Skip to content

[Dy2St] PyTorch 和 Paddle 仓库模型导出成功率对比任务 #58985

Closed
@2742195759

Description

背景

PyTorch 最近在 2.1.0 版本发布了 torch.export 功能,提供了以 PyTorch 2.x 的 dynamo 为核心的模型导出方法。其功能上与我们 Paddle 动转静模型导出 paddle.jit.save 相类似,因此我们想要建设一个 repo 能够自动对比 torch.exportpaddle.jit.save 的导出成功与否,PaddleJitLab/jit-exportable-models 就诞生了~

为了确保对比的公平性,需要两个模型使用相同/等价的组网 API,因此我们选取了一些 PaConvert 能够以较高成功率转换经典的 PyTorch 模型作为测试集合,通过 PaConvert 即可将 PyTorch 模型转换为等价的 Paddle 模型。

之后我们分别对转换前后的模型分别调用 torch.export.exportpaddle.jit.save 就可以了解两者模型导出是否成功啦~

任务详情

我们在下面的 #58985 (comment) 提供了详细的操作流程,你可以跟随下面的步骤来完成任务。

最终任务需要提交如下几部分内容:

  • Fork 后并添加 torch.export.export 代码的 PyTorch 模型 repo 1
  • 通过 PaConvert 转换并添加 paddle.jit.save 代码的 Paddle 模型 repo 2
  • PaddleJitLab/jit-exportable-models 提交 PR,增加可运行的自动化测试 shell 脚本

任务完成且通过后会邀请进入 PaddleJitLab 组织,并迁移 repo 1 和 repo 2 进组织~

参考资料

任务列表 (整体进展:12/80)

按 merge 的时间顺序,排名不分先后: @2742195759 (1)、@GreatV (11)

