LLMTimeOptions<T>

Options & Configuration · AiDotNet.Models.Options

Configuration options for LLM-Time (Zero-Shot Time Series Forecasting via LLM Tokenization).

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

For Beginners

LLM-Time takes a surprising approach:

How It Works:

  1. Convert numbers to text: [1.5, 2.3, 3.1] → “1.500, 2.300, 3.100”
  2. Feed the text to a pretrained LLM (GPT-3, LLaMA)
  3. The LLM predicts the next “tokens” (which are digits of future values)
  4. Parse the generated text back into numbers

Key Advantages:

  • Zero-shot: no training required at all
  • Leverages the vast pattern knowledge of large language models
  • Produces probabilistic forecasts via sampling

Trade-offs:

  • Requires access to a large language model API
  • Limited precision (controlled by decimal places)
  • Slower than specialized time series models

How It Works

LLM-Time converts numeric time series into text strings and uses pretrained LLMs (GPT-3, LLaMA) for zero-shot forecasting by treating the task as next-token prediction on numerical text. No fine-tuning is required—the LLM backbone is frozen.

Reference: Gruver et al., “Large Language Models Are Zero-Shot Time Series Forecasters”, NeurIPS 2023.

Constructors

Constructor Summary
LLMTimeOptions() Initializes a new instance with default values.
LLMTimeOptions(LLMTimeOptions<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.
double DropoutRate { get; set; } Gets or sets the dropout rate.
double Temperature { get; set; } Gets or sets the LLM sampling temperature.
int ContextLength { get; set; } Gets or sets the maximum number of historical time steps for the text prompt.
int ForecastHorizon { get; set; } Gets or sets the number of future time steps to forecast.
int HiddenDimension { get; set; } Gets or sets the hidden dimension of the LLM backbone.
int NumDecimalPlaces { get; set; } Gets or sets the number of decimal places for numeric-to-text conversion.
int NumHeads { get; set; } Gets or sets the number of attention heads.
int NumLayers { get; set; } Gets or sets the number of transformer layers.
int NumSamples { get; set; } Gets or sets the number of samples for probabilistic forecasting.