Mark Allen Evans PRO

I build and train my own small language models, including a 1.6B parameter model that turns natural language into SQL. My interests are small model training and fine tuning, distillation from large teacher models, GGUF quantization and local inference, deterministic model serving, and peer-to-peer inference infrastructure. I publish open-weight model conversions under Interchained and I am currently exploring self-modifying models with validation gates and a new project on what comes after AI.