已完成部分
序号 模型名称 模型地址 难度 认领人 PR链接
✅1 TiSASRec.pytorch https://github.com/pmixer/TiSASRec.pytorch/tree/master @2742195759
✅4 pytorch-hed https://github.com/sniklaus/pytorch-hed @GreatV PaddleJitLab/jit-exportable-models#1
✅5 ConSinGAN https://github.com/tohinz/ConSinGAN @GreatV PaddleJitLab/jit-exportable-models#2
✅7 ScaledYOLOv4 https://github.com/WongKinYiu/ScaledYOLOv4 ⭐⭐ @GreatV PaddleJitLab/jit-exportable-models#11
✅11 DINO https://github.com/facebookresearch/dino ⭐⭐⭐ @GreatV PaddleJitLab/jit-exportable-models#7
✅26 EAST https://github.com/SakuraRiven/EAST @GreatV PaddleJitLab/jit-exportable-models#3
✅31 recurrent-visual-attention https://github.com/kevinzakka/recurrent-visual-attention @GreatV PaddleJitLab/jit-exportable-models#4
✅38 few-shot https://github.com/Shandilya21/Few-Shot @GreatV PaddleJitLab/jit-exportable-models#5
✅39 PyTorch-BayesianCNN https://github.com/kumar-shridhar/PyTorch-BayesianCNN ⭐⭐ @GreatV PaddleJitLab/jit-exportable-models#9
✅55 DeepRecommender https://github.com/NVIDIA/DeepRecommender ⭐⭐ @GreatV PaddleJitLab/jit-exportable-models#10
✅56 pren https://github.com/RuijieJ/pren @GreatV PaddleJitLab/jit-exportable-models#8
✅61 wide-resnet.pytorch https://github.com/meliketoy/wide-resnet.pytorch @GreatV PaddleJitLab/jit-exportable-models#6
序号 模型名称 模型地址 难度 认领人 PR链接
2 FastSAM https://github.com/CASIA-IVA-Lab/FastSAM ⭐⭐
3 caser_pytorch https://github.com/graytowne/caser_pytorch
6 Real-ESRGAN https://github.com/xinntao/Real-ESRGAN ⭐⭐
8 pytorch-segmentation https://github.com/qubvel/segmentation_models.pytorch ⭐⭐
9 tabnet https://github.com/dreamquark-ai/tabnet ⭐⭐
10 ContrastiveSeg https://github.com/tfzhou/ContrastiveSeg ⭐⭐⭐
12 DAMO-YOLO https://github.com/tinyvision/DAMO-YOLO ⭐⭐⭐
13 NCR https://github.com/rutgerswiselab/NCR
14 PAN https://github.com/WenmuZhou/PAN.pytorch ⭐⭐
15 monodepth2 https://github.com/OniroAI/MonoDepth-PyTorch
16 pspnet-pytorch https://github.com/Lextal/pspnet-pytorch
17 yolact https://github.com/dbolya/yolact/tree/master ⭐⭐
18 OV-DETR https://github.com/yuhangzang/OV-DETR ⭐⭐
19 yolov5 https://github.com/ultralytics/yolov5 ⭐⭐
20 deeplab-pytorch https://github.com/kazuto1011/deeplab-pytorch ⭐⭐
21 SimCC https://github.com/leeyegy/SimCC
22 MODNet https://github.com/ZHKKKe/MODNet/tree/master ⭐⭐
23 ESPNet https://github.com/espnet/espnet
24 CORA https://github.com/AkariAsai/CORA ⭐⭐⭐
25 AoANet https://github.com/husthuaan/AoANet ⭐⭐
27 detr https://github.com/facebookresearch/detr ⭐⭐
28 pytorch-YOLOv4 https://github.com/Tianxiaomo/pytorch-YOLOv4 ⭐⭐
29 pytorch_Realtime_Multi-Person_Pose_Estimation https://github.com/tensorboy/pytorch_Realtime_Multi-Person_Pose_Estimation ⭐⭐
30 CLIP https://github.com/openai/CLIP ⭐⭐
32 MVFNet https://github.com/Fanziapril/mvfnet ⭐⭐⭐
33 ADDS-DepthNet https://github.com/LINA-lln/ADDS-DepthNet
34 deep-image-prior https://github.com/DmitryUlyanov/deep-image-prior
35 vilbert-multi-task https://github.com/facebookresearch/vilbert-multi-task ⭐⭐
36 TorchSeg https://github.com/ycszen/TorchSeg ⭐⭐
37 CenterNet https://github.com/gakkiri/simple-centernet-pytorch ⭐⭐
40 ViTPose https://github.com/gpastal24/ViTPose-Pytorch ⭐⭐⭐
41 r2c https://github.com/rowanz/r2c
42 DeepCTR-Torch https://github.com/shenweichen/DeepCTR-Torch ⭐⭐
43 hacksc https://github.com/ZoneLikeWonderland/HACK-Model/ ⭐⭐
44 YOLOX https://github.com/Megvii-BaseDetection/YOLOX ⭐⭐
45 a-PyTorch-Tutorial-to-Image-Captioning https://github.com/sgrvinod/a-PyTorch-Tutorial-to-Image-Captioning ⭐⭐
46 prototypical-networks https://github.com/jakesnell/prototypical-networks
47 AdvSemiSeg https://github.com/hfslyc/AdvSemiSeg ⭐⭐
48 FastFCN https://github.com/wuhuikai/FastFCN ⭐⭐
49 STDC-Seg https://github.com/MichaelFan01/STDC-Seg ⭐⭐
50 AdaptSegNet https://github.com/wasidennis/AdaptSegNet
51 tiny-faces-pytorch https://github.com/varunagrawal/tiny-faces-pytorch
52 pytorch_GAN_zoo https://github.com/facebookresearch/pytorch_GAN_zoo
53 efficientdet-pytorch https://github.com/zylo117/Yet-Another-EfficientDet-Pytorch ⭐⭐⭐
54 mmocr https://github.com/open-mmlab/mmocr ⭐⭐⭐
57 VideoMAE https://github.com/MCG-NJU/VideoMAE ⭐⭐⭐
58 pixel2style2pixel https://github.com/eladrich/pixel2style2pixel ⭐⭐
59 dlrm https://github.com/facebookresearch/dlrm ⭐⭐ EmbeddingBag
暂不支持
60 torecsys https://github.com/p768lwy3/torecsys ⭐⭐⭐
62 pit https://github.com/naver-ai/pit ⭐⭐⭐
63 ESPNetv2 https://github.com/sacmehta/ESPNetv2
64 SupContrast https://github.com/HobbitLong/SupContrast
65 3D-ResNets-PyTorch https://github.com/kenshohara/3D-ResNets-PyTorch
66 beit2 https://github.com/sfatimakhan/BEIT ⭐⭐⭐
67 PyTorch-GAN https://github.com/eriklindernoren/PyTorch-GAN ⭐⭐
68 vnet.pytorch https://github.com/mattmacy/vnet.pytorch
69 mmaction2 https://github.com/open-mmlab/mmaction2 ⭐⭐⭐
70 FSPNet https://github.com/ZhouHuang23/FSPNet ⭐⭐⭐
71 CMT https://github.com/gnebehay/CMT ⭐⭐⭐
72 PseCo https://github.com/ligang-cs/PseCo ⭐⭐⭐
73 BackgroundMattingV2 https://github.com/PeterL1n/BackgroundMattingV2 ⭐⭐⭐
74 MapTR https://github.com/hustvl/MapTR ⭐⭐⭐
75 mae https://github.com/facebookresearch/mae ⭐⭐⭐
76 xcit https://github.com/facebookresearch/xcit ⭐⭐⭐
77 T2T-ViT https://github.com/yitu-opensource/T2T-ViT ⭐⭐⭐
78 ContextPrior https://github.com/ycszen/ContextPrior ⭐⭐⭐
79 moco https://github.com/facebookresearch/moco ⭐⭐
80 moco-v3 https://github.com/facebookresearch/moco-v3/tree/main ⭐⭐⭐

其他模型

序号 模型名称 模型地址 难度 认领人 PR链接
81 ??? ??? ⭐⭐⭐⭐
82 ??? ??? ⭐⭐⭐⭐

Metadata

Labels

Type

No type

Projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions