfiZOWiY" rel="nofollow">https://www.youtube.com/watch?v=6AgOfiZOWiY</a></p>
]]></description><pubDate>Sun, 20 Sep 2026 05:40:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=49772779</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49772779</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49772779</guid></item><item><title><![CDATA[New comment by lossolo in "US confirms for first time it has deployed space weapons"]]></title><description><![CDATA[
<p>"The European Space Agency estimates that there have been more than 660 explosions, collisions, or anomalous events resulting in fragmentation. As it stands, there are currently more than 1.5 million space debris objects in Earth orbit.<p>It’s not clear what may have caused the Yaogan-50 (02) satellite to break apart, but it may have been a propulsion or battery failure, or it may have itself collided with another object or space debris."</p>
]]></description><pubDate>Tue, 15 Sep 2026 19:51:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=49717953</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49717953</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49717953</guid></item><item><title><![CDATA[New comment by lossolo in "US confirms for first time it has deployed space weapons"]]></title><description><![CDATA[
<p>China is not Iraq, Afghanistan, Venezuela or some other third world country that the US has fought in over the past few decades. You can't just destroy one of their satellites for free. And I can bet the US didn't do it, because that would be a major escalation, especially now that Xi is coming to the US and there are rumors that the US wants a deal.</p>
]]></description><pubDate>Tue, 15 Sep 2026 17:25:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=49715765</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49715765</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49715765</guid></item><item><title><![CDATA[New comment by lossolo in "On the Navier–Stokes Millennium Prize Problem"]]></title><description><![CDATA[
<p>Yeah, I'm just curious about the setting. It's just weird to me that this wasn't disclosed by either party while the accusations were being made, that's all.
Even if it was used in the training data, I don't believe it had that much of an impact myself, since the solutions are quite different.</p>
]]></description><pubDate>Tue, 08 Sep 2026 23:12:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=49618457</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49618457</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49618457</guid></item><item><title><![CDATA[New comment by lossolo in "On the Navier–Stokes Millennium Prize Problem"]]></title><description><![CDATA[
<p>> If they opted out of training, then we definitely did not train on them.<p>Can't you guys just check their account settings so the public knows what was set?<p>EDIT: Why was this downvoted? I'm genuinely asking because I have no idea. Opting out is just a normal setting in the profile, It's not like I'm asking for their private conversations or PII. 
If I were the person claiming that they trained on my conversations, I'd make sure to disclose that I had opted out and hadn't given them permission to do so. And if I were the accused party, I'd disclose whether that setting was turned on or off to provide evidence against the accusation.</p>
]]></description><pubDate>Tue, 08 Sep 2026 21:23:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=49617277</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49617277</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49617277</guid></item><item><title><![CDATA[New comment by lossolo in "On the Navier–Stokes Millennium Prize Problem"]]></title><description><![CDATA[
<p>Not entirely, it's just a late stage of the overall training process. It's an early checkpoint in post training (you can use the model at different stages of training), so it will probably become even stronger with more post training.</p>
]]></description><pubDate>Tue, 08 Sep 2026 20:57:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=49616917</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49616917</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49616917</guid></item><item><title><![CDATA[New comment by lossolo in "GPT-6 Astra"]]></title><description><![CDATA[
<p>Yeah, basically they are using more computation to explore the solution space before producing the final answer.</p>
]]></description><pubDate>Thu, 03 Sep 2026 20:17:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=49556249</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49556249</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49556249</guid></item><item><title><![CDATA[New comment by lossolo in "Ask HN: Why were OpenAI, Claude, and Grok simultaneously down?"]]></title><description><![CDATA[
<p>It was working for me too.<p>sample_size++;</p>
]]></description><pubDate>Thu, 03 Sep 2026 17:40:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=49553754</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49553754</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49553754</guid></item><item><title><![CDATA[New comment by lossolo in "Io_uring Without Readahead"]]></title><description><![CDATA[
<p>> a larger read is generally as fast as multiple smaller one on modern hardware.<p>Not always if by modern you mean NVMe drives.
One synchronous preadv() for 256 KiB gives the kernel/device one big request but 16 independent asynchronous 16 KiB reads can be serviced concurrently.
So the latter gives the NVMe controller 16 operations it can schedule in parallel.
