marlin

Vast.ai LM templates

See official help

Paste-ready launch recipes for rented GPU boxes (Vast custom/docker template). Not loaded by Marlin code.

Operator procedure (tunnel, worker profiles, throughput): Getting Started → Path B.

Configs

File Use
marlin-4090-gemma4-e4b.txt 1× RTX 4090 (24GB), Gemma 4 E4B, official vLLM Gemma image

Each file documents registry, image, disk, ports, env, startup, and notes. Add a new file when you design another setup.