Paddle2ONNX v2.0.2rc2

What's Changed

• [Debug] Record debug info for exporting op by @0x3878f in #1568
• Add Conv2D_Dropout test by @zhanghonggeng in #1569
• [Fix] Fix the issue with mapper of tile op by @0x3878f in #1571
• Use shutil.which instead of deprecated find_executable by @emmanuel-ferdman in #1572
• Support conversion of hardswish using opset 7 by @0x3878f in #1574
• Update validation of paddle by @0x3878f in #1576

New Contributors

• @emmanuel-ferdman made their first contribution in #1572

Full Changelog: v2.0.2rc1...v2.0.2rc2

Paddle2ONNX v2.0.2rc1

What's Changed

• Support compilation on windows by @0x3878f in #1561
• Fix the issue with optimize tools by @0x3878f in #1563
• [Optimizer] Check ONNX model after optimization by @0x3878f in #1565

Full Changelog: v2.0.1...v2.0.2rc1

Paddle2ONNX v2.0.1

What's Changed

• make libpaddle2onnx.so link libpaddle.so and include paddle_include direcotory by @risemeup1 in #1352
• support pir mode in paddle2onnx by @risemeup1 in #1354
• Support pir by @qzylalala in #1389
• Get input/output/attrvar interface by @risemeup1 in #1380
• Conversion of ResNet50 model under PIR by @0x3878f in #1382
• [PIR] Use paddlepaddle==3.0.0.dev20240904 in CI of test_pir by @0x3878f in #1395
• [PIR] Completion of unit testing in PIR mode by @0x3878f in #1394
• [OSPP][PIR] support mobilenet v2 by @qzylalala in #1397
• Automatic mapping of operator variable names to indices by @0x3878f in #1401
• [PIR] Adjust some interfaces in PIR && Support conversion of builtin.combine and builtin.split by @0x3878f in #1402
• Support op concat, split, split_with_num, transpose by @0x3878f in #1403
• [OSPP][PIR] support dropout and squeeze by @qzylalala in #1406
• [PIR]support if op by @risemeup1 in #1407
• support some ops in pir by @qzylalala in #1416
• Support some operations by @qzylalala in #1419
• Support while op in pir mode by @0x3878f in #1426
• Support some ops in PIR by @0x3878f in #1435
• Merge PR#1434 by @0x3878f in #1437
• support some ops by @qzylalala in #1434
• Improved the mechanism, added the conversion of some ops, and fixed some issues. by @0x3878f in #1444
• Improve some operations to support more scenarios by @0x3878f in #1450
• Fix issues with inference of LaTeX_OCR_rec model by @0x3878f in #1452
• Merge the develop branch into the test_pir branch by @0x3878f in #1455
• Fix the issue with multiclass_nms3 op by @0x3878f in #1456
• Support compilation and loading on macOS by @0x3878f in #1462
• Support compilation and loading on Windows by @0x3878f in #1465
• Add patch for compilaition on Windows by @0x3878f in #1470
• Update usage documentation by @0x3878f in #1473
• Enhance monitoring for macOS and Windows platforms using github action by @0x3878f in #1469
• Fix issues with inference of onnx models converted from paddlex by @0x3878f in #1484
• Fix the issue with determining whether operators in control flow are registered. by @0x3878f in #1485
• Support phi::DataType::UNDEFINED by @0x3878f in #1486
• Improve the mechanism by @0x3878f in #1487
• Update CI and README by @0x3878f in #1492
• Update the package release workflow by @0x3878f in #1497
• Fix the issue with while op by @0x3878f in #1499
• support round and fft_r2c by @risemeup1 in #1501
• Support decompose pir program in paddle2onnx by @zhanghonggeng in #1500
• fix abs bug by @risemeup1 in #1505
• Support get scalar attribute by @0x3878f in #1502
• Update README by @0x3878f in #1508
• Support PT in paddle2onnx by @risemeup1 in #1481
• Fix the issue with pd_op.max by @0x3878f in #1506
• support isinite isinf isnan by @zhanghonggeng in #1509
• Fix the issue with input data by @0x3878f in #1511
• support masked_select & fix isinf isnan tes by @zhanghonggeng in #1512
• fix tile mapper by @zhanghonggeng in #1513
• support op hardtanh & fix op unfold, empty, max_pool2d_with_index, max_pool3d_with_index by @zhanghonggeng in #1518
• fix conv2d_transpose op by @zhouleiMax in #1522
• Support PP-FormulaNet-S and PP-FormulaNet-L by @0x3878f in #1523
• Fix the issue with batch norm op exposed in LaTeX_OCR_rec by @0x3878f in #1529
• [CI] Add a merge step in workflow by @0x3878f in #1531
• Add attribute seed for multinomial by @0x3878f in #1530
• [Debug] A simple debug tool for identifying accuracy issues in the converted ONNX model by @0x3878f in #1534
• [CI] Fix the issue with action on push by @0x3878f in #1535
• fix some op mapper and tests by @zhanghonggeng in #1519
• support quantize_linear & dequantize_linear op and float8_e4m3fn, float8_e5m2 dtype by @zhanghonggeng in #1528
• Update the CI configuration and resolve the compilation issues. by @0x3878f in #1539
• Add optimization for onnx by @0x3878f in #1543
• [PIR] A new rule for input and output names. by @0x3878f in #1544
• [LOG] Standardize log information by @0x3878f in #1545
• [CodeStyle] Update pre-commit configs and refactor code formatting by @0x3878f in #1547
• [Dy2ONNX] Improve dygraph2onnx API by @0x3878f in #1549
• [Fix] fix the issue with batch norm mapper by @0x3878f in #1550
• [PT | Decomposition] Improve program translatation and decomposition by @0x3878f in #1551
• [CI][CodeStyle] Add a new workflow to check code style by @0x3878f in #1552
• [Dy2ONNX] Support conversion configurations by @0x3878f in #1555
• Update README by @0x3878f in #1554
• [Fix] Fix the issue with split op and update ONNX optimizer by @0x3878f in #1559

