Y Hacker News new | ask | show | jobs

user: visha1v created: 2025-11-14 karma: 73 CTO at Dehurdle. IIT Delhi. Building at the intersection of physics and AI systems. github.com/vishal-dehurdle | vishalvermalabs.com submissions: Show HN: Growth-Ratio Energy Function as Leading Indicator of Agent Task Failure 1 points | 0 comments Topological Control of LLMs: A Route to Trustworthy AI 1 points | 0 comments The Dawn of AI Change-Control 1 points | 0 comments Growth-Ratio Energy Functions as Leading Indicators of Agent Task Failure 4 points | 0 comments Capturing token IDs during agentic interaction for better reinforcement learning 3 points | 0 comments Comparing Machine Learning Models of Raindrop Formation 2 points | 0 comments Show HN: A control-theory approach to detecting LLM agent instability 2 points | 0 comments Why Triso Matters 2 points | 0 comments Can We Understand How Large Language Models Reason? 3 points | 1 comments Never ask a model to do something a deterministic system can do reliably 3 points | 1 comments 0 points | 0 comments Is Recursive Self-Improvement Here? 3 points | 2 comments Details on Fable 5's cyber safeguards and jailbreak framework 2 points | 0 comments 0 points | 0 comments Ebrains Crowdsources the Future of European Neuroscience 2 points | 1 comments 0 points | 0 comments Algae Microrobots Battle Bladder Cancer 4 points | 1 comments 0 points | 0 comments But Nothing Has Changed on Our Side 3 points | 1 comments 0 points | 0 comments 0 points | 0 comments Two channels for the top–antitop excess 1 points | 1 comments 0 points | 0 comments Illuminating Iron Clusters' Magnetism 2 points | 1 comments 0 points | 0 comments A New Catalog of Stellar Rotation Periods for over a Million Stars 38 points | 3 comments 0 points | 0 comments The FCC, Half a Century On 2 points | 1 comments