Merge pull request #100 from huawei-noah/zjj_1.3_release

v1.3 release

MANIFEST.in

@@ -1,5 +1,4 @@
 #dispatch files to site-packages
-recursive-include deploy *
 recursive-include docs *
 recursive-include evaluate_service *
 recursive-include examples *

README.cn.md

@@ -1,15 +1,23 @@
+
+<div align="center">
+  <img src="./docs/images/pipeline.png"/>
+</div>
+
 # Vega
 
-[English](./README.md)
+**[English](./README.md) | 中文**
 
-**Vega ver1.2.0 发布：**
+**Vega ver1.3.0 发布：**
 
 - 特性增强：
-  - 细粒度网络搜索空间：可以自由定义网络搜索空间，提供了丰富的网络架构参数供搜索空间使用，可同时搜索网络架构参数和模型训练超参，并且该搜索空间可以同时适用于Pytorch、TensorFlow和MindSpore。
+  - 支持昇腾平台，在Ascend 910搜索和训练（TensorFlow和MindSpore），在Ascend 310上模型评估。
+  - 支持在麒麟990上模型评估。
+  - 支持用户数据集在DNet预模型上FineTune，在Ascend 910/310上超越SOTA。
+  - 支持用户数据集和ResNet类模型的剪枝能力。Cifar100数据集，精度变化较小（+-0.5），时延降低15%，模型大小降低30%。
 - 新增算法：
-  - [NAGO: Neural Architecture Generator Optimization](https://arxiv.org/abs/2004.01395) 基于随机图生成器的，多阶层的神经网络搜索空间
-
-- 社区贡献者：[Chen Bo](https://github.com/chenboability), [cndylan](https://github.com/cndylan), [hasanirtiza](https://github.com/hasanirtiza), [IlyaTrofimov](https://github.com/IlyaTrofimov), [Lzc06](https://github.com/Lzc06), [marsggbo](https://github.com/marsggbo), [mengzhibin](https://github.com/mengzhibin), [qixiuai](https://github.com/qixiuai), [SHUHarold](https://github.com/SHUHarold), [sptj](https://github.com/sptj).
+  - ModularNAS: Towards Modularized and Reusable Neural Architecture Search, 支持权重共享和网络变换等方法的神经网络架构搜索算法集。
+  - DNet：提供和Davinci芯片亲和的预训练模型，和网络搜索算法。并提供DNet Model Zoo。
+  - MF-ASC: Multi-Fidelity neural Architecture Search with Co-kriging.
 
 ## Vega简介
 
@@ -19,13 +27,16 @@ Vega是诺亚方舟实验室自研的AutoML算法工具链，有主要特点：
 2. 业界标杆的自研算法：提供了诺亚方舟实验室自研的 **[业界标杆(Benchmark)](./docs/benchmark.md)** 算法，并提供 **[Model Zoo](./docs/model_zoo.md)** 下载SOTA(State-of-the-art)模型。
 3. 高并发模型训练能力：提供高性能Trainer，加速模型训练和评估。
 4. 细粒度SearchSpace：可以自由定义网络搜索空间，提供了丰富的网络架构参数供搜索空间使用，可同时搜索网络架构参数和模型训练超参，并且该搜索空间可以同时适用于Pytorch、TensorFlow和MindSpore。
-5. 多Backend支持：支持PyTorch，TensorFlow，MindSpore（试用中）。
+5. 多Backend支持：支持PyTorch(GPU), TensorFlow(GPU, Ascend 910), MindSpore(Ascend 910).。
+6. 支持昇腾平台：支持在Ascend 910搜索和训练，支持在Ascend 310上模型评估。
 
 ## 算法列表
 
 | 分类 | 算法 | 说明 | 参考 |
 | :--: | :-- | :-- | :-- |
 | NAS | [CARS: Continuous Evolution for Efficient Neural Architecture Search](https://arxiv.org/abs/1909.04977) | 基于连续进化的多目标高效神经网络结构搜索方法 | [参考](./docs/cn/algorithms/cars.md) |
+| NAS | ModularNAS: Towards Modularized and Reusable Neural Architecture Search |  支持权重共享和网络变换等方法的神经网络架构搜索算法集 | [参考](./docs/cn/algorithms/modnas.md) |
+| NAS | [MF-ASC](https://dl.acm.org/doi/10.1145/3292500.3330893) | Multi-Fidelity neural Architecture Search with Co-kriging | [参考](./docs/en/algorithms/mfasc.md) |
 | NAS | [NAGO: Neural Architecture Generator Optimization](https://arxiv.org/abs/2004.01395) | 基于随机图生成器的，多阶层的神经网络搜索空间 | [参考](./docs/cn/algorithms/nago.md) |
 | NAS | SR-EA | 适用于轻量级网络的自动网络架构搜索方法 | [参考](./docs/cn/algorithms/sr_ea.md) |
 | NAS | [ESR-EA: Efficient Residual Dense Block Search for Image Super-resolution](https://arxiv.org/abs/1909.11409) | 基于网络架构搜索的多目标图像超分方法 | [参考](./docs/cn/algorithms/esr_ea.md) |
@@ -51,15 +62,16 @@ Vega是诺亚方舟实验室自研的AutoML算法工具链，有主要特点：
 执行如下命令安装Vega和相关开源软件：
 
 ```bash
-pip3 install --user noah-vega
-python3 -m vega.tools.install_pkgs
+pip3 install --user --upgrade noah-vega
 ```
 
