<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: kbumsik</title><link>https://news.ycombinator.com/user?id=kbumsik</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Tue, 26 May 2026 19:07:52 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=kbumsik" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by kbumsik in "Eagle 3.1: Collaboration Between the EAGLE Team, vLLM Team, and TorchSpec Team"]]></title><description><![CDATA[
<p>So the draft model's performance is directly linked to the overall speed. Thank you for the explanation!<p>By the way, can it be slower than without speculative decoding in worst case then?</p>
]]></description><pubDate>Tue, 26 May 2026 14:51:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=48280650</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=48280650</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48280650</guid></item><item><title><![CDATA[New comment by kbumsik in "Eagle 3.1: Collaboration Between the EAGLE Team, vLLM Team, and TorchSpec Team"]]></title><description><![CDATA[
<p>> performance often degrades under different chat templates, long-context inputs, or out-of-distribution system prompts.<p>I heard that speculative decoding doesn't affect performance (I meant accuracy). Am I wrong about it?</p>
]]></description><pubDate>Tue, 26 May 2026 14:30:14 +0000</pubDate><link>https://news.ycombinator.com/item?id=48280355</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=48280355</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48280355</guid></item><item><title><![CDATA[Kimi K2.6]]></title><description><![CDATA[
<p>Article URL: <a href="https://huggingface.co/moonshotai/Kimi-K2.6">https://huggingface.co/moonshotai/Kimi-K2.6</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47835621">https://news.ycombinator.com/item?id=47835621</a></p>
<p>Points: 11</p>
<p># Comments: 1</p>
]]></description><pubDate>Mon, 20 Apr 2026 15:20:28 +0000</pubDate><link>https://huggingface.co/moonshotai/Kimi-K2.6</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=47835621</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47835621</guid></item><item><title><![CDATA[New comment by kbumsik in "FCC updates covered list to include foreign-made consumer routers"]]></title><description><![CDATA[
<p>Right? Even enterprise routers, e.g. Cisco, are not produced in USA.</p>
]]></description><pubDate>Mon, 23 Mar 2026 22:45:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=47496169</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=47496169</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47496169</guid></item><item><title><![CDATA[Logitech Options+ and G Hub macOS Certificate Issue]]></title><description><![CDATA[
<p>Article URL: <a href="https://support.logi.com/hc/en-us/articles/37493733117847-Options-and-G-HUB-macOS-Certificate-Issue">https://support.logi.com/hc/en-us/articles/37493733117847-Options-and-G-HUB-macOS-Certificate-Issue</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=46523062">https://news.ycombinator.com/item?id=46523062</a></p>
<p>Points: 3</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 07 Jan 2026 05:51:16 +0000</pubDate><link>https://support.logi.com/hc/en-us/articles/37493733117847-Options-and-G-HUB-macOS-Certificate-Issue</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=46523062</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46523062</guid></item><item><title><![CDATA[New comment by kbumsik in "A first look at Django's new background tasks"]]></title><description><![CDATA[
<p>We switched from Celery to Temporal. Temporal is such a great piece of distributed system.</p>
]]></description><pubDate>Sat, 29 Nov 2025 06:32:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=46085569</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=46085569</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=46085569</guid></item><item><title><![CDATA[New comment by kbumsik in "DeepSeek OCR"]]></title><description><![CDATA[
<p>> realistic written documents?<p>Just get a DEF 14A (Annual meeting) filing of a company from SEC EDGAR.<p>I have seen so many mistakes when looking at the result closely.<p>Here is a DEF 14A filing from Salseforce. You can print it to a PDF and then try converting.<p><a href="https://www.sec.gov/Archives/edgar/data/1108524/000110852425000009/crm-20250424.htm" rel="nofollow">https://www.sec.gov/Archives/edgar/data/1108524/000110852425...</a></p>
]]></description><pubDate>Mon, 20 Oct 2025 10:39:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=45642356</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=45642356</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45642356</guid></item><item><title><![CDATA[New comment by kbumsik in "DeepSeek OCR"]]></title><description><![CDATA[
<p>Right, I wouldn't use full table detection to VLM model because they tend to mistake with numbers in table...</p>
]]></description><pubDate>Mon, 20 Oct 2025 09:13:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=45641692</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=45641692</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45641692</guid></item><item><title><![CDATA[New comment by kbumsik in "DeepSeek OCR"]]></title><description><![CDATA[
<p>> My impression is that OCR is basically solved at this point.<p>Not really in practice to me. Especially they still struggle with Table format detection.</p>
]]></description><pubDate>Mon, 20 Oct 2025 08:51:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=45641470</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=45641470</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45641470</guid></item><item><title><![CDATA[New comment by kbumsik in "Auth.js is now part of Better Auth"]]></title><description><![CDATA[
