This directory contains six image classification models, which are models automatically discovered by Baidu Big Data Lab (BDL) Hierarchical Neural Architecture Search project (HiNAS), achieving 96.1% accuracy on CIFAR-10 dataset. These models are divided into two categories. The first three have no skip link, named HiNAS 0-2, and the last three networks contain skip links, which are similar to the shortcut connections in Resnet, named HiNAS 3-5.
This directory contains six image classification models, which are models automatically discovered by Baidu Big Data Lab (BDL) Hierarchical Neural Architecture Search project (HiNAS), achieving 96.1% accuracy on CIFAR-10 dataset. These models are divided into two categories. The first three have no skip link, named HiNAS 0-2, and the last three networks contain skip links, which are similar to the shortcut connections in Resnet, named HiNAS 3-5.
We also recommend users to take a look at the [IPython Notebook demo](https://aistudio.baidu.com/aistudio/projectDetail/122279)
Image classification, which is an important field of computer vision, is to classify images into pre-defined labels. Recently, many researchers have developed different kinds of neural networks and highly improved the classification performance. This page introduces how to do image classification with PaddlePaddle Fluid.
Image classification, which is an important field of computer vision, is to classify images into pre-defined labels. Recently, many researchers have developed different kinds of neural networks and highly improved the classification performance. This page introduces how to do image classification with PaddlePaddle Fluid.
We also recommend users to take a look at the [IPython Notebook demo](https://aistudio.baidu.com/aistudio/projectDetail/122278)
## Quick Start
## Quick Start
### Installation
### Installation
...
@@ -198,7 +200,7 @@ Note: if you train model with flag use_ema, and you want to evaluate your Expone
...
@@ -198,7 +200,7 @@ Note: if you train model with flag use_ema, and you want to evaluate your Expone
Running sample code in this directory requires PaddelPaddle Fluid v.1.3.0 and later. If the PaddlePaddle on your device is lower than this version, please follow the instructions in [installation document](http://paddlepaddle.org/documentation/docs/en/1.3/beginners_guide/install/index_en.html) and make an update.
Running sample code in this directory requires PaddelPaddle Fluid v.1.3.0 and later. If the PaddlePaddle on your device is lower than this version, please follow the instructions in [installation document](http://paddlepaddle.org/documentation/docs/en/1.3/beginners_guide/install/index_en.html) and make an update.
Region Convolutional Neural Network (RCNN) models are two stages detector. According to proposals and feature extraction, obtain class and more precise proposals.
Region Convolutional Neural Network (RCNN) models are two stages detector. According to proposals and feature extraction, obtain class and more precise proposals.
@@ -18,6 +18,7 @@ The Single Shot MultiBox Detector (SSD)
...
@@ -18,6 +18,7 @@ The Single Shot MultiBox Detector (SSD)
SSD is readily pluggable into a wide variant standard convolutional network, such as VGG, ResNet, or MobileNet, which is also called base network or backbone. In this tutorial we used [MobileNet](https://arxiv.org/abs/1704.04861).
SSD is readily pluggable into a wide variant standard convolutional network, such as VGG, ResNet, or MobileNet, which is also called base network or backbone. In this tutorial we used [MobileNet](https://arxiv.org/abs/1704.04861).
We also recommend users to take a look at the [IPython Notebook demo](https://aistudio.baidu.com/aistudio/projectDetail/122276)
@@ -20,6 +20,8 @@ We use many image augment and label smooth tricks from [Bag of Freebies for Trai
...
@@ -20,6 +20,8 @@ We use many image augment and label smooth tricks from [Bag of Freebies for Trai
With execution acceleration method in Paddle framework prediction library, inference speed of YOLOv3 in our impliment can be 30% faster than darknet framework.
With execution acceleration method in Paddle framework prediction library, inference speed of YOLOv3 in our impliment can be 30% faster than darknet framework.
We also recommend users to take a look at the [IPython Notebook demo](https://aistudio.baidu.com/aistudio/projectDetail/122277)
## Quick Start
## Quick Start
### Installation
### Installation
...
