PyPI · #2770 most downloaded on PyPI

onnxscript

0.7.22.1M downloads/momicrosoft/onnxscript ↗New releases in your daily digest, changelog included.

Naturally author ONNX functions and models using a subset of Python

Last release 18 days ago
16 Sep 2026
Ships on a steady schedule
a new release about every 9 days
Nearly every release is documented
notes for 25 of 26 stable releases
Nothing withdrawn
no release was ever pulled
3 years old
722 releases · first in 2023

Release timeline

722 releases since 2023

2023202420252026

20232026

One column per quarter.

Releases

0.5.5.dev2025101717 Oct 2025pre-release
Nothing published for this version
0.5.416 Oct 2025Deprecated
[torchlib] Deprecate Rank and IsScalar by Justin Chu (@justinchuby) in #2624
What's Changed

Optimizer and rewriter

Fix constant in constant folding by Ti-Tai Wang (@titaiwangms) in #2622

Create helper for comparing semantic equivalence of shapes by Justin Chu (@justinchuby) in #2620

Fix GQA fusion to produce present key/value by Justin Chu (@justinchuby) in #2634

Torch Lib

Separated implementation of aten::scatter overloads by deoxy (@linshokaku) in #2605

Enhanced type annotations and simplified implementation of scatter.value by deoxy (@linshokaku) in #2612

support for scalar args to aten::scatter by deoxy (@linshokaku) in #2613

[torchlib] Implement aten_bilinear function using Einsum by @Copilot in #2574

Simplify aten_unbind when shape is static by Justin Chu (@justinchuby) in #2597

Consolidate overloads in torchlib by Justin Chu (@justinchuby) in #2604

[torchlib] Fix implementations for bitwise_* overloads by Justin Chu (@justinchuby) in #2618

[torchlib] Deprecate Rank and IsScalar by Justin Chu (@justinchuby) in #2624

[torchlib] Fix operator add by Justin Chu (@justinchuby) in #2630

Remove redundant registration of operator::add and fix sub.Scalar by Justin Chu (@justinchuby) in #2631

Other Changes

Update torch api error message to include value names by Justin Chu (@justinchuby) in #2599

Remove beartype by Justin Chu (@justinchuby) in #2603

Allow opset_version to be set explicitly when exporting by NoRaincheck in #2615

Merge shapes only in identity op and nodel-level shape inference by Ti-Tai Wang (@titaiwangms) in #2623

Full Changelog: v0.5.3...v0.5.4

0.5.31 Oct 2025

Fix Onnx 23 Rotary Fusion by G. Ramalingam (@gramalingam) in #2576

What's Changed

Optimizer and rewriter

Fix Onnx 23 Rotary Fusion by G. Ramalingam (@gramalingam) in #2576
Record names of contributing values in the constant folding pass by Justin Chu (@justinchuby) in #2575
Merge output shape with input shape instead of override by Daniel Zhang (@wodesuck) in #2578
Extend utilities for checking a scalar value by G. Ramalingam (@gramalingam) in #2587
Merge input and output shape when removing identity by Daniel Zhang (@wodesuck) in #2588
Add NaN handling in softmax pattern in SDPA fusion by G. Ramalingam (@gramalingam) in #2593
Fix collapse slices rewrite rules to handle unknown dims by Justin Chu (@justinchuby) in #2583
Expose the should_fold option to optimize() by Justin Chu (@justinchuby) in #2594

Torch Lib

[torchlib] Add trace_only flag to aten_copy, aten_tril, aten_triu by Justin Chu (@justinchuby) in #2572
[torchlib] Support integers in logical_and/or ops and update other logical ops by Justin Chu (@justinchuby) in #2582
[torchlib] Add back operator and/or by Justin Chu (@justinchuby) in #2590
Improve aten_floor_divide for int inputs by Justin Chu (@justinchuby) in #2592

Other Changes

Remove usages of ir.Input in test by Justin Chu (@justinchuby) in #2591

New Contributors

Daniel Zhang (@wodesuck) made their first contribution in #2578

Full Changelog: v0.5.2...v0.5.3

Open source →

0.5.223 Sept 2025

What's Changed

Optimizer and rewriter

[rewriter] Remove generic pattern matcher by @justinchuby in https://github.com/microsoft/onnxscript/pull/2567
Add GQA fusion to ONNX fusions by @gramalingam in https://github.com/microsoft/onnxscript/pull/2524

