<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: numba888</title><link>https://news.ycombinator.com/user?id=numba888</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Tue, 07 Apr 2026 23:58:17 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=numba888" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by numba888 in "Jensen Huang – Nvidia GTC 2025 Keynote"]]></title><description><![CDATA[
<p>well, "RTX Pro Blackwell" desktop and workstation version looks to be the answer for midrange. Some should be available starting in April.</p>
]]></description><pubDate>Wed, 19 Mar 2025 03:44:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=43408108</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43408108</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43408108</guid></item><item><title><![CDATA[New comment by numba888 in "Nvidia DGX Spark (formerly DIGITS) available for reservation"]]></title><description><![CDATA[
<p>Edge inference most likely. Its FP4 performance is about 1/3 of 5090, power 170W for the whole thing. It can run big model or several small. Shifting balance to memory favors MoE. Would be nice to see FP32 numbers, they are used in training. My guess about 20 TFLOP, may be more, but 5090 is still times better.</p>
]]></description><pubDate>Wed, 19 Mar 2025 03:35:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=43408064</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43408064</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43408064</guid></item><item><title><![CDATA[New comment by numba888 in "Jensen Huang – Nvidia GTC 2025 Keynote"]]></title><description><![CDATA[
<p>> glad to hear about DGX Station<p>how much is it expected, my guess barely fits in 5 digits. Would be nice to have something in between Spark and Station. I.e. some desktop withing $20K.</p>
]]></description><pubDate>Tue, 18 Mar 2025 22:54:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=43406167</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43406167</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43406167</guid></item><item><title><![CDATA[New comment by numba888 in "Jensen Huang – Nvidia GTC 2025 Keynote"]]></title><description><![CDATA[
<p>what's the better alternative, DIY AMD? No, thanks. I'm interested in the result, not the process. In terms of robotic software there is just nobody close. Any attempt to port it to different hardware will be just a big pain and limited result at best. So NVidia looks to be the way to go till China catches up.</p>
]]></description><pubDate>Tue, 18 Mar 2025 19:28:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=43403867</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43403867</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43403867</guid></item><item><title><![CDATA[New comment by numba888 in "Jensen Huang – Nvidia GTC 2025 Keynote"]]></title><description><![CDATA[
<p>cool robot at the end ;)</p>
]]></description><pubDate>Tue, 18 Mar 2025 19:17:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=43403699</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43403699</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43403699</guid></item><item><title><![CDATA[New comment by numba888 in "Jensen Huang – Nvidia GTC 2025 Keynote"]]></title><description><![CDATA[
<p>It looks to be the whole line with different performance. But no dates and times</p>
]]></description><pubDate>Tue, 18 Mar 2025 19:09:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=43403588</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43403588</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43403588</guid></item><item><title><![CDATA[New comment by numba888 in "Jensen Huang – Nvidia GTC 2025 Keynote"]]></title><description><![CDATA[
<p>2 hours in, don't see anything about Digits, their 'personal supercomputer'. Did I miss something?</p>
]]></description><pubDate>Tue, 18 Mar 2025 19:06:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=43403542</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43403542</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43403542</guid></item><item><title><![CDATA[New comment by numba888 in "Image Processing in C (2000) [pdf]"]]></title><description><![CDATA[
<p>It doesn't mean it's wrong. Like partial derivatives are centuries old. But we are moving on. Just recently I've seen a book on image recognition. While recent it was focused on math and not a word about NN. That looked so outdated. Someone spent the whole carer on this way. Another example would be linguistic approach to text translation. 'Stupid' LLM does it way better today. I can only guess how pro linguists feel about it.</p>
]]></description><pubDate>Mon, 17 Mar 2025 23:14:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=43393729</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43393729</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43393729</guid></item><item><title><![CDATA[New comment by numba888 in "Image Processing in C (2000) [pdf]"]]></title><description><![CDATA[
<p>2000-2003, both are pre-historic. We have neural networks now to do things like upscaling and colorization.</p>
]]></description><pubDate>Mon, 17 Mar 2025 06:51:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=43385783</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43385783</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43385783</guid></item><item><title><![CDATA[New comment by numba888 in "A 62M-year-old skeleton sheds light on an enigmatic mammal"]]></title><description><![CDATA[
<p>> At almost three pounds (.4 kilograms), Mixodectes was quite large<p>Something is  wrong here with the numbers..</p>
]]></description><pubDate>Mon, 17 Mar 2025 06:47:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=43385769</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43385769</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43385769</guid></item><item><title><![CDATA[New comment by numba888 in "Arbitrary-Scale Super-Resolution with Neural Heat Fields"]]></title><description><![CDATA[
