Model packs / M0
M0 · inferix-model-foundations
Model Foundations
Why this pack? Era A literacy — train and inspect from scratch
nanoGPT training loop and open-weight inspection. Teaches the team how checkpoints, eval harnesses, and promote gates work before wiring agents.
status: done
Training focus
Character-level LM from scratch; tokenizer + checkpoint hygiene
Inferix role
Baseline for FineForge promote/rollback semantics and eval set format
From toy model to owned inference
M0 teaches the loop. M1–M3 wire real agents. Start with the quick start, then open the console.