Torch Lib

[torchlib] Fix aten_gather to correctly handle scalar indices by @linshokaku in https://github.com/microsoft/onnxscript/pull/2566
[torchlib] Simplify linalg_vector_norm to remove the redundant Abs by @justinchuby in https://github.com/microsoft/onnxscript/pull/2570

New Contributors

@linshokaku made their first contribution in https://github.com/microsoft/onnxscript/pull/2566

Full Changelog: https://github.com/microsoft/onnxscript/compare/v0.5.1...v0.5.2

Open source →

0.5.012 Sept 2025

What's Changed

Breaking Changes

Refactor rewrite rules into the rewriter.rules namespace by @justinchuby in https://github.com/microsoft/onnxscript/pull/2531
Update constant folding behavior for large tensors by @justinchuby in https://github.com/microsoft/onnxscript/pull/2488

Optimizer and rewriter

[Rewriter] Prevent out of range when matching node outputs by @iksnagreb in https://github.com/microsoft/onnxscript/pull/2508
Add RMS Normalization variant by @gramalingam in https://github.com/microsoft/onnxscript/pull/2519
[Optimizer] Fix reinterpretation of strings in _get_numpy_value by @iksnagreb in https://github.com/microsoft/onnxscript/pull/2514
Improve symbolic dim tracking by @gramalingam in https://github.com/microsoft/onnxscript/pull/2520
Remove function extraction in ONNX rotary embedding by @gramalingam in https://github.com/microsoft/onnxscript/pull/2525
[Optimizer] Avoid accessing None value in _process_constant_node by @iksnagreb in https://github.com/microsoft/onnxscript/pull/2513
Increase DEFAULT_CONSTANT_FOLD_INPUT_SIZE_LIMIT by @justinchuby in https://github.com/microsoft/onnxscript/pull/2527
[Rewriter(matmul_add_to_gemm)]: check shapes by @AyoubMDL in https://github.com/microsoft/onnxscript/pull/2528
Add ort-specific passes to ort_fusion by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2532
[Reland] Update SplitToSequence in constant folding by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2544
[Rewriter]: add fusion rules for successive Min/Max patterns by @AyoubMDL in https://github.com/microsoft/onnxscript/pull/2500
[rewriter] Unify reshape flatten ops by @Johansmm in https://github.com/microsoft/onnxscript/pull/2518
Do not try to fold op.SplitToSequence when split is None by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2550
[Rewriter] Add fuse batchnorm to default rules by @AyoubMDL in https://github.com/microsoft/onnxscript/pull/2553
Add Conv-Affine(Mul+Add) and hardswish fusion by @Stonesjtu in https://github.com/microsoft/onnxscript/pull/2472
Fix rewriter and CI tests for the latest onnx-ir version by @justinchuby in https://github.com/microsoft/onnxscript/pull/2554

Torch Lib

[torchlib] Modify aten_unbind to use None for split_sizes by @justinchuby in https://github.com/microsoft/onnxscript/pull/2536
[torchlib] Improve pixel_shuffle by @justinchuby in https://github.com/microsoft/onnxscript/pull/2537
[torchlib] Improve handling of SymInt[] by @justinchuby in https://github.com/microsoft/onnxscript/pull/2522
[torch] Fix incorrect Concat when processing dynamic paddings by @kistenklaus in https://github.com/microsoft/onnxscript/pull/2540
Add test for dynamic padding by @justinchuby in https://github.com/microsoft/onnxscript/pull/2541
[torchlib] Mark atan2 as trace_only and map NaN to 0 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2557
[torchlib] Fix repeat_interleave when repeats is a symbolic tensor by @xadupre in https://github.com/microsoft/onnxscript/pull/2548
Support enable_gqa and only support 4D Q, K, and V by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2558
[torchlib] Implement torch.ops.prims.broadcast_in_dim.default by @Copilot in https://github.com/microsoft/onnxscript/pull/2382

