Skip to content

Instantly share code, notes, and snippets.

View JunEnomoto's full-sized avatar

Jun Enomoto JunEnomoto

View GitHub Profile
@JunEnomoto
JunEnomoto / compose.yaml
Created April 22, 2026 15:20
RedHatAI/Qwen3.6-35B-A3B-NVFP4 + DFlash on DGX Spark
# RedHatAI/Qwen3.6-35B-A3B-NVFP4 + DFlash on DGX Spark
#
# - Image : ghcr.io/aeon-7/vllm-spark-omni-q36:v1.2
# - Model : RedHatAI/Qwen3.6-35B-A3B-NVFP4 (~22GB)
# - Drafter: z-lab/Qwen3.6-35B-A3B-DFlash (~905MB)
#
# Usage:
# 1. download model: docker compose run --rm -d model-download
# 2. vLLM: docker compose up -d
services: