Command example to run Qwen3.8-27B GGUF on DGX Spark

ariG23498 · x · 2026-08-17

Demonstrates the specific command to run the Qwen3.8-27B GGUF model using llama serve on the DGX Spark platform. The configuration enables speculative decoding with spec-default and spec-type draft-mtp, along with reasoning preservation and agent mode.

Original post →

More from Infra

Infra channel →