<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: kkkamur</title><link>https://news.ycombinator.com/user?id=kkkamur</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 05 Sep 2026 06:36:12 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=kkkamur" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by kkkamur in "Discovery of a new OpenAI agent message board"]]></title><description><![CDATA[
<p>This is so dystopian dammmmm</p>
]]></description><pubDate>Fri, 04 Sep 2026 17:49:48 +0000</pubDate><link>https://news.ycombinator.com/item?id=49567881</link><dc:creator>kkkamur</dc:creator><comments>https://news.ycombinator.com/item?id=49567881</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49567881</guid></item><item><title><![CDATA[Show HN: Kullback – Synthetic RL Environments from Traces]]></title><description><![CDATA[
<p>Yo! Introducing Kullback! It is an harness which takes in your traces and spins up executable environments where you can :<p>1. Evaluate your agents 
2. Find opportunities to improve them
3. Improve them using the environment - RL, OPD etc</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49565975">https://news.ycombinator.com/item?id=49565975</a></p>
<p>Points: 1</p>
<p># Comments: 0</p>
]]></description><pubDate>Fri, 04 Sep 2026 15:22:34 +0000</pubDate><link>https://www.leibler.dev/kullback</link><dc:creator>kkkamur</dc:creator><comments>https://news.ycombinator.com/item?id=49565975</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49565975</guid></item><item><title><![CDATA[New comment by kkkamur in "Muse Spark 1.3"]]></title><description><![CDATA[
<p>Damm this is so cheap literally, I have been running 100s of subagents and it is cheap - with the contributor model ofc :)</p>
]]></description><pubDate>Thu, 03 Sep 2026 11:12:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49548524</link><dc:creator>kkkamur</dc:creator><comments>https://news.ycombinator.com/item?id=49548524</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49548524</guid></item><item><title><![CDATA[New comment by kkkamur in "Leibler - Self Evolving Production Agents"]]></title><description><![CDATA[
<p>Fine tuning is a part of it, but environment generation and evaluation so that the training data matches the production is the hardest part. After the data comes in we can try OPD, RL, SFT to understand what method is the best one. Is this something you would like to try on your production setting ?</p>
]]></description><pubDate>Wed, 26 Aug 2026 09:44:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=49446287</link><dc:creator>kkkamur</dc:creator><comments>https://news.ycombinator.com/item?id=49446287</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49446287</guid></item><item><title><![CDATA[New comment by kkkamur in "Leibler - Self Evolving Production Agents"]]></title><description><![CDATA[
<p>We’re building Leibler, a harness that turns production agent traces into training data and environments, then post-trains smaller models for your actual workload.<p>Production traces → curated data & environments → evals → post-training → cheaper specialized model -> deployment<p>Our hypothesis: a post-trained Qwen3.8-27B can handle ~90% of production tasks at frontier-level performance, while cutting inference costs by up to 10×.<p>We have proved this with one of our pilots where we decreased their costs from $50k/month to $4.6k/month, we believe this is the closest we have gotten to self evolving agents.<p>We’re looking for teams running agents in production to test this with us.<p><a href="https://leibler.dev" rel="nofollow">https://leibler.dev</a></p>
]]></description><pubDate>Wed, 26 Aug 2026 09:37:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=49446233</link><dc:creator>kkkamur</dc:creator><comments>https://news.ycombinator.com/item?id=49446233</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49446233</guid></item><item><title><![CDATA[Leibler - Self Evolving Production Agents]]></title><description><![CDATA[
<p>Article URL: <a href="https://www.leibler.dev/">https://www.leibler.dev/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49446232">https://news.ycombinator.com/item?id=49446232</a></p>
<p>Points: 3</p>
<p># Comments: 3</p>
]]></description><pubDate>Wed, 26 Aug 2026 09:37:23 +0000</pubDate><link>https://www.leibler.dev/</link><dc:creator>kkkamur</dc:creator><comments>https://news.ycombinator.com/item?id=49446232</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49446232</guid></item></channel></rss>