vibe-service-guardian
Local-first service attribution, stale development server detection, AI runtime health monitoring, and open-weight model capacity planning.
- 108
- Python
- MIT
- 3 wk ago
Local-first service attribution, stale development server detection, AI runtime health monitoring, and open-weight model capacity planning.
Source-tracked catalog of abliterated, low-refusal, and community-reported open-weight and hosted language models
A developer's guide to uncensored, less-restricted language models: concepts, deployment options, setup, troubleshooting and a model comparison tool.
Index of source-tracked low-filter, abliterated, and community-reported language, image, and video model catalogs
Fine-tune small open models into reliable JSON tool callers. Reproducible, end to end, on a single GPU.
“A clean, from-scratch implementation of the OLMo architecture with KV caching, RoPE, and an efficient autoregressive inference pipeline. Designed as a minimal yet extensible foundation for post-training research, including RLHF, preference optimization, and reasoning-focused systems.”
An independent, evidence-graded read on how much you own the open models and inference providers you rely on.
Multi-agent debate for code generation with local 7-9B LLMs (Ollama): propose, critique, revise, vote. Code, 90-task benchmark and HPC results of a Vilnius University bachelor's thesis (2026).
Open-weight coding and agent models (Qwen Coder, DeepSeek, GLM, Kimi, Devstral, StarCoder) plus the tools to run them locally: llama.cpp, Ollama, vLLM, MLX, ranked by GitHub stars
Text to speech API, speech to text and LLM provider for Vercel AI SDK: official Telnyx package for voice AI apps.
Local, offline web studio for game, video, podcast and social assets: images, 3D models, SFX, music and voices from ~100 open-weight models. Shared asset library, single-GPU job queue, prompt improver, commercial-licence filter, loudness-matched export presets and a built-in MCP server.
Local-first open-weight AI browser demo recorder using Ollama, PHP 8.5, Playwright, FFmpeg and Docker. Automates dummy-data workflows and generates silent MP4, screenshots, SVG, PNG and GIF assets.
A controlled research framework for studying when additional test-time reasoning helps, hurts, or wastes compute in open-weight LLMs.