Preprint
Aug 2026
Dual-Node NVIDIA DGX Spark over Tailscale: A Remote-Access Testbed for Distributed LLM Training and Cyber-Threat-Intelligence Fine-Tuning
A proof-of-concept deployment of distributed NanoChat pretraining across two NVIDIA DGX Spark systems is presented, establishing feasibility rather than a scaling-efficiency claim, since single-node throughput used for comparison was estimated, not measured under matched conditions.
V. Iyer
· 0 citations