KTT Hierarchical Classification System

Contents:

  • Usage
    • Installation
      • Downloading
      • Setting up dependencies
    • Quickstart
      • Data preparation
      • Training a model
      • Exporting the trained model
      • Serving up a Bento
      • Shipping Bentos in a container
    • CLI usage
      • Adapters
  • System design
    • Data adapters
      • Intermediate format specification
        • Parquet schema
        • Hierarchy JSON schema
      • Theory
        • The SQL adapter
        • The flatfile adapter
    • Training stage
      • The process
      • Classes
        • Common classes
        • PyTorch utilities
        • Scikit-learn utilities
    • Exporting your models
      • ONNX exporting
      • BentoML exporting
        • Packaging
  • Encoders
    • DistilBERT
      • API
    • Scikit-learn text feature extractors
      • API
  • Prebuilt models
    • Tf-idf + Leaf SGD
      • API
      • Configuration schema
      • Default tuning configuration
      • Theory
    • Tf-idf + Hierarchy SGD
      • API
      • Configuration schema
      • Default tuning configuration
      • Theory
    • DB-BHCN
      • API
      • Configuration schema
      • Default tuning configuration
      • Checkpoint schema
      • Theory
        • DB-BHCN
    • DB-BHCN+AWX
      • API
      • Configuration schema
      • Default tuning configuration
      • Checkpoint schema
      • Theory
    • DistilBERT + Adapted HMCN-F
      • API
      • Configuration schema
      • Default tuning configuration
      • Checkpoint schema
      • Theory
    • DistilBERT + Adapted C-HMCNN
      • API
      • Configuration schema
      • Default tuning configuration
      • Checkpoint schema
      • Theory
    • DistilBERT + Linear
      • API
      • Configuration schema
      • Checkpoint schema
      • Theory
  • Developing new encoders
    • Where encoders come in
    • Adding encoders
    • Implementing preprocessors
  • Developing new models
    • Frameworks
    • General model folder structure
    • The model itself
    • Checkpointing
    • Preprocessing needs
    • Exporting
      • ONNX
      • export_bento_resources
        • The service implementation
        • The service configuration files
        • The reference dataset
        • The export_bento_resources method
    • Specifying your hyperparameters (optional)
    • Registering your model with the rest of the system
      • The model lists
    • Test-run your model
      • Grafana dashboard design (optional)
      • Testing automatic dashboard provisioning
    • Framework-specific guides
      • Implementing a model with PyTorch+DistilBERT
        • The model
        • PyTorch model module structure
        • PyTorch utilities
        • The process
        • Registering, testing & conclusion
      • Implementing a model with Scikit-learn
        • The model
        • Scikit-learn utilities
        • The process
        • Registering, testing & conclusion
  • Advanced guides
    • Using DVC with our system
    • Inferencing with GPUs
      • Prerequisites
      • GPU-based inference using Bentos
      • GPU-based inference for Dockerised services
        • Without monitoring capabilities
        • With monitoring capabilities
    • Automatic hyperparameter tuning
      • CLI usage
      • Tune configuration format
  • References
KTT Hierarchical Classification System
  • »
  • Advanced guides
  • View page source

Advanced guides

Here are some topics that are optional to daily usage, but may provide significant benefits if used.

  • Using DVC with our system
  • Inferencing with GPUs
    • Prerequisites
    • GPU-based inference using Bentos
    • GPU-based inference for Dockerised services
      • Without monitoring capabilities
      • With monitoring capabilities
  • Automatic hyperparameter tuning
    • CLI usage
    • Tune configuration format
Previous Next

© Copyright 2022, Huynh Thien Khiem, Voong Xay Tac, Huynh Truong Tu.

Built with Sphinx using a theme provided by Read the Docs.