Other Changes

chore(deps): bump actions/upload-pages-artifact from 3 to 4 by @dependabot[bot] in https://github.com/microsoft/onnxscript/pull/2517
Generate opset24 ops by @justinchuby in https://github.com/microsoft/onnxscript/pull/2523
Disable unstable tests by @xadupre in https://github.com/microsoft/onnxscript/pull/2512
Bump version to 0.4.1 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2529
More robust checks for FLOAT8E8M0 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2530
chore(deps): bump ruff from 0.12.10 to 0.12.11 in /requirements/lintrunner by @dependabot[bot] in https://github.com/microsoft/onnxscript/pull/2535
Implements repeat_interleave by @xadupre in https://github.com/microsoft/onnxscript/pull/2477
Update onnx-weekly version to 1.20.0 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2545
[Rewriter] Support specifying node name in rewrites by @AyoubMDL in https://github.com/microsoft/onnxscript/pull/2474
Bump version to 0.5.0 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2538

New Contributors

@iksnagreb made their first contribution in https://github.com/microsoft/onnxscript/pull/2508
@kistenklaus made their first contribution in https://github.com/microsoft/onnxscript/pull/2540
@Stonesjtu made their first contribution in https://github.com/microsoft/onnxscript/pull/2472

Full Changelog: https://github.com/microsoft/onnxscript/compare/v0.4.0...v0.5.0

Open source →

0.4.022 Aug 2025

What's Changed

Breaking Changes

Remove torchscript graph builder by @justinchuby in https://github.com/microsoft/onnxscript/pull/2444

New Features

[torch api] Create torch API for PyTorch 2.9 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2503

Optimizer and rewriter

Expose match functionality of rewrite-rule by extracting base classes by @Copilot in https://github.com/microsoft/onnxscript/pull/2447
Implement MatchContext class for rewriter pattern matching by @Copilot in https://github.com/microsoft/onnxscript/pull/2455
[Rewriter] Implement value/node level checkers for pattern matching infrastructure by @Copilot in https://github.com/microsoft/onnxscript/pull/2459
fix: handling of default attrs in SimplifiedLayerNormalization + LayerNormalization🐛 by @KarelZe in https://github.com/microsoft/onnxscript/pull/2396
Add initial support for RotaryEmbedding fusion for onnx opset 23 by @gramalingam in https://github.com/microsoft/onnxscript/pull/2450
Attention mask for GQA fusion by @gramalingam in https://github.com/microsoft/onnxscript/pull/2452
Handle matching against None explicitly by @gramalingam in https://github.com/microsoft/onnxscript/pull/2460
[docs] Document rewriter pattern options by @Copilot in https://github.com/microsoft/onnxscript/pull/2406
Rename fusion files by @gramalingam in https://github.com/microsoft/onnxscript/pull/2476
[Rewriter] Add optimizer to fold Pad operators into Conv by @Johansmm in https://github.com/microsoft/onnxscript/pull/2363
MHA fusion cleanup by @gramalingam in https://github.com/microsoft/onnxscript/pull/2481
Remove double transpose flag in MHA fusion by @gramalingam in https://github.com/microsoft/onnxscript/pull/2483
Add reproduction test case for incorrect slice rewrite and add potential fix by @sebimarkgraf in https://github.com/microsoft/onnxscript/pull/2478
[ort_fusuion] Support fp16 in rms_norm fusion by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2491
Introduce layer-norm fusion by @gramalingam in https://github.com/microsoft/onnxscript/pull/2492
Correctly create empty ints for Constant in rewriter by @justinchuby in https://github.com/microsoft/onnxscript/pull/2497
Add Erf-based Gelu fusion rule by @gramalingam in https://github.com/microsoft/onnxscript/pull/2495
Extend rewriter to handle subgraphs by @gramalingam in https://github.com/microsoft/onnxscript/pull/2494
Minor fix for getting function's graph by @gramalingam in https://github.com/microsoft/onnxscript/pull/2504
Fixes for when attr type can be ambiguous for empty lists by @justinchuby in https://github.com/microsoft/onnxscript/pull/2505

Torch Lib

[torchlib] Improve aten_chunk conversion by @justinchuby in https://github.com/microsoft/onnxscript/pull/2469
Make onnx export SDPA match aten behavior by @IlyasMoutawwakil in https://github.com/microsoft/onnxscript/pull/2479
Add condition to dropout and ref to isnan by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2482
Support aten::scatter.src by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2490

