Closed cloud2009 closed 1 year ago
您好,我们已经收到了您的问题,会安排技术人员尽快解答您的问题,请耐心等待。请您再次检查是否提供了清晰的问题描述、复现代码、环境&版本、报错信息等。同时,您也可以通过查看官网API文档、常见问题、历史Issue、AI社区来寻求解答。祝您生活愉快~
Hi! We've received your issue and please be patient to get responded. We will arrange technicians to answer your questions as soon as possible. Please make sure that you have posted enough message to demo your request. You may also check out the API,FAQ,Github Issue and AI community to get the answer.Have a nice day!
任务:195 ⭐⭐ 多学科物理场可视化组件开发 疑问### 1、是基于基础渲染库开发可视化组件,还是在成熟的可视化组件上扩展适配paddle的可视化组件。以什么方式提供给使用者,独立可视化组件还是paddle api。 2、会涉及到机器人/运动控制的物理模拟部分吗? 3、技术方案有限制,如用opengl/glfw,webgl之类的,兼容性要求等。
任务:195 ⭐⭐ 多学科物理场可视化组件开发 疑问### 1、是基于基础渲染库开发可视化组件,还是在成熟的可视化组件上扩展适配paddle的可视化组件。以什么方式提供给使用者,独立可视化组件还是paddle api。 2、会涉及到机器人/运动控制的物理模拟部分吗? 3、技术方案有限制,如用opengl/glfw,webgl之类的,兼容性要求等。
同学,你好!针对以上疑问,回复如下: 1、使用方式是Paddle科学计算套件PaddleScience API; 2、不涉及; 3、不做限制,但要求使用性能,快速、轻便、易用。 欢迎参与开源贡献,随时交流。
任务177, 提了issue一周多了 仍无回复,麻烦加速下, https://github.com/PaddlePaddle/FastDeploy/issues/1416 @cloud2009
【PaddlePaddle Hackathon 4】模型套件开源贡献任务合集
(此 ISSUE 为 PaddlePaddle Hackathon 第四期活动的任务 ISSUE,更多详见 【PaddlePaddle Hackathon 第四期】任务总览)
注:开发请参考 贡献指南,任务列表如下,其他说明事项在任务列表后:
No.98:升级paddlenlp.transformers内的模型结构并且增加基础单测
技术标签:深度学习、Python、NLP
任务难度:基础⭐️
详细描述:
提交内容:
技术要求:
No.99:升级paddlenlp.transformers内的模型结构并且增加基础单测
技术标签:深度学习、Python、NLP
任务难度:基础⭐️
详细描述:
提交内容:
技术要求:
No.100:升级paddlenlp.transformers内的模型结构并且增加基础单测
技术标签:深度学习、Python、NLP
任务难度:基础⭐️
详细描述:
提交内容:
技术要求:
No.101:升级paddlenlp.transformers内的模型结构并且增加基础单测
技术标签:深度学习、Python、NLP
任务难度:基础⭐️
详细描述:
提交内容:
技术要求:
No.102:给AutoConverter增加新的模型组网的支持
任务难度:基础⭐️
详细描述:
为PaddleNLP的AutoConverter增加支持的模型结构,使得更多的PaddleNLP模型可以无缝一行代码加载HuggingFace Hub上的torch模型
每个模型算单独的子任务,总共待升级的模型共5个,完成全部5个模型算完成一个基础任务
clip
distilbert
bart
albert
electra
每个模型添加CompatibilityTest, 能够完成hf-internal-testing内相应torch模型的自动转换与精度对齐,见范例PR
开发流程和环境配置请参考 CONTRIBUTING.md,开发资源可以考虑使用GitHub Codespaces
提交内容:
技术要求:
No.103:新增tie_weights能力
No.104:生成式API对齐HF,包括sample和contrastive_search
No.105:基于PaddleNLP PPDiffusers 训练 AIGC 趣味模型
技术标签: Python、NLP、扩散模型
奖励设置:
详细描述: 结合 PPDiffusers 最新版本,基于自己的数据集,训练并开源趣味模型。可参考 模型训练 QuickStart
提交内容:
提交流程:
No.106:论文名称:Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion
No.107:论文名称:Plug-and-Play Diffusion Features for Text-Driven Image-to-Image Translation
No.108:论文名称:AudioLDM: Text-to-Audio Generation with Latent Diffusion Models
No.109:论文名称:Zero-shot Image-to-Image Translation
No.110:论文名称:Multi-Concept Customization of Text-to-Image Diffusion
No.111:论文名称:DeBERTa: Decoding-enhanced BERT with Disentangled Attention
No.112:论文复现:Multi-Granularity Prediction for Scene Text Recognition
No.113:论文复现:PageNet: Towards End-to-End Weakly Supervised Page-Level Handwritten Chinese Text Recognition
No.114:论文复现:GLASS: Global to Local Attention for Scene-Text Spotting
No.115:论文复现:SPTS: Single-Point Text Spotting
No.116:论文复现:ABCNet v2: Adaptive Bezier-Curve Network for Real-time End-to-end Text Spotting
No.117:论文复现:CoMER: Modeling Coverage for Transformer-based Handwritten Mathematical Expression Recognition
No.118:论文复现:Syntax-Aware Network for Handwritten Mathematical Expression Recognition
技术标签:Python、深度学习
任务难度:基础️⭐️
详细描述:
提交内容:
技术要求:
No.119:论文复现:Learning From Documents in the Wild to Improve Document Unwarping