@@ -94,7 +96,7 @@ Please make sure that pre-trained model is downloaded and loaded correctly, othe
...
@@ -94,7 +96,7 @@ Please make sure that pre-trained model is downloaded and loaded correctly, othe
--data_dir=${path_to_data} \
--data_dir=${path_to_data} \
--class_num=${category_num}
--class_num=${category_num}
- Set `export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7` to specifiy 8 GPUs to train.
- Set `export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7` to specifiy 8 GPUs to train.
- It is recommended to set `--use_multiprocess_reader=False` when training on Windows.
- It is recommended to set `--use_multiprocess_reader=False` when training on Windows.
- Set `--worker_num=` to specifiy multiprocess reader worker number, which is default 8, if the number of CPU cores in the training environment is small, it is recommended to set worker number to a small value.
- Set `--worker_num=` to specifiy multiprocess reader worker number, which is default 8, if the number of CPU cores in the training environment is small, it is recommended to set worker number to a small value.
同时我们也给出了 id 化后的部分训练数据：[`demo_wiki_train.gz`](./data/train/demo_wiki_train.gz)、和测试数据：[`demo_wiki_validation.gz`](./data/validation/demo_wiki_validation.gz)，每行数据为1个训练样本，示例如下:
同时我们也给出了 id 化后的部分训练数据：[`demo_wiki_train.gz`](./data/train/demo_wiki_train.gz)、和测试数据：[`demo_wiki_validation.gz`](./data/validation/demo_wiki_validation.gz)，每行数据为1个训练样本，示例如下:
ELMo(Embeddings from Language Models) 是重要的通用语义表示模型之一，以双向 LSTM 为网路基本组件，以 Language Model 为训练目标，通过预训练得到通用的语义表示，将通用的语义表示作为 Feature 迁移到下游 NLP 任务中，会显著提升下游任务的模型性能。本项目是 ELMo 在 Paddle Fluid 上的开源实现, 基于百科类数据训练并发布了预训练模型。
ELMo(Embeddings from Language Models) 是重要的通用语义表示模型之一，以双向 LSTM 为网路基本组件，以 Language Model 为训练目标，通过预训练得到通用的语义表示，将通用的语义表示作为 Feature 迁移到下游 NLP 任务中，会显著提升下游任务的模型性能。本项目是 ELMo 在 Paddle Fluid 上的开源实现, 基于百科类数据训练并发布了预训练模型。
In personalized recommendation scenario, a user often is provided with several items from personalized interest matching model. In real world application, a user may have multiple views of features, say user-id, age, click-history of items, search queries. A item, e.g. news, may also have multiple views of features like news title, news category, images in news and so on. Multi-view Simnet is matching a model that combine users' and items' multiple views of features into one unified model. The model can be used in many industrial product like Baidu's feed news. The model is adapted from the paper A Multi-View Deep Learning(MV-DNN) Approach for Cross Domain User Modeling in Recommendation Systems, WWW 2015. The difference between our model and the MV-DNN is that we also consider multiple feature views of users.
In personalized recommendation scenario, a user often is provided with several items from personalized interest matching model. In real world application, a user may have multiple views of features, say user-id, age, click-history of items, search queries. A item, e.g. news, may also have multiple views of features like news title, news category, images in news and so on. Multi-view Simnet is matching a model that combine users' and items' multiple views of features into one unified model. The model can be used in many industrial product like Baidu's feed news. The model is adapted from the paper A Multi-View Deep Learning(MV-DNN) Approach for Cross Domain User Modeling in Recommendation Systems, WWW 2015. The difference between our model and the MV-DNN is that we also consider multiple feature views of users.
We also recommend users to take a look at the [IPython Notebook demo](https://aistudio.baidu.com/aistudio/projectDetail/122294)
## Dataset
## Dataset
Currently, synthetic dataset is provided for proof of concept and we aim to add more real world dataset in this project in the future. The result is inaccurate because of synthetic dataset.
Currently, synthetic dataset is provided for proof of concept and we aim to add more real world dataset in this project in the future. The result is inaccurate because of synthetic dataset.
