# YarrowLab

Post-training research on **bounded deviation from a statistical prior** for
event forecasting.

Instead of asking a model to output a probability directly, we train it to
output `Δ` — a bounded deviation from a prior `p₀` computed upstream from
reference-class statistics — plus an attribution category for *why* the
prior might be wrong, and an evidence span supporting that attribution.
The final probability `p = σ(logit(p₀) + Δ)` is composed in code; the model
never sees or touches market prices.

## Repositories

- [`yarrow-delta-sft-v1-smoke`](https://huggingface.co/YarrowLab/yarrow-delta-sft-v1-smoke) —
  a pipeline-validation checkpoint (LoRA over a randomly initialized tiny
  model). Confirms the SFT loop runs end to end; not a trained model.

More will be added here as real training runs land.