TOSA Related Software

TOSA related software projects

Description of the TOSA related software pieces on MLPlatform

TOSA Reference Model & Test Generator

In addition to the TOSA specification document, there is also a reference model written as C code which implements the exact behaviour as set out in the specification. This reference implementation is intended to be the “golden reference” for other TOSA implementations to be compared against and incorporated into regression test suites and such. The reference model consumes a FlatBuffers serialization of the network subgraph generated by the TOSA Serialization Library, along with input tensors for placeholder nodes in NumPy format. By default, the model validates and evaluates the network subgraph, and writes out the resulting output tensors in NumPy format. Alongside the TOSA Reference Model there is also a comprehensive set of tools for generating unit tests which exercise the behaviour of the TOSA operators. The tests are created as TOSA operations serialized in the flatbuffer format with inputs created in files using the NumPy format, as used by the reference model.

For details on how to build and use the TOSA Reference Model, review the in the root of that repo.

TOSA MLIR Translator

The TOSA MLIR translator repository contains an MLIR pass which can serialize MLIR TOSA dialect content into a TOSA flatbuffer file. For details on how to build and use the TOSA MLIR translator, see the in the root of that repository.

TOSA Checker

The TOSA Checker is a tool that provides with an easy way to ensure that a neural network model is compatible with the TOSA specification. The checker validates if the model can be expressed using exclusively TOSA primitive operators. It currently supports TensorFlow™ Lite models, support for additional model types will be added in the future.

The Git repository containing the TOSA Checker is here:

Python packages are available on PyPi:

Repository URL Description
Conformance Test Suite Conformance tests for the Base Inference profile
Reference Model Reference implementation and testing infrastructure for TOSA
Serialization Library Methods to read and write serialized TOSA graphs using a flatbuffer schema
TOSA Checker Checks if a TensorFlow Lite model is compatible with the TOSA specification
TOSA MLIR Translator Uses the serialization library to convert a TOSA MLIR graph to a serialized flatbuffer