HYPERKAM: A Human Brain-Inspired AI Mind 2nd run скачать в хорошем качестве

HYPERKAM: A Human Brain-Inspired AI Mind 2nd run

HYPERKAM is a biologically-inspired cognitive architecture that maps 44 brain structures onto modular software components powered by Large Language Models. It's not a chatbot — it's an attempt to build a living cognitive system that perceives, thinks, remembers, sleeps, dreams, and maintains a persistent identity. In this live stream, we're running the HYPERKAM dashboard in real time, showing how all cognitive modules interact tick by tick. What makes HYPERKAM different: • 44 brain-mapped modules organized into 12 neuroscientific clusters (Sensory, Thalamic, Limbic, Memory, Executive, Social, Decision, Neuromodulation, Metacognition, Reward, Cerebellum, Inner Life) • Global Workspace Theory — consciousness implemented as an information bottleneck, with three activation levels (OFF / PARTIAL / FULL) that dynamically gate which modules process each tick • Dual-store memory with episodic encoding (Qdrant vector DB) and sleep-based consolidation — the system literally dreams to reorganize its memories • Full sleep cycle: NREM1 → NREM2 → SWS → REM in 90-minute cycles, with memory replay and creative recombination during dreams • Neuromodulation bus simulating dopamine, serotonin, norepinephrine, acetylcholine, and GABA — affecting decision-making, mood, and attention across all modules • Personality, moral reasoning, social cognition, and emotional regulation — not hardcoded, but emergent from interacting modules • Inner monologue and imagination via Default Mode Network activation during idle ticks • Autonomous goal generation, action selection (basal ganglia Go/NoGo pathways), and web interaction capabilities • Real-time monitoring dashboard showing module activity, consciousness state, emotional valence, and cognitive events • Runs entirely on local/self-hosted infrastructure: Built with Python, asyncio, and a tick-based cognitive loop. 3000+ tests. Open for discussion in the chat. #AI #CognitiveArchitecture #Neuroscience #ArtificialMind #LLM #OpenSource #LiveCoding