FlowStateOptions<T>

Options & Configuration · AiDotNet.Models.Options

Configuration options for FlowState (IBM’s SSM-based Time Series Foundation Model).

Inheritance: ModelOptions → RegressionOptions<T> → TimeSeriesRegressionOptions<T> → FlowStateOptions<T>

For Beginners

FlowState proves bigger isn’t always better:

State-Space Model Architecture: Instead of attention (which is quadratic in sequence length), FlowState uses structured state spaces (like S4/Mamba) that are linear in sequence length. This makes it extremely efficient for long sequences.

Key Advantages:

  • Only 9.1M parameters (smallest in GIFT-Eval top 10)
  • Outperforms models 20x its size
  • Generalizes to unseen timescales
  • Linear-time computation for long sequences

How It Works

FlowState is IBM’s State-Space Model (SSM) based time series foundation model with only 9.1M parameters. Despite being the smallest model in the GIFT-Eval top 10, it outperforms models 20x its size and generalizes to unseen timescales.

Reference: IBM Research, “SSM Time Series Model”, 2025. https://research.ibm.com/blog/SSM-time-series-model

Constructors

Constructor Summary
FlowStateOptions() Initializes a new instance with default values.
FlowStateOptions(FlowStateOptions<T> other) Initializes a new instance by copying from another instance.

Properties

Property Summary
FoundationModelSize ModelSize { get; set; } Gets or sets the model size variant.
bool UseDiscretization { get; set; } Gets or sets whether to use discretization for continuous-time SSM.
double DropoutRate { get; set; } Gets or sets the dropout rate.
int ContextLength { get; set; } Gets or sets the context length.
int ForecastHorizon { get; set; } Gets or sets the forecast horizon.
int HiddenDimension { get; set; } Gets or sets the hidden dimension.
int NumLayers { get; set; } Gets or sets the number of SSM layers.
int SSMRank { get; set; } Gets or sets the SSM rank for low-rank parameterization.
int StateDimension { get; set; } Gets or sets the state dimension for the SSM.