Package com.codedstream.otterstream.inference.model

Interface ModelLoader<T>

  • Type Parameters:
    T - the type of loaded model (e.g., OrtSession, SavedModelBundle)
    All Known Implementing Classes:
    OnnxModelLoader
    public interface ModelLoader<T>
    Interface for loading ML models from various sources.

    Implementations handle the specifics of loading different model formats (ONNX, TensorFlow, PyTorch, etc.) from various storage systems (file system, HDFS, S3, HTTP, etc.).

    Responsibilities:

    • Load models from configured paths
    • Validate model format and integrity
    • Extract model metadata
    • Handle different storage backends

    Implementation Example:

    {@code
 public class OnnxModelLoader implements ModelLoader {
    Since:
    1.0.0
    Author:
    Nestor Martourez, Sr Software and Data Streaming Engineer @ CodedStreams
    See Also:
    ModelConfig, ModelMetadata

    • Method Detail

      • loadModel

        T loadModel​(ModelConfig config)
     throws ModelLoadException
        Loads a model from the path specified in configuration.
        Parameters:
        config - model configuration containing path and options
        Returns:
        loaded model instance
        Throws:
        ModelLoadException - if loading fails

      • loadModel

        T loadModel​(InputStream inputStream,
            ModelConfig config)
     throws ModelLoadException
        Loads a model from an input stream.

        Useful for loading models from non-file sources.

        Parameters:
        inputStream - stream containing model data
        config - model configuration
        Returns:
        loaded model instance
        Throws:
        ModelLoadException - if loading fails

      • validateModel

        boolean validateModel​(T model,
                      ModelConfig config)
        Validates that a loaded model matches the configuration.
        Parameters:
        model - the loaded model
        config - model configuration
        Returns:
        true if model is valid

      • getSupportedFormats

        ModelFormat[] getSupportedFormats()
        Gets the model formats supported by this loader.
        Returns:
        array of supported formats

      • getModelMetadata

        ModelMetadata getModelMetadata​(T model)
        Extracts metadata from a loaded model.
        Parameters:
        model - the loaded model
        Returns:
        model metadata including inputs, outputs, and format