Pinned Loading

1. dotfilesdotfilesPublic

   Personal dotfiles managed with GNU Stow

   Shell

2. vllm-qwen36vllm-qwen36Public

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

   Shell

3. vllm-gemma4vllm-gemma4Public

   Serve Gemma 4 NVFP4 on Blackwell GPUs with vLLM - OpenAI-compatible API with full 262K context, vision, thinking, and tool calling via Docker Compose

   Shell

4. nvfp4-vllmnvfp4-vllmPublic

   Quantize HuggingFace models to NVFP4 and serve them with vLLM on NVIDIA Blackwell GPUs.

   Python 2

5. llama-qwen36llama-qwen36Public

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

   1

6. slugvisionslugvisionPublic

   Tiny VLMs that turn an image + optional article title into a 3-5 word permalink slug — distillation pipeline, training, eval, and GGUF export

   Python