MiniTensor Documentation

Welcome to the MiniTensor documentation. This directory contains focused guides for installing, using, extending, benchmarking, and contributing to MiniTensor.

Start here

  • Installation guide – supported installation paths for PyPI, source builds, editable installs, and contributor environments.

  • API reference – consolidated public Python API surface, including top-level helpers, tensors, functional operations, neural-network layers, optimizers, NumPy compatibility, serialization, plugins, and debug utilities.

  • Development guide – repository layout, environment setup, validation commands, documentation workflow, and release checks.

Feature guides

  • Custom operations – how the Rust custom-operation trait, registry, builder, and Python helpers fit together.

  • Plugin system – plugin metadata, compatibility checks, Python plugin registries, dynamic loading, and best practices.

  • Performance benchmarks – benchmark commands, interpretation guidance, profiling tips, and optimization checklist.

Examples and notebooks

The repository also includes executable examples under ../examples and notebooks under ../examples/notebooks. Rebuild the Rust extension before running examples if you changed code in engine/ or bindings/.

Documentation maintenance checklist

When adding or changing public functionality:

  1. Update the API reference for new symbols, parameters, or behavior changes.

  2. Add or update a focused guide when the feature requires more than a short API entry.

  3. Add an executable example or test for important user-facing behavior.

  4. Run the validation commands in the development guide.