package caisar

  1. Overview
  2. Docs
package caisar
Legend:
Page
Library
Module
Module type
Parameter
Class
Class type
Source

Module Onnx.Version

Versioning

ONNX versioning is specified in docs/IR.md and elaborated on in docs/Versioning.md

To be compatible with both proto2 and proto3, we will use a version number that is not defined by the default value but an explicit enum number.

Overview

ONNX is an open specification that is comprised of the following components:

  1. A definition of an extensible computation graph model.
  2. Definitions of standard data types.
  3. Definitions of built-in operators.

This document describes the syntax of models and their computation graphs, as well as the standard data types. Together, they are referred to as the ONNX Intermediate Representation, or 'IR' for short.

The normative semantic specification of the ONNX IR is found in docs/IR.md. Definitions of the built-in neural network operators may be found in docs/Operators.md.

Notes

Protobuf compatibility

To simplify framework compatibility, ONNX is defined using the subset of protobuf that is compatible with both protobuf v2 and v3. This means that we do not use any protobuf features that are only available in one of the two versions.

Here are the most notable contortions we have to carry out to work around these limitations:

  • No 'map' (added protobuf 3.0). We instead represent mappings as lists of key-value pairs, where order does not matter and duplicates are not allowed.
Sourcetype t =
  1. | P_START_VERSION
    (*

    proto3 requires the first enum value to be zero. We add this just to appease the compiler.

    *)
  2. | IR_VERSION_2017_10_10
    (*

    The version field is always serialized and we will use it to store the version that the graph is generated from. This helps us set up version control. For the IR, we are using simple numbers starting with 0x00000001, which was the version we published on Oct 10, 2017.

    *)
  3. | IR_VERSION_2017_10_30
    (*

    IR_VERSION 2 published on Oct 30, 2017

    • Added type discriminator to AttributeProto to support proto3 users
    *)
  4. | IR_VERSION_2017_11_3
    (*

    IR VERSION 3 published on Nov 3, 2017

    • For operator versioning:
      • Added new message OperatorSetIdProto
      • Added opset_import in ModelProto
    • For vendor extensions, added domain in NodeProto
    *)
  5. | IR_VERSION_2019_1_22
    (*

    IR VERSION 4 published on Jan 22, 2019

    • Relax constraint that initializers should be a subset of graph inputs
    • Add type BFLOAT16
    *)
  6. | IR_VERSION_2019_3_18
    (*

    IR VERSION 5 published on March 18, 2019

    • Add message TensorAnnotation.
    • Add quantization annotation in GraphProto to map tensor with its scale and zero point quantization parameters.
    *)
  7. | IR_VERSION_2019_9_19
    (*

    IR VERSION 6 published on Sep 19, 2019

    • Add support for sparse tensor constants stored in model.
      • Add message SparseTensorProto
      • Add sparse initializers
    *)
  8. | IR_VERSION_2020_5_8
    (*

    IR VERSION 7 published on May 8, 2020

    • Add support to allow function body graph to rely on multiple external opreator sets.
    • Add a list to promote inference graph's initializers to global and mutable variables. Global variables are visible in all graphs of the stored models.
    • Add message TrainingInfoProto to store initialization method and training algorithm. The execution of TrainingInfoProto can modify the values of mutable variables.
    • Implicitly add inference graph into each TrainingInfoProto's algorithm.
    *)
  9. | IR_VERSION
    (*

    IR VERSION 8 published on Introduce TypeProto.SparseTensor Introduce TypeProto.Optional Added a list of FunctionProtos local to the model Deprecated since_version and operator status from FunctionProto

    *)
Sourceval name : unit -> string

Fully qualified protobuf name of this enum