Other Changes

Update VERSION to 0.4.0 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2449
Remove legacy_ir usage in testutil by @justinchuby in https://github.com/microsoft/onnxscript/pull/2451
Remove legacy ir by @justinchuby in https://github.com/microsoft/onnxscript/pull/2456
Remove DORT related tests since it was removed from PyTorch by @justinchuby in https://github.com/microsoft/onnxscript/pull/2465
Update requirements-ort-nightly.txt by @justinchuby in https://github.com/microsoft/onnxscript/pull/2471
Fix logic for converting np array to text by @justinchuby in https://github.com/microsoft/onnxscript/pull/2470
Add a test for boolean attention mask within SDPA by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2480
chore(deps): bump actions/checkout from 4 to 5 by @dependabot[bot] in https://github.com/microsoft/onnxscript/pull/2484
Use onnx-ir 0.1.7 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2509
Minor fixes to onnx to onnxscript converter by @gramalingam in https://github.com/microsoft/onnxscript/pull/2510

New Contributors

@IlyasMoutawwakil made their first contribution in https://github.com/microsoft/onnxscript/pull/2479
@sebimarkgraf made their first contribution in https://github.com/microsoft/onnxscript/pull/2478

Full Changelog: https://github.com/microsoft/onnxscript/compare/v0.3.2...v0.4.0

Open source →

0.3.211 Jul 2025

What's Changed

Optimizer and rewriter

Eliminate unnecessary ScatterND by @gramalingam in https://github.com/microsoft/onnxscript/pull/2422
Change loop order during rewrite by @gramalingam in https://github.com/microsoft/onnxscript/pull/2427
Cleanup elimination of redundant scatter-nd: consolidate rules and improve organization by @Copilot in https://github.com/microsoft/onnxscript/pull/2426
[pass][reland] Enable DeduplicateInitializersPass by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2429
[Rewriter]: fuse successive Relu/Clip nodes by @AyoubMDL in https://github.com/microsoft/onnxscript/pull/2410
Add support for onnx fusions by @gramalingam in https://github.com/microsoft/onnxscript/pull/2412
Fix MatchResult.fail() call signature in redundant_scatter_nd.py by @Copilot in https://github.com/microsoft/onnxscript/pull/2431
A couple of minor fixes on rewrite rules by @gramalingam in https://github.com/microsoft/onnxscript/pull/2432
Reorder optimization passes by @gramalingam in https://github.com/microsoft/onnxscript/pull/2433
Cleanup uses of onnxscript.rewriter.pattern - export symbols from main module by @Copilot in https://github.com/microsoft/onnxscript/pull/2437
Improve redundant slice removal by @gramalingam in https://github.com/microsoft/onnxscript/pull/2441
Make TransposeIdentity more robust by @justinchuby in https://github.com/microsoft/onnxscript/pull/2443

ONNX IR

Use onnx_ir common passes by @justinchuby in https://github.com/microsoft/onnxscript/pull/2420
[pass] Enable DeduplicateInitializersPass by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2416

Torch Lib

[torchlib] Fix sdpa dtype in attn_mask by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2445

Other Changes

Update VERSION to 0.3.2 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2421
Copilot instructions by @gramalingam in https://github.com/microsoft/onnxscript/pull/2448

Full Changelog: https://github.com/microsoft/onnxscript/compare/v0.3.1...v0.3.2

Open source →

0.3.126 Jun 2025

What's Changed

Breaking Changes

[rewriter] Decouple llama rule sets and make API explicit by @copilot-swe-agent in https://github.com/microsoft/onnxscript/pull/2388

Core ONNX Script

Update autocast.py to fix attribute creation error by @justinchuby in https://github.com/microsoft/onnxscript/pull/2365

Optimizer and rewriter

SDPA fusion cleanup by @gramalingam in https://github.com/microsoft/onnxscript/pull/2352
Enable CSE in optimizer by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2361
Test SDPA fusion via MHA by @gramalingam in https://github.com/microsoft/onnxscript/pull/2366
Always fold the Transpose node in the constant folder by @justinchuby in https://github.com/microsoft/onnxscript/pull/2355
Fix fused matmul check/rewrite functions by @bmehta001 in https://github.com/microsoft/onnxscript/pull/2331
Cast-cast elimination by @gramalingam in https://github.com/microsoft/onnxscript/pull/2368
Re-enable fused matmul rules by @bmehta001 in https://github.com/microsoft/onnxscript/pull/2370
Fusion extensions to improve GQA fusion by @gramalingam in https://github.com/microsoft/onnxscript/pull/2374
fix: pattern match gelu from contrib and onnx ops🐛 by @KarelZe in https://github.com/microsoft/onnxscript/pull/2364
[Rewriter]: Add ∘ MatMul -> Gemm by @AyoubMDL in https://github.com/microsoft/onnxscript/pull/2356
Fixes to MHA fusion by @gramalingam in https://github.com/microsoft/onnxscript/pull/2380
fix: check for rank of bias in bias-gelu fusion🐛 by @KarelZe in https://github.com/microsoft/onnxscript/pull/2393
[torchlib] Implement aten__upsample_bicubic2d_aa and aten__upsample_bilinear2d_aa functions by @copilot-swe-agent in https://github.com/microsoft/onnxscript/pull/2383
Move gemm_to_matmul_add rule to ort fusion rules by @justinchuby in https://github.com/microsoft/onnxscript/pull/2398
Fix fusion ordering for partial rotary embedding by @gramalingam in https://github.com/microsoft/onnxscript/pull/2402
[optimizer] Replace value.nbytes with value.size by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2399

