Tutorials

Prebuilt Binaries and Source Code

When prompted during installation of NEAT, select SiMa NEAT extras (press Space to toggle), then continue installation.

The extras package includes prebuilt tutorials/tests plus source code, and is automatically unpacked under the NEAT installation folder.

Prebuilt layout:

.
├── lib
│   └── sima-neat
│       ├── tests
│       └── tutorials
└── share
    └── sima-neat
        ├── tests
        └── tutorials

• lib/...: prebuilt binaries
• share/...: source code

Build Tutorials From Source

From the tutorial source directory, use build.sh:

./build.sh --list-targets
./build.sh

Build a single tutorial target:

./build.sh --target tutorial_v2_015_graph_model_hybrid

Use these tutorials in order. Each card links to a chapter with concept-first guidance and matching C++ and Python implementation.

Beginner

Model in 5 Minutes <5 minutes

This tutorial teaches the quickest practical path to run inference with a compiled model in NEAT.

modelmpkinferencefoundations

Async Push Pull 10-15 minutes

This tutorial explains how to use asynchronous APIs to build high-performance production quality applications.

asyncpush-pullthroughputruntime

Session Build And Run 5 minutes

This tutorial introduces `Session`, the runtime composition entry point in NEAT.

sessionbuildrunpipeline

Model Options Chapter 5 minutes

Model options define the runtime contract between your input data, model pipeline stages, and output decoding behavior.

model-optionsconfigurationcontracts

Intermediate

Preproc Chapter 15-20 minutes

Preprocessing is where raw input data is transformed into the exact tensor contract your model expects.

preprocessingnormalizationimage

Postproc Boxdecode 15-20 minutes

Postprocessing converts raw model outputs into actionable detection results (boxes, scores, classes).

postprocessingboxdecodedetection

Session Patterns 15 minutes

This chapter compares practical **session composition patterns** for building real applications.

sessioncompositionpatterns

Numpy Torch Tensor IO 10-15 minutes

This tutorial explains how NEAT tensors map to the data structures most Python and ML users already know: NumPy arrays and Py...

numpypytorchtensorio

Multi Input Samples 15 minutes

This tutorial shows how to send and receive **multi-field samples** (bundle samples) instead of only single-tensor payloads.

multi-inputsamplessync

Output Handling Patterns 10-15 minutes

This tutorial focuses on **output interpretation patterns**: how to reliably inspect what came back from a run and decide wha...

outputpatternssink

Diagnostics In 3 Commands <10 minutes

This tutorial gives you a compact three-step diagnostics workflow you can run before deep debugging.

diagnosticsdebuggingobservability

YOLO Quickstart 15-20 minutes

This tutorial is the fastest path to run a YOLO-style detector in NEAT and confirm the detection pipeline is wired correctly.

yolodetectionmpk

Resnet Quickstart 10-15 minutes

This tutorial is a quick classifier bring-up path using a ResNet-style MPK.

resnetclassificationmpk

Graph Basics 15-20 minutes

This tutorial introduces the NEAT graph runtime using the smallest useful graph: push one sample through connected nodes and...

graphtraversalmetadata

Advanced

Graph Model Hybrid 20-25 minutes

This tutorial shows how to combine graph orchestration with model execution in one hybrid flow.

graphhybridstage-modelmpk

Graph Multistream 20-25 minutes

This tutorial teaches multistream graph scheduling fundamentals: how multiple stream/frame inputs are fairly scheduled, branc...

graphmultistreamschedulerjoin

Performance Tuning 15-20 minutes

This tutorial explains the first performance knobs most teams tune in async pipelines: queue depth, overflow policy, and metr...

performancetuningasyncqueues

Production Blueprint 20-25 minutes

This tutorial assembles a practical production blueprint from patterns introduced earlier: explicit run options, controlled a...

productionreliabilitydeployment