v0.0.36

Features

• T0 family checkpoints: Document and compare T0 alpha and beta via the existing T0 class (foundationforecast>=0.1.8, tfc-t0>=0.5.0). Pass repo_id to select the checkpoint:

  • theforecastingcompany/t0-alpha (default) — ~102M parameters
  • theforecastingcompany/t0-beta — ~256M parameters (model card)

  import pandas as pd
  from timecopilot.models.foundation.t0 import T0
  
  df = pd.read_csv(
      "https://timecopilot.s3.amazonaws.com/public/data/air_passengers.csv",
      parse_dates=["ds"],
  )
  
  model = T0(
      repo_id="theforecastingcompany/t0-beta",
      alias="t0-beta",
      context_length=512,
  )
  fcst_df = model.forecast(df, h=12, freq="MS", level=[80, 90])

• T0 family notebook: Added t0-family example notebook comparing t0-alpha and t0-beta on event page-views, including prediction intervals and evaluation metrics.

Fixes

• GIFT-Eval CI: Point Seasonal Naive baseline downloads at results/Seasonal_Naive/ on the GIFT-Eval Hugging Face space (the old seasonal_naive path now 404s).

---

Full Changelog: v0.0.35...v0.0.36