Awaiting GPU hardware

⏳ TimesFM 3.0 — zero-shot time-series forecasting

A Gradio demo of google/timesfm-3.0-pytorch, Google Research's time-series foundation model. Hand it history, get back a distribution over the future — no training, no fitting, no per-dataset tuning.

What the app does

Forecast a series

Pick a bundled dataset or upload a CSV, choose how much history to feed the model and how far ahead to predict. The plot shows the median forecast with the 80% (q0.1–q0.9) and 40% (q0.3–q0.7) quantile bands.

A backtest that scores itself

The last horizon observations are held out rather than shown to the model, then the forecast is scored against them: MAE, RMSE, sMAPE, the ratio against a naive baseline, and what fraction of the truth actually landed inside the 80% interval — the honest test of whether the intervals mean anything.

Multivariate forecasting with covariates

TimesFM 3.0's headline feature. Several correlated columns are forecast jointly, with attention across variates, optionally conditioned on past-only covariates and on covariates known over the forecast window too.

Bundled datasets

Airline passengers, monthly 1949–1960 — multiplicative yearly seasonality on a trend
Melbourne minimum temperature, daily 1981–1990 — smooth annual cycle, noisy day to day
Monthly sunspots, 1749–1983 — the irregular ~11-year solar cycle
Female births in California, daily 1959 — near-stationary counts, very little structure
ETTh1 electricity transformer, hourly — 7 correlated variates, daily and weekly cycles

Why this page instead of the app. Running a Gradio Space on Hugging Face requires PRO or a community GPU grant. The full application lives in this repo — see app.py — and targets ZeroGPU. Once hardware is attached, switching sdk: static to sdk: gradio in the README frontmatter brings it up; no code changes are needed.