https://locollm.org/https://locollm.org/adapter-guide/https://locollm.org/adapter-primer/https://locollm.org/adr/https://locollm.org/adr/0001-base-model-qwen3-4b/https://locollm.org/adr/0002-adapter-registry-design/https://locollm.org/adr/0003-single-evolving-router/https://locollm.org/adr/0004-retire-cerebro-adopt-b250-multi-gpu/https://locollm.org/adr/0005-weiho-8gpu-replace-b250-colmena/https://locollm.org/adr/0006-gguf-ollama-inference-standard/https://locollm.org/architecture-vision/https://locollm.org/architecture/https://locollm.org/base-model-selection/https://locollm.org/benchmarking-guide/https://locollm.org/benchmarks/https://locollm.org/evaluation-standards/https://locollm.org/faq/https://locollm.org/ideas/https://locollm.org/known-challenges/https://locollm.org/locoagente-integration/https://locollm.org/project-ideas/https://locollm.org/research-roadmap/https://locollm.org/scaffolding-studies/https://locollm.org/scaffolding-studies/re2/https://locollm.org/small-model-strategies/https://locollm.org/train-math-adapter/https://locollm.org/training-new-adapters/https://locollm.org/why-locollm/