TimesFM
Summary
TimesFM is Google’s decoder-only time-series forecasting foundation-model family. It treats patched numeric history as a causal sequence-modeling problem and forecasts future patches autoregressively.
Role In The Wiki
TimesFM is one of the canonical decoder-only forecasting TSFMs. It is useful as a contrast to masked-encoder models such as Moirai, compact mixer models such as TTM, and unified multi-task models such as UniTS.
Official Artifacts
- Official code: https://github.com/google-research/timesfm
- Official checkpoint: https://huggingface.co/google/timesfm-2.5-200m-pytorch
Evidence
Relation To Foundation TSFM Agenda
Use the source-level agenda mapping in timesfm-2023 rather than duplicating verdict rows here.
At the entity level, TimesFM is one of the canonical decoder-only forecasting TSFMs. It is useful as a contrast to masked-encoder models such as Moirai, compact mixer models such as TTM, and unified multi-task models such as UniTS. This page should stay as the object card; source pages carry slot-level verdicts, evidence, and missing pieces.