What's Changed

• Fix quantize OP Matmul and Relu by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/788
• Add partial_sum and partial_concat support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/791
• Add temporal_shift support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/792
• Fix multiclass_nms by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/794
• Keep input/output same order with paddle by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/789
• Fix gather aixs bug of nms by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/795
• Add leaky_relu support in quantize model by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/797
• Add scatter_nd_add op support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/800
• Add conv3d and pool3d support, fix a bug in tile by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/798
• Add quantize model related doc by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/801
• Add ONNX util function by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/806
• Add Paddle2Caffe under Contrib by @ltcs11 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/799
• Add Paddle model util function by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/808
• Support operator export as custom op by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/814
• Enhance OnnxReader by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/816
• Fix bug in PaddleReader by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/817
• Support remove multiclass_nms by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/821
• Add deploy backend option and TRT backend support for quantize model by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/819
• Add bilinear_interp OP by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/829
• Fix quantize model export for TRT backend by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/830
• Add take_along _axis OP support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/831
• Support read parameters of nms operator by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/834
• Modify install directory on windows by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/835
• Add clip extra in test by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/822
• Set enable_dev_version to True by default by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/836
• Fix windows build bug by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/839
• Fix ONNX reader while initializer exists by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/845
• [OpAttr]Support to parse Attribute with type(Variable) in 2ONNX by @Aurelius84 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/854
• Update document by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/842
• separate setup for Paddle2Caffe by @ltcs11 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/843
• Add linspace support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/860
• Add group_norm support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/863
• Add top_k OP support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/861
• Support new ir of paddle while by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/865
• Make roi of interploate be empty by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/867
• Replace Explicit Quantization to Implicit Quantization for TensorRT backend by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/856
• Update compile.md to support previous version ONNXRuntime by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/873
• Update compile.md by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/874
• Add index_select support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/876
• Rename paddle2caffe by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/886
• Enhance ONNX reader to get input/output information from ONNX model by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/885
• Add RKNN deploy backend support for quantized model by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/887
• Add grid_sampler op convert support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/889

New Contributors

• @ltcs11 made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/799
• @Aurelius84 made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/854

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v0.9.9...v1.0.0

What's Changed

• Support read parameters of nms operator by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/834
• Modify install directory on windows by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/835
• Add clip extra in test by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/822
• Set enable_dev_version to True by default by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/836
• Fix windows build bug by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/839
• Fix onnx reader while initializer exists by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/845
• [OpAttr]Support to parse Attribute with type(Variable) in 2ONNX by @Aurelius84 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/854
• Update document by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/842
• Separate setup for Paddle2Caffe by @ltcs11 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/843
• Fix test by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/864
• Add linspace support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/860
• Add group_norm support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/863
• Add top_k OP support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/861
• Support new ir of paddle while by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/865
• Make roi of interploate be empty by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/867
• Replace Explicit Quantization to Implicit Quantization for TensorRT backend by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/856

New Contributors

• @Aurelius84 made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/854

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v1.0.0rc2...v1.0.0rc4

What's Changed

• Add deploy backend option and TRT backend support for quantize model by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/819
• Fix quantize OP Matmul and Relu by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/788
• Add partial_sum and partial_concat support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/791
• Add temporal_shift support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/792
• Fix multiclass_nms by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/794
• Keep input/output same order with paddle by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/789
• Fix gather axis bug of nms by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/795
• Add leaky_relu support in quantize model by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/797
• Add scatter_nd_add op support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/800
• Add conv3d and pool3d support, fix a bug in tile by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/798
• Add quantize model related doc by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/801
• Add onnx util function by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/806
• Fix quantize bugs by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/804
• Add Paddle2Caffe under Contrib by @ltcs11 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/799
• Add Paddle model util function by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/808
• Support operator export as custom op by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/814
• Enhance OnnxReader by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/816
• Fix bug in PaddleReader by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/817
• Support remove multiclass_nms by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/821
• Add bilinear_interp OP by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/829

New Contributors

• @ltcs11 made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/799

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v0.9.9...v1.0.0rc2

What's Changed

• Update ONNXRuntime to version 1.11.1 in ci test by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/787
• Fix quantize OP Matmul and Relu by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/788
• Add partial_sum and partial_concat support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/791
• Add temporal_shift support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/792
• Fix multiclass_nms by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/794
• Keep input/output same order with paddle by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/789
• Fix gather aixs bug of nms and add nms test by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/795
• Add leakyrelu OP support in quantize model by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/797
• Add conv3d, pool3d OPs support and fix a bug in tile by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/798
• Add scatter_nd_add OP support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/800
• Add quantize model related doc and change abs to fabs by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/801