技术标签:Python、深度学习
任务难度:进阶⭐️⭐️
详细描述:
提交内容:
技术要求:
No.120:论文复现:C3-STISR: Scene Text Image Super-resolution with Triple Clues
技术标签:Python、深度学习
任务难度:进阶⭐️⭐️
详细描述:
提交内容:
技术要求:
No.240:论文复现:Learning Enriched Features for Fast Image Restoration and Enhancement
技术标签:Python、深度学习
任务难度:进阶⭐️⭐️
详细描述:
提交内容:
技术要求:
No.121:PaddleOCR js部署
技术标签:JavaScript、C++
任务难度:进阶⭐️⭐️
详细描述:通过c++打包PP-OCR多语言模型与PP-Structure版面恢复功能,完成js调用示例,开发js demo展示上述能力,撰写使用文档
提交内容:
技术要求:
No.122:《动手学OCR》升级
No.123:模型库中文适配
No.124:论文复现:More ConvNets in the 2020s: Scaling up Kernels Beyond 51x51 using Sparsity
No.125:论文复现:Scaling Up Your Kernels to 31x31: Revisiting Large Kernel Design in CNNs
No.126:论文复现:Revisiting ResNets: Improved Training and Scaling Strategies
No.127:论文复现:Separable Self-attention for Mobile Vision Transformers
No.128:论文复现:MobileViTv3: Mobile-Friendly Vision Transformer with Simple and Effective Fusion of Local, Global and Input Features
No.129:论文复现:Model Rubik’s Cube: Twisting Resolution, Depth andWidth for TinyNets
No.130:论文复现:FlexMatch: Boosting Semi-Supervised Learning with Curriculum Pseudo Labeling
No.131:论文复现:Hypergraph-Induced Semantic Tuplet Loss for Deep Metric Learning
No.132:论文复现:iBOT: Image BERT Pre-Training with Online Tokenizer
No.133:论文复现:Forward Compatible Training for Large-Scale Embedding Retrieval Systems
No.134:论文复现:ICE: Inter-instance Contrastive Encoding for Unsupervised Person Re-identification
No.135:论文复现:VL-LTR: Learning Class-wise Visual-Linguistic Representation for Long-Tailed Visual Recognition
No.136:论文复现:Recall@k Surrogate Loss with Large Batches and Similarity Mixup
No.137:论文题目:Learning Transferable Visual Models From Natural Language Supervision(CLIP)
论文题目:BEIT V2: Masked Image Modeling with Vector-Quantized Visual Tokenizers
No.138:论文题目:Expanding language-image pretrained models for general video recognition(X-CLIP)
论文题目:Context Autoencoder for Self-Supervised Representation Learning(CAE)
No.139:论文题目:BEIT: BERT Pre-Training of Image Transformers
论文题目:Masked Autoencoders Are Scalable Vision Learners(MAE)
论文题目:Exploring Simple Siamese Representation Learning(SimSam)
No.140:论文题目:Improved baselines with momentum contrastive learning(MoCov2)
论文题目:Bootstrap your own latent: A new approach to self-supervised learning(BYOL)
论文题目:Unsupervised Learning of Visual Features by Contrasting Cluster Assignments (SwAV)
No.141:PP-LCNet v3 下游场景验证
No.142:PP-HGNet v2下游场景验证
No.143:万类通用识别数据集制作
No.144:SeaFormer: Squeeze-enhanced Axial Transformer for Mobile Semantic Segmentation
No.145:EfficientFormerV2:Rethinking Vision Transformers for MobileNet Size and Speed
No.146:Fully Convolutional Networks for Panoptic Segmentation
No.147:Per-Pixel Classification is Not All You Need for Semantic Segmentation
No.148:K-Net: Towards Unified Image Segmentation
No.149:Highly Accurate Dichotomous Image Segmentation (ECCV 2022)
No.150:PaddleRS集成PaddleDetection的旋转框检测能力
No.151:PaddleRS运行环境打包,并制作端到端遥感建筑物提取教程
No.152:PaddleRS API 文档完善
No.153:PaddleRS 英文文档
No.154:论文复现:YOLOv6 v3.0: A Full-Scale Reloading
No.155:YOLOv8模型复现
No.156:论文复现:Open-Vocabulary DETR with Conditional Matching
No.157:论文复现:PseCo: Pseudo Labeling and Consistency Training for Semi-Supervised Object Detection
