<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: runeblaze</title><link>https://news.ycombinator.com/user?id=runeblaze</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 06 Apr 2026 03:38:37 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=runeblaze" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by runeblaze in "OpenAI raises $110B on $730B pre-money valuation"]]></title><description><![CDATA[
<p>1. openrouter is API usage. There is obviously consumer side<p>2. people often use openrouter for the sole purpose of using a unified chat completions API<p>3. OpenAI invented chat completions; if you use openrouter for chat completions often you can just switch your endpoint URL to point to the OAI endpoint to avoid the openrouter surcharge!<p>4. Hence anyone with large enough volume will very likely <i>not</i> use openrouter for OpenAI; there is an active incentive to take the easy route of changing the endpoint URL to OAI’s</p>
]]></description><pubDate>Fri, 27 Feb 2026 17:21:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=47182980</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=47182980</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47182980</guid></item><item><title><![CDATA[New comment by runeblaze in "LLM Structured Outputs Handbook"]]></title><description><![CDATA[
<p>Schemas can get pretty complex (and LLMs might not be the best at counting). Also schemas are sometimes the first way to guard against the stochasticity of LLMs.<p>With that said, the model is pretty good at it.</p>
]]></description><pubDate>Sat, 17 Jan 2026 06:12:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=46655767</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=46655767</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46655767</guid></item><item><title><![CDATA[New comment by runeblaze in "“Erdos problem #728 was solved more or less autonomously by AI”"]]></title><description><![CDATA[
<p>Resouce-affording, if you are chasing the frontier of some more niche task you redo your training regime on the new-gen LLMs</p>
]]></description><pubDate>Sun, 11 Jan 2026 17:35:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=46577700</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=46577700</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46577700</guid></item><item><title><![CDATA[New comment by runeblaze in "“Erdos problem #728 was solved more or less autonomously by AI”"]]></title><description><![CDATA[
<p>Is it though? There is a reason gpt has codex variants. RL on a specific task raises the performance on that task</p>
]]></description><pubDate>Sun, 11 Jan 2026 02:09:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=46572065</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=46572065</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46572065</guid></item><item><title><![CDATA[New comment by runeblaze in "Tesla sales fell by 9 percent in 2025, its second yearly decline"]]></title><description><![CDATA[
<p>Sure never again is totally fair and I am sure a lot of people hate it. I was mostly objecting to the radioactivity of it. Your friends will be more like “I am looking to sell my Tesla in 3 months” if it is truly radioactive.<p>Let’s be realistic in our portrayal here.</p>
]]></description><pubDate>Sat, 03 Jan 2026 01:09:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=46471662</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=46471662</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46471662</guid></item><item><title><![CDATA[New comment by runeblaze in "Tesla sales fell by 9 percent in 2025, its second yearly decline"]]></title><description><![CDATA[
<p>I think radioactive is a strong word here… I have talked to a lot of people in tech</p>
]]></description><pubDate>Fri, 02 Jan 2026 22:05:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=46469992</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=46469992</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46469992</guid></item><item><title><![CDATA[New comment by runeblaze in "Court report detailing ChatGPT's involvement with a recent murder suicide [pdf]"]]></title><description><![CDATA[
<p>Reading what you wrote scares me</p>
]]></description><pubDate>Wed, 31 Dec 2025 19:42:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=46447564</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=46447564</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46447564</guid></item><item><title><![CDATA[New comment by runeblaze in "Critical vulnerability in LangChain – CVE-2025-68664"]]></title><description><![CDATA[
<p>> And if for some ungodly reason you had to do it in Python<p>I literally invoke sglang and vllm in Python. You are supposed to (if not using them over-the-network) use the two fastest inference engines there is via Python.</p>
]]></description><pubDate>Fri, 26 Dec 2025 09:58:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=46390739</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=46390739</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46390739</guid></item><item><title><![CDATA[New comment by runeblaze in "Yann LeCun to depart Meta and launch AI startup focused on 'world models'"]]></title><description><![CDATA[
<p>Agreed, I am surprised he is happy to stay this long. He would have been on paper a far better match at a place like pre-Gemini-era Google</p>
]]></description><pubDate>Wed, 12 Nov 2025 12:58:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=45899588</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45899588</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45899588</guid></item><item><title><![CDATA[New comment by runeblaze in "Using Generative AI in Content Production"]]></title><description><![CDATA[
<p>I don't know the data distribution, but are you sure that's generated by an Adobe model? I can only see that it is in Stock + it is tagged as AI generated (that is, was that image generated by some other model?)<p>Disclaimer: I used to work at Adobe GenAI. Opinions are of my own ofc.</p>
