가입 후 초대 링크를 공유하면 동영상 재생 및 초대 보상을 받을 수 있습니다.

Mia
@MiaAI_lab
Building with AI & LLMs // Running frontier open models locally
가입 July 2022
359 팔로잉 중    41.8K 팬
I get a lot of data on what people actually run on their Sparks. Here’s what I have: Solo DGX Spark - Most run Qwen3.8 Flash Next 2x DGX Sparks - GLM 5.3 Flash or DeepSeek v4.1 Flash 3x DGX Sparks, conf 1: - GLM 5.3 Flash on 2x DGX Sparks - Qwen3.8-Flash on solo DGX Spark - GLM as Orchestrator and Qwen 3.8 Flash Next as worker 3x DGX Sparks, conf 2: - GLM 5.3 Flash on all three, or - DeepSeek v4.1 Flash on all three 4x DGX Sparks, conf 1: - GLM 5.3 Flash on 2x DGX Sparks - Qwen3.8 Flash on 2x DGX Sparks - Logic: GLM as Orchestrator and Qwen 3.8 Flash Next as worker 4x DGX Sparks, conf 2: - DeepSeek v4.1 Flash on all four, or - GLM 5.3 Flash on all four 6+ DGX Sparks: With 6 or more, you have a LOT of flexibility. You can run many different combinations, but I hear from a lot of users who run this specific setup: - GLM 5.3 Flash on 2x DGX Sparks - Qwen3.8 Flash as on 2x DGX Sparks - DeepSeek v4 Flash on 2x DGX Sparks - Logic: GLM 5.3 Flash as orchestrator) + Qwen 3.8 Flash as worker + Deepseek v4 Flash as a security scanner/cyber. Me personally? I run GLM 5.3 Flash on all three of my DGX Sparks What are you running?
더 보기