MI Prometheus documentation

MI Prometheus is an open source Python library, built using PyTorch, that enables reproducible Machine Learning research.

Notes

• Installation
• Updating the documentation
• License

MI-Prometheus Primer

• MI-Prometheus Explained
• Problems Explained
• Models Explained
• Workers Explained
• Grid Workers Explained
• Helpers Explained

Reproducible Research

• Reproducible Research Primer
• Compositional generalization of the MAC and S-MAC models
• Testing working memory using Cog dataset and model
• Training the Sequential MAC model on the Cog dataset

Tutorials

• Basics: LeNet-5 on MNIST

Package Reference

• miprometheus.utils
  • AppState
  • DataDict
  • ParamInterface
  • ParamRegistry
  • SamplerFactory
  • Split Indices
  • StatisticsCollector
  • StatisticsAggregator
  • TimePlot
  • Losses
    • Masked BCEWithLogitsLoss
    • Masked CrossEntropyLoss
  • Problems Utils
    • GenerateFeatureMaps
    • Language
• miprometheus.problems
  • Problem
  • ProblemFactory
  • ImageTextToClass Problems
    • CLEVR
    • Sort-Of-CLEVR
    • ShapeColorQuery
    • VQAMED
  • ImageToClass Problems
    • CIFAR-10
    • MNIST
  • SequenceToSequence Problems
  • VideoTextToClass Problems
    • COG
  • Algorithmic SequenceToSequence Problems
    • Dual Comparison
      • SequenceComparisonCommandLines
      • SequenceEqualityCommandLines
      • SequenceSymmetryCommandLines
    • Dual Distraction
      • DistractionCarry
      • DistractionForget
      • DistractionIgnore
      • ReadingSpan
    • Dual Interruption
      • InterruptionNot
      • InterruptionReverseRecall
      • InterruptionSwapRecall
      • OperationSpan
    • Manipulation Spatial
      • ManipulationSpatialNot
      • ManipulationSpatialRotation
    • Manipulation Temporal
      • ManipulationTemporalRotation
      • RepeatReverseRecallCommandLines
      • ReverseRecallCommandLines
      • SkipRecallCommandLines
    • Recall
      • RepeatSerialRecallCommandLines
      • ScratchPadCommandLines
      • SerialRecallCommandLines
  • TextToText Problems
    • TranslationAnki
  • VideoToClass Problems
    • Permuted Sequential Row Mnist
    • Sequential Pixel MNIST
• miprometheus.models
  • Model
  • SequentialModel
  • ModelFactory
  • Visual Question Answering baselines
    • CNN + LSTM
    • Stacked Attention Networks
    • MAC
    • Simplified MAC
    • Relational Networks
    • COG
  • Image Classification models
  • Controllers for MANNs models
  • Memory-Augmented Neural Network (MANN) models
    • DWM
    • DNC
    • NTM
    • Encoder-Solver models
    • Mental Model prototype
  • Others Models
    • LSTM
    • ThalNet
• miprometheus.workers
  • Worker
  • Trainer
  • OfflineTrainer
  • OnlineTrainer
  • Tester
• miprometheus.grid_workers
  • GridWorker
  • GridTrainerCPU
  • GridTrainerGPU
  • GridTesterCPU
  • GridTesterGPU
  • GridAnalyzer
• miprometheus.helpers
  • IndexSplitter
  • ProblemInitializer

Indices and tables

• Index
• Module Index
• Search Page