Skip to content

Files

Latest commit

 

History

History
57 lines (43 loc) · 1.95 KB

index.md

File metadata and controls

57 lines (43 loc) · 1.95 KB

API Reference

This section contains detailed API documentation for Ember ML.

Core Modules

  • ember_ml.ops: Core operations for tensor manipulation

    • tensor: Tensor creation and manipulation
    • math: Mathematical operations
    • random: Random number generation
    • solver: Linear algebra operations

  • ember_ml.backend: Backend abstraction system

    • numpy: NumPy backend implementation
    • torch: PyTorch backend implementation
    • mlx: MLX backend implementation
    • ember: Ember backend implementation

  • ember_ml.features: Feature extraction and processing

    • TerabyteFeatureExtractor: Extracts features from large datasets
    • TemporalStrideProcessor: Processes temporal data with variable strides
    • GenericFeatureEngineer: General-purpose feature engineering
    • GenericTypeDetector: Automatic column type detection

  • ember_ml.models: Machine learning models

    • liquid: Liquid neural networks
    • rbm: Restricted Boltzmann Machines

Neural Network Components

  • ember_ml.core: Core neural implementations

    • ltc: Liquid Time Constant neurons
    • geometric: Geometric processing
    • spherical_ltc: Spherical variants
    • stride_aware_cfc: Stride-Aware Continuous-time Fully Connected cells

  • ember_ml.attention: Attention mechanisms

    • temporal: Time-based attention
    • causal: Causal attention
    • multiscale_ltc: Multiscale LTC attention

  • ember_ml.nn: Neural network components

    • wirings: Network connectivity patterns
    • modules: Neural network modules

Utility Modules

  • ember_ml.utils: Utility functions

    • math_helpers: Mathematical utilities
    • metrics: Evaluation metrics
    • visualization: Plotting tools

  • ember_ml.initializers: Weight initialization

    • glorot: Glorot/Xavier initialization
    • binomial: Binomial initialization

For detailed documentation on specific functions and classes, refer to the docstrings in the source code.