So depending on the workload and hardware, offsets, filesystem and request sizes that can give you lower aggregate latency or higher throughput.</p>
]]></description><pubDate>Tue, 01 Sep 2026 16:33:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=49524311</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49524311</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49524311</guid></item><item><title><![CDATA[New comment by lossolo in "Google Has Removed MV2 Extensions from the Chrome Web Store, Including UBO"]]></title><description><![CDATA[
<p>A few days ago, my mother asked me to fix her YouTube because she kept getting something telling her to update. I looked at YouTube on her iPhone and saw an autoplaying video ad that displayed an animated fake system style pop-up button, designed to look like an important iPhone OS security update. The ad even showed an animation of the button appearing on the screen, making it look even more like a real system notification.<p>This needs to stop.</p>
]]></description><pubDate>Mon, 31 Aug 2026 23:48:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=49516218</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49516218</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49516218</guid></item><item><title><![CDATA[New comment by lossolo in "Disruption with Some GitHub Services"]]></title><description><![CDATA[
<p>How can you do so badly with Git when its architecture is basically so friendly to partitioning and horizontal scaling?</p>
]]></description><pubDate>Wed, 26 Aug 2026 17:16:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=49452598</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49452598</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49452598</guid></item><item><title><![CDATA[New comment by lossolo in "I were 17, I'd learn how to build LLMs from scratch"]]></title><description><![CDATA[
<p>It's like saying that horse can understand physics, because he knows when to jump using his intuition.</p>
]]></description><pubDate>Mon, 24 Aug 2026 23:24:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=49427199</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49427199</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49427199</guid></item><item><title><![CDATA[New comment by lossolo in "Fable and the end of the free lunch"]]></title><description><![CDATA[
<p>>  At some point chatting with Fable inevitably leads to it thinking about the security related aspects, tripping the safeguards.<p>It happens to me all the time with things that have nothing to do with security, Fable spawns a subagent that then adversarially checks the code Fable just wrote and hits guardrails, with zero prompting from me.</p>
]]></description><pubDate>Sun, 23 Aug 2026 21:38:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=49412891</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49412891</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49412891</guid></item><item><title><![CDATA[New comment by lossolo in "Canada will match US tariffs 'dollar for dollar' as trade talks break down"]]></title><description><![CDATA[
<p>> Gross debt to GDP ratio in the US has fallen from 130% in 2021 to under 125% now.<p>Most of that was inflation, so it hurts U.S. citizens. I mean, you could reduce it even further pretty easily, but I'm not sure you would want to pay $1,000+ for a Big Mac.</p>
]]></description><pubDate>Sat, 22 Aug 2026 16:28:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=49401292</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49401292</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49401292</guid></item><item><title><![CDATA["Two 2030 AMD racks are expected to deliver same compute as 570 racks in 2024"]]></title><description><![CDATA[
<p>Article URL: <a href="https://newsroom.amd.com/news/amd-tracks-ahead-of-rack-scale-ai-energy-efficiency-goal/">https://newsroom.amd.com/news/amd-tracks-ahead-of-rack-scale-ai-energy-efficiency-goal/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49366670">https://news.ycombinator.com/item?id=49366670</a></p>
<p>Points: 12</p>
<p># Comments: 3</p>
]]></description><pubDate>Wed, 19 Aug 2026 20:20:19 +0000</pubDate><link>https://newsroom.amd.com/news/amd-tracks-ahead-of-rack-scale-ai-energy-efficiency-goal/</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49366670</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49366670</guid></item><item><title><![CDATA[New comment by lossolo in "US announces new sanctions on top ICC figures"]]></title><description><![CDATA[
<p>Khan (ICC prosecutor) said Lindsey Graham (US senator) told him during a May 2024 call:<p>"This court is for Africa and thugs like Putin. It is not for democracies like Israel and the United States of America."</p>
]]></description><pubDate>Tue, 18 Aug 2026 21:10:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=49352759</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49352759</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49352759</guid></item><item><title><![CDATA[New comment by lossolo in "Claude Code May–August 2026 weekly limits promotion"]]></title><description><![CDATA[
<p>If they remove that extra 50%, I'm moving my $200 to Codex. Good thing we will know tomorrow, just a day before renewal.<p>"Your subscription will auto renew on Aug 20, 2026."<p>But I doubt they will remove it, just like they didn't remove Fable from the subscription. There's simply too much competition.</p>
]]></description><pubDate>Tue, 18 Aug 2026 19:20:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=49351162</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49351162</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49351162</guid></item><item><title><![CDATA[New comment by lossolo in "Qwen 3.8 27B"]]></title><description><![CDATA[
<p>Why weren't the points merged again from the "dupe" thread that had 289 points?<p><a href="https://news.ycombinator.com/item?id=49299684">https://news.ycombinator.com/item?id=49299684</a><p>What a weird mechanism. If someone is judging a thread/topic/event impact by the number of points it got, then doing this unfairly degrades that thread.<p>It should have deduped by user and combined the 168(at the time of writing this comment) + 289 points. Just add the twitter link from the previous thread as an additional link in the description, like you normally do, move all the points over, and remove the old thread.</p>
]]></description><pubDate>Fri, 14 Aug 2026 17:09:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=49301629</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49301629</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49301629</guid></item><item><title><![CDATA[New comment by lossolo in "GLM-5.3: Frontier coding with emergent cyber capabilities"]]></title><description><![CDATA[
<p>> but this is still just GLM 5.2 with post-training magic.<p>So exactly the same as Opus 5 and GPT 5.6 Sol. It's all "post-training magic".</p>
]]></description><pubDate>Fri, 14 Aug 2026 14:43:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=49299372</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49299372</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49299372</guid></item><item><title><![CDATA[New comment by lossolo in "Grok 4.6"]]></title><description><![CDATA[
<p>This is basically the answer, they generate A LOT of synthetic task rollouts in parallel, then use RL on the resulting reward signals to improve the model. Add scale to this and you have a Fable class model.</p>
]]></description><pubDate>Wed, 12 Aug 2026 17:20:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49275765</link><dc:creator>lossolo</dc:creator><comments>https://news.ycombinator.com/item?id=49275765</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49275765</guid></item></channel></rss>