Altan "Atabarezz" Atabarut·Nov 1When Claude Starts Scoring Its Users: Social Credit or Just Another Risk Engine? 🤖Some people call it a social credit score. Others say it’s just another risk-scoring mechanism. I think the interesting question is…
Insofia’s tech cubiclebysofia·Oct 27no one has ANY idea how China dropped a model like thisand it is more frontier than frontierA response icon1A response icon1
Inmy_aimlbySanjay Basu, PhD·Sep 22Dark Knowledge and the Enclosure ProblemA compression trick from 2015 became the subject of a joint NSA, CISA and FBI advisory. The path between those two facts is worth walking…
Ayush Varma·Sep 18Distillation: The Intelligence DiscountReaching frontier level capability for a fraction of what the frontier cost to build and why the industry is now at war over whether it…
InEnrique DansbyEnrique Dans·Sep 10Why is it that when US companies steal it’s innovation, but when China does, it’s theft?On September 1, the US Department of Justice filed a brief supporting OpenAI’s position in defense of the lawsuit brought against it by the…A response icon7A response icon7
Dr. Olga D. Khon·Sep 7Note 171. “Cheap Open-Source AI Models: Yet Another Illegal Chinese Replica?”The launch of DeepSeek in January 2025 severely hit US tech stocks, causing a loss of nearly USD 1 trillion back then (VOA, 2025). The…
Felipe Penhorate·Aug 23Teaching an LLM to Take Turns Without Forgetting Everything ElseSelf-distillation with plain supervised fine-tuning: how I trained a full-duplex voice model that keeps its function-calling skills intactA response icon1A response icon1
ComfyAI·Aug 21Close in KL, Far in RepresentationIf two models produce nearly the same output distribution, are their internal representations nearly the same too?
Almohtady Bellah·Aug 13Meta’s “Open Source Comeback” Skips One NumberThe model’s free. The math isn’t what you think.