<p>Well, that has nothing to do with JS itself though.</p>
]]></description><pubDate>Sat, 27 Sep 2025 09:06:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=45394273</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=45394273</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45394273</guid></item><item><title><![CDATA[New comment by kbumsik in "Auth.js is now part of Better Auth"]]></title><description><![CDATA[
<p>In case if you don't know, Auth.js is not a frontend-only framework. It uses a backend server to make it secure.<p>So it basically has no difference from the alternatives you mentioned.</p>
]]></description><pubDate>Sat, 27 Sep 2025 00:08:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=45392290</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=45392290</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45392290</guid></item><item><title><![CDATA[New comment by kbumsik in "UUIDv47: Store UUIDv7 in DB, emit UUIDv4 outside (SipHash-masked timestamp)"]]></title><description><![CDATA[
<p>It is really easy to tell the difference btw. You will always see "4" or "7" in the middle.</p>
]]></description><pubDate>Fri, 19 Sep 2025 01:44:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=45297089</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=45297089</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45297089</guid></item><item><title><![CDATA[New comment by kbumsik in "Show HN: Pangolin – Open source alternative to Cloudflare Tunnels"]]></title><description><![CDATA[
<p>How does it compare to frp, one of the most popular Open Source Cloudflare Tunnel alternative?<p><a href="https://github.com/fatedier/frp">https://github.com/fatedier/frp</a></p>
]]></description><pubDate>Fri, 11 Jul 2025 20:36:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=44536560</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=44536560</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44536560</guid></item><item><title><![CDATA[New comment by kbumsik in "Deepseek R1-0528"]]></title><description><![CDATA[
<p>Artificial Analysis is the only stable source. Don't look at others like HF Leaderboard.<p><a href="https://artificialanalysis.ai/" rel="nofollow">https://artificialanalysis.ai/</a></p>
]]></description><pubDate>Wed, 28 May 2025 23:38:56 +0000</pubDate><link>https://news.ycombinator.com/item?id=44121573</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=44121573</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44121573</guid></item><item><title><![CDATA[New comment by kbumsik in "CVE program faces swift end after DHS fails to renew contract"]]></title><description><![CDATA[
<p>Because gov infra also relies on CVE?</p>
]]></description><pubDate>Wed, 16 Apr 2025 08:00:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=43702722</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=43702722</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43702722</guid></item><item><title><![CDATA[New comment by kbumsik in "CVE program faces swift end after DHS fails to renew contract [updated]"]]></title><description><![CDATA[
<p>> it probably is an anomaly that this was government funded<p>Companies can definitely fund it. But to be fair the gov, including NIST, also relies on CVE.</p>
]]></description><pubDate>Wed, 16 Apr 2025 07:58:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=43702699</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=43702699</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43702699</guid></item><item><title><![CDATA[New comment by kbumsik in "AWS S3 sync does not sync all the files"]]></title><description><![CDATA[
<p>Yup. aws-cli is not dedicated for file sync. It was strange that there is sync command when aws-cli is meant to be a wrapper for a single api calls at the beginning.<p>Use a tool that does one thing well like rclone.</p>
]]></description><pubDate>Tue, 01 Apr 2025 12:42:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=43546095</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=43546095</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43546095</guid></item><item><title><![CDATA[New comment by kbumsik in "Show HN: Telescope – an open-source web-based log viewer for logs in ClickHouse"]]></title><description><![CDATA[
<p>It is not same as OP, but according to a similar o11y stack on top of Clickhouse, Signoz, Clickhouse based logging costs less than ELK for storage and performs better:<p><a href="https://github.com/SigNoz/logs-benchmark">https://github.com/SigNoz/logs-benchmark</a></p>
]]></description><pubDate>Wed, 26 Feb 2025 15:33:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=43184585</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=43184585</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43184585</guid></item><item><title><![CDATA[New comment by kbumsik in "Show HN: Telescope – an open-source web-based log viewer for logs in ClickHouse"]]></title><description><![CDATA[
<p>How is it different from Signoz, a complete observability stack (including Logs) built on top of Clickhouse?</p>
]]></description><pubDate>Wed, 26 Feb 2025 15:27:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=43184536</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=43184536</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43184536</guid></item><item><title><![CDATA[New comment by kbumsik in "New 16GB Raspberry Pi 5 on sale now at $120"]]></title><description><![CDATA[
<p>Right. PoE will be fantastic.</p>
]]></description><pubDate>Thu, 09 Jan 2025 15:12:48 +0000</pubDate><link>https://news.ycombinator.com/item?id=42646413</link><dc:creator>kbumsik</dc:creator><comments>https://news.ycombinator.com/item?id=42646413</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=42646413</guid></item></channel></rss>