Time series · Google
TimesFM 2.5
200M decoder-only forecaster from Google Research.
Datasheet
Maker | |
|---|---|
Parameters | 231 MVerifiedsafetensors.total from the HF API Source |
Licence | Apache-2.0Verified |
Access | Public download |
Task (HF pipeline) | time-series-forecasting |
First published on HF | 2025-09-02 |
Last updated on HF | 2025-10-02 |
Weights in memory | 0.5 GB at 16-bit · 0.1 GB at ~4.85 bitsEstimateParameters × bits ÷ 8. Excludes context cache and runtime overhead; not every model has a 4-bit build. |
Benchmark
0.705
GIFT-Eval MASE, geometric mean relative to seasonal naive (lower is better) · #2 of 7 scored in this category
Computed: MASE 0.705, CRPS 0.490; no leakage per config.json. Source
Used in
The description summarises the model card at https://huggingface.co/google/timesfm-2.5-200m-pytorch. Parameters and licence are read from the Hugging Face API; the weights-in-memory line is our arithmetic, not a measurement.