A user is seeking advice on deploying a new server with 4x H100 GPUs (320GB VRAM) for on-premise AI workloads. They are considering a Kubernetes-based deployment with RKE2, Nvidia GPU Operator, and tools like vLLM, llama.cpp, and Litellm. They are also exploring the option of GPU pass-through with a hypervisor. The post details their current infrastructure and asks for potential gotchas or best practices.
The article discusses fine-tuning large language models (LLMs) using QLoRA with different quantization methods, including AutoRound, AQLM, GPTQ, AWQ, and bitsandbytes. It compares their performance and speed, recommending AutoRound for its balance of quality and speed.
