Neuramorphic Research
Technical primers, architecture notes and engineering write-ups on neuromorphic computing, spiking neural networks, state-space models and the production neuromorphic LLM stack — Caroline / NeuratronLLM-Edge 4B — running air-gapped on NVIDIA Jetson AGX Orin.
How Neuramorphic runs a 4B-parameter neuromorphic LLM (Caroline / NeuratronLLM-Edge 4B) on NVIDIA Jetson AGX Orin: memory budget, sparse activation, event-driven inference and the trade-offs that make air-gapped foundation-model deployment viable on edge silicon.
Read the primer →SNNs and SSMs are often grouped together as the alternatives to attention, but they encode different assumptions about time and behave very differently on real hardware. A practical comparison from the perspective of building a production neuromorphic LLM.
Read the primer →Technical primer on spiking neural networks, state-space models and event-driven inference — and how Neuramorphic ships these as a production neuromorphic LLM (Caroline / NeuratronLLM-Edge 4B) on commodity NVIDIA Jetson silicon.
Read the primer →