Module neural_forecasting 

Neural Forecasting Models for Time Series

This module provides cutting-edge implementations for neural network-based time series forecasting, including LSTM, GRU, Transformer, Mamba/State Space Models, Temporal Fusion Transformers, and Mixture of Experts architectures. These implementations focus on core algorithmic components and can be extended with actual neural network frameworks.

Advanced Architectures

• LSTM Networks: Long Short-Term Memory networks for sequence modeling
• Transformer Models: Self-attention based architectures
• N-BEATS: Neural basis expansion analysis for time series forecasting
• Mamba/State Space Models: Linear complexity for long sequences with selective state spaces
• Flash Attention: Memory-efficient attention computation for transformers
• Temporal Fusion Transformers: Specialized architecture for time series forecasting
• Mixture of Experts: Conditional computation for model scaling

Re-exports

pub use self::attention::*;

pub use self::config::*;

pub use self::lstm::*;

pub use self::mamba::*;

pub use self::mixture_of_experts::*;

pub use self::nbeats::*;

pub use self::temporal_fusion::*;

pub use self::transformer::*;

Modules

attention

Advanced Attention Mechanisms

config

Configuration and Common Types for Neural Forecasting

lstm

LSTM Network Components for Time Series Forecasting

mamba

Mamba/State Space Models for Time Series

mixture_of_experts

Mixture of Experts for Conditional Computation

nbeats

N-BEATS Neural Basis Expansion Analysis for Time Series

temporal_fusion

Temporal Fusion Transformer Components

transformer

Transformer Networks for Time Series Forecasting