<p>I did it some time ago. It works at low levels. Doing it in hard cases is non-trivial but possible. It's difficult to monetize that's why nobody is doing it seriously.</p>
]]></description><pubDate>Sun, 16 Mar 2025 18:36:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=43381102</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43381102</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43381102</guid></item><item><title><![CDATA[New comment by numba888 in "Mayo Clinic's secret weapon against AI hallucinations: Reverse RAG in action"]]></title><description><![CDATA[
<p>That would require keeping a library of 'given facts'. Which would include reproducible copyrighted materials or their derivatives.</p>
]]></description><pubDate>Sun, 16 Mar 2025 18:28:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=43381036</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43381036</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43381036</guid></item><item><title><![CDATA[New comment by numba888 in "Ask HN: Any insider takes on Yann LeCun's push against current architectures?"]]></title><description><![CDATA[
<p>That would mean with current resources AI can get so much more intelligent than humans, right?  Aren't you scared?</p>
]]></description><pubDate>Sun, 16 Mar 2025 18:23:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=43381014</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43381014</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43381014</guid></item><item><title><![CDATA[New comment by numba888 in "Ask HN: Any insider takes on Yann LeCun's push against current architectures?"]]></title><description><![CDATA[
<p>In some cases it doesn't recover even without physical or chemical damage. Psychiatric clinics are full of this stories.</p>
]]></description><pubDate>Sun, 16 Mar 2025 18:20:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=43380997</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43380997</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43380997</guid></item><item><title><![CDATA[New comment by numba888 in "Ask HN: Any insider takes on Yann LeCun's push against current architectures?"]]></title><description><![CDATA[
<p>> where every LLM hits some arbitrary amount of context that causes it to "lose focus" and develop a sort of LLM ADD.<p>Humans brains have the same problem. As any intelligence probably. Solution for this is structural thinking. One piece at a time, often top-down. Educated humans do it, LLM can be orchestrated to do it too. Effective context window will be limited even though some claim millions of tokens.</p>
]]></description><pubDate>Sun, 16 Mar 2025 18:16:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=43380964</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43380964</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43380964</guid></item><item><title><![CDATA[New comment by numba888 in "Ask HN: Any insider takes on Yann LeCun's push against current architectures?"]]></title><description><![CDATA[
<p>Great, but how do you imagine multimodal with text, video. Just 2 for simplicity, what will be in the training set. With text model tries to predict next, then more steps were added. But what to do with multimodal?</p>
]]></description><pubDate>Sun, 16 Mar 2025 04:32:31 +0000</pubDate><link>https://news.ycombinator.com/item?id=43376947</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43376947</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43376947</guid></item><item><title><![CDATA[New comment by numba888 in "Gemini Robotics"]]></title><description><![CDATA[
<p>At some point it will to punch him is the face and call a*hole. It should be somewhere in the training set if it has real life data.</p>
]]></description><pubDate>Fri, 14 Mar 2025 05:09:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=43359759</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43359759</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43359759</guid></item><item><title><![CDATA[New comment by numba888 in "Gemini Robotics"]]></title><description><![CDATA[
<p>> What are they gonna do then when all the discussion boards where that data would originate are either gone or optimized into algorithmic metric farms like all the other social media sites?<p>As people are using AI more and more for coding and problems solving providing company can keep records and train on them. I.e. if person 1 solved the problem of doing 2 on product 3 then when 4 is trying to do the same it can be either already trained into the model or model can lookup similar problems and solutions. This way the knowledge isn't gone or isolated, it's being saved and reused. Ideally it requires permission from the user, but price cuts can be motivating. Like all main players today have free versions which can collect interaction data. With millions of users it's much more than online forums have.</p>
]]></description><pubDate>Fri, 14 Mar 2025 04:59:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=43359720</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43359720</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43359720</guid></item><item><title><![CDATA[New comment by numba888 in "Gemini Robotics"]]></title><description><![CDATA[
<p>you probably need robotic leg for walking. or better pony. but doing anything physically requires at least working torso.</p>
]]></description><pubDate>Fri, 14 Mar 2025 04:31:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=43359588</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43359588</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43359588</guid></item><item><title><![CDATA[New comment by numba888 in "Gemini Robotics"]]></title><description><![CDATA[
<p>and a buddy who can get drunk too.</p>
]]></description><pubDate>Fri, 14 Mar 2025 04:21:51 +0000</pubDate><link>https://news.ycombinator.com/item?id=43359545</link><dc:creator>numba888</dc:creator><comments>https://news.ycombinator.com/item?id=43359545</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43359545</guid></item></channel></rss>