PrototypeCommon
raullenchai/twinspark
Production-grade recipe for DeepSeek-V4-Flash-0731 (284B MoE) on 2x NVIDIA DGX Spark: self-healing 2-node vLLM cluster, reboot-verified, tuned DSpark speculative decoding (~75 tok/s), full benchmarks, OpenAI Codex CLI integration
- Owner
- raullenchai (personal account)
- Stars
- Stars: 176
- Forks
- Forks: 20
- Language
- Shell
- License
- MIT
- Created
- 2026
- Topics
- deepseekdgx-sparkllm-inferencelocal-llmvllm
- Tier
- Prototype (100 to 249 ★) What this tier means
More repos by raullenchai
All 4 repos by raullenchai- raullenchai/Rapid-MLXStars: 4,098PythonRC
- raullenchai/esp32-thinksStars: 262ShellAlpha
- raullenchai/vnshStars: 158TypeScriptPrototype
Nearby in Shell
All Shell repos- jmckinley/claude-code-resourcesStars: 176ShellPrototype
- enp6/ZelayStars: 176ShellPrototype
- V2RaySSR/3x-ui-cn-installerStars: 176ShellPrototype
- rudimeier/bash_ini_parserStars: 176ShellPrototype
- joukewitteveen/netctlStars: 176ShellPrototype
- mm2270/JamfProScriptsStars: 176ShellPrototype