No.158:论文复现:DiffusionDet: Diffusion Model for Object Detection
No.159:论文复现:PromptDet: Towards Open-vocabulary Detection using Uncurated Images
No.160:论文复现:Coarse-to-Fine Vision-Language Pre-training with Fusion in the Backbone
No.161:论文复现:CLRNet: Cross Layer Refinement Network for Lane Detection
No.162:论文复现:Attentional Graph Neural Network for Parking Slot Detection
No.163:基于PaddleDetection PP-TinyPose,新增手势关键点检测模型
No.164:PaddleDetection重点模型接入huggingface
No.165:Camera标定LiDAR标定
No.166:Paddle3D目标检测结果可视化
No.167:Paddle3D&ROS联合开发Demo
No.168:Geometry Uncertainty Projection Network for Monocular 3D Object Detection
No.169:DETR3D: 3D Object Detection from Multi-view Images via 3D-to-2D Queries
No.170:TransFusion: Robust LiDAR-Camera Fusion for 3D Object Detection with Transformers
No.171:FUTR3D: A Unified Sensor Fusion Framework for 3D Detection
No.172:Direct Voxel Grid Optimization: Super-fast Convergence for Radiance Fields Reconstruction
No.173:Point-based Neural Radiance Fields
No.174:Representing Scenes as Neural Radiance Fields for View Synthesis
No.175:TensoRF: Tensorial Radiance Fields
No.176:相机去畸变C++自定义算子开发
No.177:将PP-YOLOE-R在算能BM1684部署。利用FastDeploy,将PP-YOLOE-R在算能BM1684X部署
No.178:集成SOLOv2模型到FastDpeloy,并在Paddle Infenence、ONNX Runtime、TernsorRT后端测试验证
No.179:将PointPillars集成到FastDeploy,并在Jetson Orin硬件上部署验证精度和速度
No.180:在FastDeploy中集成集成地平线推理引擎,在PP-YOLOE完成模型转换测试
No.181:完成TVM接入FastDeploy,并在PP-YOLOE模型上验证正确性
No.182:完成pp-ocrv3在RK3588上的部署,并验证正确性
No.183:使用FastDeploy完成 ELECTRA 模型GLUE任务模型部署
No.184:在FastDeploy C API的基础上,使用rust完成PaddleDetection部署
No.185:在FastDeploy C++ API的基础上,使用java完成PaddleDetection部署
No.186:在FastDeploy C API的基础上,使用go完成PaddleDetection部署
No.187:模型复现:pruned_transducer_stateless8
No.188:模型复现:hubert
No.189:模型复现:wavlm
No.190:模型复现:iSTFTNet
No.191:模型复现:JETS
No.192:使用 Gradio 为 PaddleSpeech 语音识别训练过程绘制WebUI工具箱(以conformer模型为例)
No.193:使用 Gradio 为 PaddleSpeech 语音合成声学模型训练过程绘制WebUI工具箱(以fastspeech2模型为例)
No.194:使用 Gradio 为 PaddleSpeech 语音合成声学模型训练过程绘制WebUI工具箱(以conformer模型为例)
No.195:多学科物理场可视化组件开发
No.196:FVM/FEM/LBM等主流CAE结果提取组件开发
No.197:论文复现:Robust Regression with Highly Corrupted Data via Physics Informed Neural Networks
No.198:论文复现:SPNets: Differentiable Fluid Dynamics for Deep Neural Networks
No.199:论文复现:Reduced-order Model for Flows via Neural Ordinary Differential Equations
No.200:论文复现:An AI-based Domain-Decomposition Non-Intrusive Reduced-Order Model for Extended Domains applied to Multiphase Flow in Pipes
No.201:论文复现:Learning to regularize with a variational autoencoder for hydrologic inverse analysis
No.202:论文复现:Disentangling Generative Factors of Physical Fields Using Variational Autoencoders
No.203:论文复现:Deep Learning Methods for Reynolds-Averaged Navier-Stokes Simulationsof Airfoil Flows
机翼的“翼型”优化一直是航空领域中分析的重点,尤其是在高雷诺情况下,可能存在的湍流等,目前传统方法是采用RANS、LES等数值计算模型,对于神经网络,我们希望能够将上述数值求解的思路或者泛化性应用于网络模型的定义中。具体任务分为两个阶段,可分阶段完成。
阶段一:
阶段二:
No.204:开放赛题:车辆标准部件受力、变形分析
“本任务属于开放性赛题”,围绕汽车、飞机等装备的零部件,进行结构变形、受力分析。本课题选择汽车某标准部件,如下图所示。具体解决两类问题:
车辆零部件示意(截面图)
其中提供的数据集生成流程如下:
note: 数据集会由合作单位产生,无需选手制作
任务具体描述如下:
具体任务要求如下:
请将代码提交至AI4S 开源仓库,并在该路径下创建自己的文件夹,格式为对应任务id的taskid/。
优先实现AIStudio项目,针对任务id建立AIStudio项目,若涉及部分工具无法安装,则提供完整的本地代码合入即可
熟练使用Python;
了解UNET、Transformer等网络结构;
简单了解结构变形原理
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
答疑交流