sbagencyMoshi, conversational AIIn just 6 months, with a team of 8, the Kyutai research lab developed from scratch an AI model with unprecedented vocal capabilities called…1d ago1d ago
sbagencyBest embedding models // vector searchbi-encoder and cross-encoder // top-k and re-ranking2d ago2d ago
sbagencyCombining vector & “scalar” (key words) search in a RAG pipelines // BM42 & beyondRAG is a hot (most popular) generative AI application for enterprises today3d ago3d ago
sbagencyNew computing paradigm we are entering to // Andrej Karpathy’s KeynoteLLM OS // big cybersecurity concern but “keep hacking”4d ago4d ago
sbagencyCriticGPT to critique other GPTs //Reinforcement learning from human feedback (RLHF) is fundamentally limited by the capacity of humans to correctly evaluate model output…5d ago5d ago
sbagencyKGs + LLMs // convergence of structured and unstructuredHybrid graph and vector search is the future of RAGJun 27Jun 27
sbagencyARC Prize 2024 // $1MJean Piaget, defined intelligence as “what you use when you don’t know what to do.”Jun 261Jun 261
sbagencyHallucination-free AI // is it possible?)There are some techniques, like RAG, GraphRAG, Knowledge graphs, etc. to reduce hallucinations…Jun 26Jun 26
sbagencyAdvanced RAG techniques // for better accuracyRetrieval augmented generation (RAG) is a core function of modern AI pipelinesJun 23Jun 23
sbagencyAI generated content {humanization, detection} // AI detecting AIHumanize AI generated contentJun 20Jun 20