Crate onnx_std 

onnx-std

A pure-Rust ONNX standard library — model I/O, textual format, and an extensible validator — built on the shared runtime IR. This is the first wave of the design captured in docs/ONNX_RS.md.

Relationship to the rest of the workspace

onnx-std does not define its own IR. Per ONNX_RS §4.1 (“Shared Crate”), the Graph / Node / Value / Tensor / WeightRef types are reused verbatim from onnx_runtime_ir, and the protobuf parse/encode stack is reused from onnx_runtime_loader. onnx-std is the standard-library layer on top: ergonomic model I/O, a human-readable text format, and a checker — the concerns the design assigns to the ONNX library rather than the inference runtime.

onnx-std  ──(standard lib: load/save/print/check)
   │
   └─ depends on ─▶ onnx-runtime-ir      (shared Graph/Node/Value IR)
   └─ depends on ─▶ onnx-runtime-loader  (protobuf serde + weight mmap)

What this wave covers

ONNX_RS §	Feature	Status
§3.4 / §4	load_model / save_model round-trip	✅
§5	Model::to_text / Model::from_text	✅
§7	schema op schemas and opset-aware registry	✅
§8	check::OnnxChecker extensible validator	✅ (schema-aware)
§9	shape::infer_shapes symbolic shape inference	✅
§10	version::VersionConverter opset conversion	✅
§6.2	textproto protobuf TextFormat I/O	✅

JSON and protobuf TextFormat are descriptor-driven from the exact vendored ONNX proto, so every bound message, field, oneof, and enum is covered. The readable text DSL appends a protobuf-TextFormat extension containing only fields outside the graph syntax. The parsed DSL is authoritative for every field it represents, while the explicit extension preserves the remainder.

Deferred to later waves (see // FOLLOW-UP markers): the complete operator catalog, remaining function/IR-gate checker rules, custom-op registration (§11), and Python bindings (§12). Training semantics are out of scope.

Example

let model = onnx_std::load_model("model.onnx")?;
println!("{}", model.to_text());
let report = model.validate();
assert!(report.is_valid());
onnx_std::save_model(&model, "roundtrip.onnx")?;

Re-exports

pub use check::OnnxChecker;

pub use check::Severity;

pub use check::ValidationResult;

pub use check::ValidationRule;

pub use check::Violation;

pub use schema::AttributeDefault;

pub use schema::AttributeSpec;

pub use schema::AttributeType;

pub use schema::InputSpec;

pub use schema::OpSchema;

pub use schema::OutputSpec;

pub use schema::SchemaError;

pub use schema::SchemaRegistry;

pub use schema::TypeConstraint;

pub use shape::ShapeInferenceResult;

pub use shape::infer_shapes;

pub use text::PrintOptions;

pub use text::from_text;

pub use text::to_text;

pub use text::to_text_with;

pub use version::AdaptResult;

pub use version::ConvertError;

pub use version::ConvertReport;

pub use version::IncompatibleOp;

pub use version::OpAdapter;

pub use version::VersionConverter;

pub use onnx_runtime_ir as ir;

Modules

check

Extensible model checker / validator (ONNX_RS §8).

json

Canonical descriptor-driven ONNX protobuf JSON interchange.

schema

Declarative operator schemas and an opset-aware registry (ONNX_RS §7).

shape

Symbolic shape inference over the shared runtime IR (ONNX_RS §9).

simple_sharded_dim_proto

Generated ONNX IR v13 multi-device/sharding protobuf model types.

text

Textual representation of ONNX models (ONNX_RS §5).

textproto

Descriptor-driven ONNX protobuf TextFormat (.onnxtxt / .pbtxt).

version

Opset version conversion with composable per-operator adapters (ONNX_RS §10).

Structs

DeviceConfigurationProto

Generated ONNX IR v13 multi-device/sharding protobuf model types.

IntIntListEntryProto

Generated ONNX IR v13 multi-device/sharding protobuf model types.

Json

Canonical protobuf JSON mapping (ONNX_RS §6).

Model

An owned ONNX model: the shared-IR Graph plus the model-level metadata and (optionally) the live weight store backing external initializers.

ModelMetadata

Model-level metadata that the IR Graph does not itself carry.

NodeDeviceConfigurationProto

Generated ONNX IR v13 multi-device/sharding protobuf model types.

ShardedDimProto

Generated ONNX IR v13 multi-device/sharding protobuf model types.

ShardingSpecProto

Generated ONNX IR v13 multi-device/sharding protobuf model types.

SimpleShardedDimProto

Generated ONNX IR v13 multi-device/sharding protobuf model types.

Text

The human-readable ONNX text DSL (ONNX_RS §5).

TextProto

Protobuf TextFormat (.onnxtxt / .pbtxt) (ONNX_RS §6).

WeightStore

A resolved set of initializer weights, keyed by the value they populate, plus the live memory maps backing any external data files.

Enums

Error

An error produced by an onnx-std operation.

ShapeError

An error produced while inferring shapes.

Traits

TextCodec

A textual serialization format for Model.

Functions

load_model

Load an ONNX model from a .onnx protobuf file (ONNX_RS §3.4).

save_model

Save a Model to a .onnx protobuf file (ONNX_RS §3.4).

Type Aliases

OpaqueProto

ONNX-ML opaque type payload (TypeProto.opaque_type, field 7).

Result

The result type used throughout the onnx-std public API.