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v0.9.9...v1.0.0rc

What's Changed

• Fix bug of quantize by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/784

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v0.9.8...v0.9.9

What's Changed

• Add where_index operator by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/759
• Add eye operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/763
• Add elementwise_floordiv operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/770
• Add dot operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/764
• Add gather_nd operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/765
• Add norm operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/766
• Add flip operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/771
• Add p_norm operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/769
• Add one_hot_v2 operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/768
• Add instance_norm by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/778
• Add scatter operator support by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/767
• Add version to library by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/777
• Quantize model support @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/761 https://github.com/PaddlePaddle/Paddle2ONNX/pull/782
• Add more debug information while export ONNX by @zh794390558 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/775
• Fix README.md by @onecatcn in https://github.com/PaddlePaddle/Paddle2ONNX/pull/783

New Contributors

• @zh794390558 made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/775
• @onecatcn made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/783

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v0.9.7...v0.9.8

What's Changed

• Remove older version code to legacy folder @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/736
• Add survey notice to help collect feedback from users by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/739
• Set paddle.onnx.export use new version of Paddle2ONNX by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/740
• Modify export interface to solve ABI fault by @heliqi in https://github.com/PaddlePaddle/Paddle2ONNX/pull/744
• Add silu support by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/745
• Fix LoadParamsFromMemoryBuffer warning by @heliqi in https://github.com/PaddlePaddle/Paddle2ONNX/pull/751
• Fix multiclass nms by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/756

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v0.9.6...v0.9.7

What's Changed

• Make layer_norm similar with torch by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/718
• Add expand and expand_as op by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/713
• Add argsort and dist op by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/703
• Optmize layer norm conversion by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/728
• [Eager] Support eager data structure by @JiabinYang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/730
• Support faster-rcnn/mask-rcnn/cascade_rcnn/cascade_mask_rcnn by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/636
• Add unstack op by @yeliang2258 in https://github.com/PaddlePaddle/Paddle2ONNX/pull/733

New Contributors

• @JiabinYang made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/730

Full Changelog: https://github.com/PaddlePaddle/Paddle2ONNX/compare/v0.9.5...v0.9.6

New operator

• Add rnn/gaussian_random/pad3d to support speech model by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/678
• Add 33 operators(activations, reduce) by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/679
• Add operator less_equal and greater_equal by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/680
• Add Topk_v2 op by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/521
• Add resize op test case by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/418
• Add Scatter op by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/565
• Assign / Assign_value op by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/520
• Add pow op by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/676
• Add flatten2 op and fill_constant_batch_size_like test case by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/698
• Fix gcc5.4 error by @heliqi in https://github.com/PaddlePaddle/Paddle2ONNX/pull/674
• Update some op level by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/675
• Remove some deprecate warning and enable test for dev version by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/677
• Add data_norm and mul by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/701

Bug Fix

• Fix problem detected by testing by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/681
• Fix trilinear op bug by @neonhuang in https://github.com/PaddlePaddle/Paddle2ONNX/pull/684
• Fix attr for yolo_box by @ming-ch in https://github.com/PaddlePaddle/Paddle2ONNX/pull/690
• Fix conv2d by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/692
• Fix matmul by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/693
• Add optimization for mul_one and add_zero by @jiangjiajun in https://github.com/PaddlePaddle/Paddle2ONNX/pull/708

New Contributors

• @ming-ch made their first contribution in https://github.com/PaddlePaddle/Paddle2ONNX/pull/690

Experimental Feature

• A new version of Paddle2ONNX is integrated in v0.9.4, which will make conversion faster and also optimize the converted onnx model. You can enable the new feature with flag --enable_dev_version=True https://github.com/PaddlePaddle/Paddle2ONNX/pull/669

Bug Fix

• Fix bug of shape inference while exporting OCR model https://github.com/PaddlePaddle/Paddle2ONNX/pull/656
• Fix bug of box_coder under some onnx opset_version https://github.com/PaddlePaddle/Paddle2ONNX/pull/667
• Fix bug of unsqueeze under some onnx opset_version https://github.com/PaddlePaddle/Paddle2ONNX/pull/652

New Operators

• Support mish、eye and one_hot_v2 https://github.com/PaddlePaddle/Paddle2ONNX/pull/663 https://github.com/PaddlePaddle/Paddle2ONNX/pull/655