NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #2819 most downloaded on PyPI
Efficient in-memory representation for ONNX
Last release 1 months ago
11 Aug 2026
Release timing varies
gaps range from 2 weeks to 4 months
Nearly every release is documented
notes for 20 of 20 stable releases
1 version withdrawn
withdrawn after publishing
2 years old
21 releases · first in 2025
One column per month.
Note This is the first stable release of onnx-ir . Thank you to all contributors to this project! 🎉
Note
This is the first stable release of onnx-ir. Thank you to all contributors to this project! 🎉
Full Changelog: v0.2.1...v1.0.0
Optional MetadataStore deep copy when cloning by @AyoubMDL in #354
Full Changelog: v0.2.0...v0.2.1
Note This release adds sympy as a dependency of the library.
Note
This release adds sympy as a dependency of the library.
Full Changelog: v0.1.16...v0.2.0
Create the journaling system by @justinchuby in #322
Full Changelog: v0.1.15...v0.1.16
Support saving external data as safetensors by @justinchuby in #306
Full Changelog: v0.1.14...v0.1.15
Implement graph cloning methods by @justinchuby in #286
Value.merge_shapes() by @justinchuby in #290Full Changelog: v0.1.13...v0.1.14_1
Note This release brings support for the newly added 2-bit data types in onnx 1.20.
Note
This release brings support for the newly added 2-bit data types in onnx 1.20.
[...] in text for rich display by @justinchuby in #243Tape by @justinchuby in #266Full Changelog: v0.1.12...v0.1.13
This update introduces the Value.replace_all_uses_with method, added for its convenience and discoverability. A new replace_graph_outputs option has b
This update introduces the Value.replace_all_uses_with method, added for its convenience and discoverability. A new replace_graph_outputs option has been included, allowing users to replace graph outputs as well.
Note
Users are still responsible for assigning the original output name to the replacement value if they want to preserve the graph's signature.
Full Changelog: v0.1.11...v0.1.12
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
This release introduces a new API tofile() on Tensors. It helps avoid unnecessary copies with tobytes() when a model is saved with external data. The API is introduced in a backward-compatible manner. See https://github.com/onnx/ir-py/pull/210 for more information.
Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.10...v0.1.11
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.9...v0.1.10
Deprecate ir.Input in favor of ir.val by @justinchuby in https://github.com/onnx/ir-py/pull/185
<!-- Release notes generated using configuration in .github/release.yml at main -->
ir.Input in favor of ir.val by @justinchuby in https://github.com/onnx/ir-py/pull/185__str__ to use repr() by @justinchuby in https://github.com/onnx/ir-py/pull/190Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.8...v0.1.9
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
[!NOTE] This patch release includes a fix in the protobuf serialization logic to handle Boolean attributes for compatibility with protobuf>=6.
Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.7...v0.1.8
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
meta field to attributes for metadata storage by @Copilot in https://github.com/onnx/ir-py/pull/154ir.Function by @justinchuby in https://github.com/onnx/ir-py/pull/158subgraph() by @justinchuby in https://github.com/onnx/ir-py/pull/156Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.6...v0.1.7
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
[!NOTE] This patch release includes a fix for compatibility with PyTorch <2.7.0 in the
tensor_adaptersmodule. It is recommended that you update to this version if you leverageonnx_ir.tensor_adaptersto interact with torch tensors.
Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.5...v0.1.6
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.4...v0.1.5
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
[!NOTE] This release includes a fix to the 4 bit dtype handling logic in the
TensorProtoTensor.numpy()method. You should update to this version if you use onnx-ir to process 4 bit tensor data (uint4, int4, float4e2m1) numerically with NumPy.
Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.3...v0.1.4
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
Tensor by @justinchuby in https://github.com/onnx/ir-py/pull/102Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.2...v0.1.3
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.1...v0.1.2
<!-- Release notes generated using configuration in .github/release.yml at main -->
<!-- Release notes generated using configuration in .github/release.yml at main -->
add methods on initializers and attributes by @justinchuby in https://github.com/onnx/ir-py/pull/33Full Changelog: https://github.com/onnx/ir-py/compare/v0.1.0...v0.1.1
First official release of the onnx_ir package.
First official release of the onnx_ir package.
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →