Pinned Loading

1. dotfiles dotfiles Public

   Personal dotfiles managed with GNU Stow

   Shell

2. vllm-qwen36 vllm-qwen36 Public

   Serve Qwen3.6 NVFP4 on Blackwell GPUs with vLLM - full 262K context, fp8 KV cache, and MTP speculative decoding via Docker Compose

   Shell

3. llama-qwen36 llama-qwen36 Public

   Dockerized llama.cpp Vulkan server setup for running Qwen3.6 27B GGUF on AMD GPUs.

   1

4. nvfp4-vllm nvfp4-vllm Public

   Scripts for quantizing HuggingFace models to NVFP4 (4-bit) and serving them with vLLM on NVIDIA Blackwell GPUs. Includes an interactive chat client for testing. Confirmed working on RTX PRO 6000 Bl…

   Python 1

5. llm-serve llm-serve Public

   Self-hosted OpenAI-compatible LLM inference server for NVFP4 models on NVIDIA Blackwell GPUs, powered by vLLM and Docker.

   Shell