New Contributors

• @qzylalala made their first contribution in #1389
• @zhanghonggeng made their first contribution in #1500
• @zhouleiMax made their first contribution in #1522

Full Changelog: v1.2.6...v2.0.1

Paddle2ONNX v2.0.1a1

What's Changed

• [Dy2ONNX] Support conversion configurations by @0x3878f in #1555
• Update README by @0x3878f in #1554

Full Changelog: v2.0.0a5...v2.0.1a1

Paddle2ONNX v2.0.0a5

What's Changed

• [CodeStyle] Update pre-commit configs and refactor code formatting by @0x3878f in #1547
• [Dy2ONNX] Improve dygraph2onnx API by @0x3878f in #1549
• [Fix] fix the issue with batch norm mapper by @0x3878f in #1550
• [PT | Decomposition] Improve program translatation and decomposition by @0x3878f in #1551
• [CI][CodeStyle] Add a new workflow to check code style by @0x3878f in #1552

Full Changelog: v2.0.0a4...v2.0.0a5

Paddle2ONNX v2.0.0a4

What's Changed

• [ONNX optimizer] Add optimization for onnx by @0x3878f in #1543
• [PIR] A new rule for input and output names. by @0x3878f in #1544
• [LOG] Standardize log information by @0x3878f in #1545

Full Changelog: v2.0.0a3...v2.0.0a4

Paddle2ONNX v2.0.0a3

What's Changed

• Fix the issue with batch norm op exposed in LaTeX_OCR_rec by @0x3878f in #1529
• [CI] Add a merge step in workflow by @0x3878f in #1531
• Add attribute seed for multinomial by @0x3878f in #1530
• [Debug] A simple debug tool for identifying accuracy issues in the converted ONNX model by @0x3878f in #1534
• [CI] Fix the issue with action on push by @0x3878f in #1535
• fix some op mapper and tests by @zhanghonggeng in #1519
• support quantize_linear & dequantize_linear op and float8_e4m3fn, float8_e5m2 dtype by @zhanghonggeng in #1528
• Update version number by @0x3878f in #1538
• Update the CI configuration and resolve the compilation issues. by @0x3878f in #1539