]]></description><pubDate>Tue, 11 Nov 2025 09:23:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=45885519</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45885519</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45885519</guid></item><item><title><![CDATA[New comment by runeblaze in "Using Generative AI in Content Production"]]></title><description><![CDATA[
<p>Emmmm sure, but throw this to a human artist who has not heard of Indiana Jones and see if they draw something alike.</p>
]]></description><pubDate>Tue, 11 Nov 2025 09:17:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=45885470</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45885470</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45885470</guid></item><item><title><![CDATA[New comment by runeblaze in "Using Generative AI in Content Production"]]></title><description><![CDATA[
<p>I work in this space. In traditional diffusion-based regimes (paired image and text), one can absolutely check the text to remove all occurrences of Indiana Jones. Likewise, Adobe Stock has content moderation that ensures (up to human moderation limit) no dirty content. It is a world without Indiana Jones to the model</p>
]]></description><pubDate>Mon, 10 Nov 2025 23:20:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=45882246</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45882246</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45882246</guid></item><item><title><![CDATA[New comment by runeblaze in "At the end you use `git bisect`"]]></title><description><![CDATA[
<p>My personal mantra (that I myself cannot uphold 100%) is that every dev should at least do the exercise of implementing binary search from scratch in a language with arbitrary-precision integers (e.g., Python) once in a while. It is the best exercise in invariant-based thinking, useful for software correctness at large</p>
]]></description><pubDate>Mon, 03 Nov 2025 08:26:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=45796970</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45796970</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45796970</guid></item><item><title><![CDATA[New comment by runeblaze in "DeepSeek OCR"]]></title><description><![CDATA[
<p>each text token is often subword unit, but in VLMs the visual tokens are in semantic space. Semantic space obviously compresses much more than subword slices.<p>disclaimer: not expert, on top of my head</p>
]]></description><pubDate>Mon, 20 Oct 2025 08:20:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=45641225</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45641225</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45641225</guid></item><item><title><![CDATA[New comment by runeblaze in "Monads are too powerful: The expressiveness spectrum"]]></title><description><![CDATA[
<p>beats me. I spent so much time learning what a fundamental group is and I still cannot tell ppl what a fundamental group is convincingly.<p>I can’t even make stuff with fundamental groups.</p>
]]></description><pubDate>Thu, 16 Oct 2025 08:01:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=45602662</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45602662</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45602662</guid></item><item><title><![CDATA[New comment by runeblaze in "Memory access is O(N^[1/3])"]]></title><description><![CDATA[
<p>Ummm guys when we talk about memory access in theory can we just be rigorous and talk about the computational model?<p>The real RAM model “in theory” tells me that memory access is O(1). Of course real RAM is a spherical cow but like we could use a bit more rigor</p>
]]></description><pubDate>Wed, 08 Oct 2025 22:24:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=45521294</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45521294</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45521294</guid></item><item><title><![CDATA[New comment by runeblaze in "I built ChatGPT with Minecraft redstone [video]"]]></title><description><![CDATA[
<p>Of course it seems to be the same person to have done minecraft inside minecraft using redstone :)). peak!! done it again</p>
]]></description><pubDate>Thu, 02 Oct 2025 09:10:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=45447634</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45447634</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45447634</guid></item><item><title><![CDATA[New comment by runeblaze in "There is a huge pool of exceptional junior engineers"]]></title><description><![CDATA[
<p>Yea experience is very useful (it is very hard to exert influence without power and be the tech lead of a team unless you have some corporate experience) but at some point one needs to realize that if someone was writing Haskell at the age of 12 and they got decent scores at USACO they would make quite a decent engineer...</p>
]]></description><pubDate>Tue, 30 Sep 2025 20:02:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=45430519</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45430519</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45430519</guid></item><item><title><![CDATA[New comment by runeblaze in "Python developers are embracing type hints"]]></title><description><![CDATA[
<p>Python types - all the onus of static types, with none of the power of calculus of constructions. /s</p>
]]></description><pubDate>Mon, 29 Sep 2025 05:44:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=45410652</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45410652</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45410652</guid></item><item><title><![CDATA[New comment by runeblaze in "Identity Types"]]></title><description><![CDATA[
<p>sure of course one does not need types to do natural deduction… if you throw what you showed me to your average maths undergrad in the US they will get confused — I truly don’t see how proofs are simple</p>
]]></description><pubDate>Sun, 28 Sep 2025 23:40:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=45409051</link><dc:creator>runeblaze</dc:creator><comments>https://news.ycombinator.com/item?id=45409051</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45409051</guid></item></channel></rss>