timesfm-ts - v1.0.0
    Preparing search index...

    Module decode-loop

    Autoregressive decode loop for TimesFM.

    This is the core inference algorithm:

    Phase 1 — Prefill: Feed all context patches through the transformer. The model produces output patches (each covering outputPatchLen future steps) and populates the KV cache.

    Phase 2 — Autoregressive Decode: Take the last output patch's median value, split it into sub-patches, feed each back through the model to generate the next chunk of the forecast horizon.

    Mirrors the decode() method in timesfm_2p5_torch.py.

    Interfaces

    DecodeResult

    Functions

    decode