Mamba
Summary
Mamba is a selective state space model architecture that uses input-dependent recurrent dynamics and hardware-aware parallel scans to make attention-free sequence modeling competitive with Transformers.
Role In The Wiki
Mamba is the baseline background model for selective SSMs and efficient recurrent sequence mixers. It is most relevant as an architecture and systems source rather than as direct evidence about numeric time-series forecasting.
Evidence
Relation To Foundation TSFM Agenda
Use the source-level agenda mapping in mamba-2023 rather than duplicating verdict rows here.
At the entity level, Mamba is the baseline background model for selective SSMs and efficient recurrent sequence mixers. It is most relevant as an architecture and systems source rather than as direct evidence about numeric time-series forecasting. This page should stay as the object card; source pages carry slot-level verdicts, evidence, and missing pieces.
Overlap Notes
Mamba is the implicit compact-state counterpart to RMT and ARMT. RMT-style models expose memory slots as tokens; Mamba keeps the state in structured recurrent dynamics with selective updates. RWKV-TS is the direct local time-series bridge for a related recurrent-serving intuition, while MIRAS is useful when the comparison is framed as memory architecture and retention rather than SSM mechanics.