Full Changelog: v2.0.0a2...v2.0.0a3

Paddle2ONNX v2.0.0a2

What's Changed

• Update the package release workflow by @0x3878f in #1497
• Fix the issue with while op by @0x3878f in #1499
• support round and fft_r2c by @risemeup1 in #1501
• Support decompose pir program in paddle2onnx by @zhanghonggeng in #1500
• fix abs bug by @risemeup1 in #1505
• Support get scalar attribute by @0x3878f in #1502
• Update README by @0x3878f in #1508
• Support PT in paddle2onnx by @risemeup1 in #1481
• Fix the issue with pd_op.max by @0x3878f in #1506
• support isinite isinf isnan by @zhanghonggeng in #1509
• Fix the issue with input data by @0x3878f in #1511
• support masked_select & fix isinf isnan tes by @zhanghonggeng in #1512
• fix tile mapper by @zhanghonggeng in #1513
• support op hardtanh & fix op unfold, empty, max_pool2d_with_index, max_pool3d_with_index by @zhanghonggeng in #1518
• fix conv2d_transpose op by @zhouleiMax in #1522
• Support PP-FormulaNet-S and PP-FormulaNet-L by @0x3878f in #1523
• Update version number by @0x3878f in #1524

New Contributors

• @zhanghonggeng made their first contribution in #1500
• @zhouleiMax made their first contribution in #1522

Full Changelog: v2.0.0a1...v2.0.0a2

v2.0.0a1

What's Changed

• [Cmake]Fix undefined symbol for std::filesystem under GCC8 + Linux by @Aurelius84 in #1333
• make libpaddle2onnx.so link libpaddle.so and include paddle_include direcotory by @risemeup1 in #1352
• support pir mode in paddle2onnx by @risemeup1 in #1354
• Support pir by @qzylalala in #1389
• Get input/output/attrvar interface by @risemeup1 in #1380
• Conversion of ResNet50 model under PIR by @0x3878f in #1382
• [PIR] Use paddlepaddle==3.0.0.dev20240904 in CI of test_pir by @0x3878f in #1395
• [PIR] Completion of unit testing in PIR mode by @0x3878f in #1394
• [OSPP][PIR] support mobilenet v2 by @qzylalala in #1397
• Automatic mapping of operator variable names to indices by @0x3878f in #1401
• [PIR] Adjust some interfaces in PIR && Support conversion of builtin.combine and builtin.split by @0x3878f in #1402
• 【P2O-PIR】Support op concat, split, split_with_num, transpose by @0x3878f in #1403
• [OSPP][PIR] support dropout and squeeze by @qzylalala in #1406
• [PIR]support if op by @risemeup1 in #1407
• support some ops in pir by @qzylalala in #1416
• Support some operations by @qzylalala in #1419
• 【PIR】Support while op in pir mode by @0x3878f in #1426
• Support some ops in PIR by @0x3878f in #1435
• Merge PR#1434 by @0x3878f in #1437
• support some ops by @qzylalala in #1434
• 【PIR】Improved the mechanism, added the conversion of some ops, and fixed some issues. by @0x3878f in #1444
• 【PIR】Improve some operations to support more scenarios by @0x3878f in #1450
• 【PIR】Fix issues with inference of LaTeX_OCR_rec model by @0x3878f in #1452
• 【PIR】Merge the develop branch into the test_pir branch by @0x3878f in #1455
• Fix the issue with multiclass_nms3 op by @0x3878f in #1456
• Support compilation and loading on macOS by @0x3878f in #1462
• Support compilation and loading on Windows by @0x3878f in #1465
• Add patch for compilaition on Windows by @0x3878f in #1470
• Update usage documentation by @0x3878f in #1473
• Enhance monitoring for macOS and Windows platforms using github action by @0x3878f in #1469
• Fix issues with inference of onnx models converted from paddlex by @0x3878f in #1484
• Fix the issue with determining whether operators in control flow are registered. by @0x3878f in #1485
• Support phi::DataType::UNDEFINED by @0x3878f in #1486
• Improve the mechanism by @0x3878f in #1487
• Update CI and README by @0x3878f in #1492
• Update release_linux_x86_64.yml by @risemeup1 in #1490
• Update release_linux_x86_64.yml by @risemeup1 in #1493
• Update release_linux_x86_64.yml by @risemeup1 in #1494
• Update version number by @0x3878f in #1496

New Contributors

• @qzylalala made their first contribution in #1389

Full Changelog: v1.2.6...v2.0.0a1

Paddle2ONNX v1.3.1

What's Changed

• Create clear_issues.yml by @Zheng-Bicheng in #1443
• Add MacRelease by @ccsuzzh in #1445

Full Changelog: v1.3.0...v1.3.1