Torch Lib

Add missing converter for _local_scalar_dense by @xadupre in https://github.com/microsoft/onnxscript/pull/2367
[torchlib] Unregister aten::max.other by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2377
Support dynamic shapes for aten_unfold by @xenova in https://github.com/microsoft/onnxscript/pull/2407

Documentation

docs: cleanup documentation for function-based rewrites📄 by @KarelZe in https://github.com/microsoft/onnxscript/pull/2359
Updates to the rewriter tutorial by @gramalingam in https://github.com/microsoft/onnxscript/pull/2397
Add sphinx inter link to onnx_ir by @justinchuby in https://github.com/microsoft/onnxscript/pull/2415

Other Changes

Fix test names for pytest by @justinchuby in https://github.com/microsoft/onnxscript/pull/2358
Require onnx-ir 0.1.1 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2360
[CI] Fix execnet.gateway_base.DumpError: can't serialize <class 'numpy.int64'> by @titaiwangms in https://github.com/microsoft/onnxscript/pull/2379
Move _c_api_utils.py to version_converter package by @copilot-swe-agent in https://github.com/microsoft/onnxscript/pull/2401
Fix proto handling in version converter by @justinchuby in https://github.com/microsoft/onnxscript/pull/2411
Add phi2/phi4 test cases for mha/gqa fusion by @gramalingam in https://github.com/microsoft/onnxscript/pull/2409
Update VERSION to 0.3.1 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2414
Bump onnx ir requirement to 0.1.3 by @justinchuby in https://github.com/microsoft/onnxscript/pull/2418
Create torch_2_8 apis by @justinchuby in https://github.com/microsoft/onnxscript/pull/2419

New Contributors

@KarelZe made their first contribution in https://github.com/microsoft/onnxscript/pull/2359
@copilot-swe-agent made their first contribution in https://github.com/microsoft/onnxscript/pull/2383
@xenova made their first contribution in https://github.com/microsoft/onnxscript/pull/2407

Full Changelog: https://github.com/microsoft/onnxscript/compare/v0.3.0...v0.3.1

Open source →

0.3.030 May 2025

[!NOTE] Starting from v0.3, onnxscript.ir is an alias of onnx_ir. Visit https://github.com/onnx/ir-py for source code.

What's Changed

Breaking Changes

[pass] Update LiftSubgraphInitializersToMainGraphPass to disallow variable shadowing by @justinchuby in https://github.com/microsoft/onnxscript/pull/2348
Formally drop python 3.8 support by @justinchuby in https://github.com/microsoft/onnxscript/pull/2354

Optimizer and rewriter

Fix cross attention in MHA by @gramalingam in https://github.com/microsoft/onnxscript/pull/2337
Implement __repr__ for MatchResult by @justinchuby in https://github.com/microsoft/onnxscript/pull/2353

ONNX IR

Use onnx_ir as a dependency by @justinchuby in https://github.com/microsoft/onnxscript/pull/2324

Torch Lib

[torchlib] Set allowzero=True on Reshape where appropriate by @justinchuby in https://github.com/microsoft/onnxscript/pull/2346
[torchlib] Fix pow.Tensor_Scalar type promotion by @justinchuby in https://github.com/microsoft/onnxscript/pull/2335

Other Changes

Update publish pipelines to use 1ES templates by @justinchuby in https://github.com/microsoft/onnxscript/pull/2349

Full Changelog: https://github.com/microsoft/onnxscript/compare/v0.2.7...v0.3.0

Open source →