8 projects
typevet
Typed answers from model next-token probabilities, on llama.cpp and vLLM
gepa-adk
Evolutionary optimization for Google ADK agents
judgevet
Typed client, CLI and MCP server for TypeSafe's Jev (System One) judgment model
docvet
Comprehensive docstring quality vetting for Python projects
saucier
Structured procedure extraction from culinary sources.
vramfit
Selective per-layer quantization to fit large open models on a single GPU
adk-secure-sessions
Encrypted session persistence for Google ADK agents
turboquant-vllm
TurboQuant KV cache compression for vLLM — fused Triton kernels, 3.76x compression, 3.7x faster decode on RTX 4090
