In‑context Time Series Predictor

Applies core ICL principles to TSF, turning forecasting tasks into tokens and leveraging context to induce suitable predictors with efficient, non‑LLM‑scale models.