+若需要在Ascend 910训练环境上安装，请联系我们。
+
 ## 使用参考
 
 | 对象 | 参考 |
 | :--: | :-- |
-| [**用户**<br>(用户指南)](./docs/cn/user/README.md) | [安装指导](./docs/cn/user/install.md)、[部署指导](./docs/cn/user/deployment.md)、[配置指导](./docs/cn/user/config_reference.md)、[示例参考](./docs/cn/user/examples.md)、[评估服务](../docs/cn/user/evaluate_service.md)、任务参考([分类](./docs/cn/tasks/classification.md)、[检测](./docs/cn/tasks/detection.md)、[分割](./docs/cn/tasks/segmentation.md)、[超分](./docs/cn/tasks/segmentation.md)) |
+| [**用户**<br>(用户指南)](./docs/cn/user/README.md) | [安装指导](./docs/cn/user/install.md)、[部署指导](./docs/cn/user/deployment.md)、[配置指导](./docs/cn/user/config_reference.md)、[示例参考](./docs/cn/user/examples.md)、[评估服务](./docs/cn/user/evaluate_service.md)、任务参考([分类](./docs/cn/tasks/classification.md)、[检测](./docs/cn/tasks/detection.md)、[分割](./docs/cn/tasks/segmentation.md)、[超分](./docs/cn/tasks/segmentation.md)) |
 | [**开发者**<br>(开发者指南)](./docs/cn/developer/README.md) | [开发者指导](./docs/cn/developer/developer_guide.md)、[快速入门指导](./docs/cn/developer/quick_start.md)、[数据集指导](./docs/cn/developer/datasets.md)、[算法开发指导](./docs/cn/developer/new_algorithm.md)、[细粒度搜索空间指导](./docs/cn/developer/fine_grained_space.md) |
 
 ## FAQ

README.md

@@ -1,15 +1,22 @@
+<div align="center">
+  <img src="./docs/images/pipeline.png"/>
+</div>
+
 # Vega
 
-[中文](./README.cn.md)
+**English | [中文](./README.cn.md)**
 
 **Vega ver1.2.0 released:**
 
 - Feature enhancement:
-  - Fine-grained network search space: The network search space can be freely defined, and rich network architecture parameters are provided for use in the search space. The network architecture parameters and model training hyperparameters can be searched at the same time, and the search space can be applied to Pytorch, TensorFlow and MindSpore.
+  - Ascend platform, search and training on the Ascend 910 (TensorFlow and MindSpore), and model evaluation on the Ascend 310.
+  - Model evaluation is supported on the Kirin 990.
+  - Allows user datasets to be FineTune on DNet pretrained models and surpass SOTA on Ascend 910/310.
+  - Support the pruning capability of user datasets and ResNet models. For the Cifar100 data set, the precision changes slightly (+– 0.5), the latency decreases by 15%, and the model size decreases by 30%.
 - New algorithm:
-  - [NAGO: Neural Architecture Generator Optimization](https://arxiv.org/abs/2004.01395): An Hierarchical Graph-based Neural Architecture Search Space
-
-- Community Contributors: [Chen Bo](https://github.com/chenboability), [cndylan](https://github.com/cndylan), [hasanirtiza](https://github.com/hasanirtiza), [IlyaTrofimov](https://github.com/IlyaTrofimov), [Lzc06](https://github.com/Lzc06), [marsggbo](https://github.com/marsggbo), [mengzhibin](https://github.com/mengzhibin), [qixiuai](https://github.com/qixiuai), [SHUHarold](https://github.com/SHUHarold), [sptj](https://github.com/sptj).
+  - ModularNAS: Towards Modularized and Reusable Neural Architecture Search, A code library for various neural architecture search methods including weight sharing and network morphism.
+  - DNet: Network architecture search algorithms and Model Zoo that are affinity with Davinci chips.
+  - MF-ASC: Multi-Fidelity neural Architecture Search with Co-kriging.
 
 ## Introduction
 
@@ -19,13 +26,16 @@ Vega is an AutoML algorithm tool chain developed by Noah's Ark Laboratory, the m
 2. Industry-leading AutoML algorithms: provides Noah's Ark Laboratory's self-developed **[industry-leading algorithm(Benchmark)](./docs/benchmark.md)** and  **[Model Zoo](./docs/model_zoo.md)** to download the State-of-the-art (SOTA) model.
 3. Fine-grained network search space: The network search space can be freely defined, and rich network architecture parameters are provided for use in the search space. The network architecture parameters and model training hyperparameters can be searched at the same time, and the search space can be applied to Pytorch, TensorFlow and MindSpore.
 4. High-concurrency neural network training capability: Provides high-performance trainers to accelerate model training and evaluation.
-5. Multi-Backend: PyTorch, TensorFlow, MindSpore(trial)
+5. Multi-Backend: PyTorch(GPU), TensorFlow(GPU, Ascend 910), MindSpore(Ascend 910).
+6. Ascend platform: Search and training on the Ascend 910 and model evaluation on the Ascend 310.
 
 ## Algorithm list
 
 | Category | Algorithm | Description | reference |
 | :--: | :-- | :-- | :-- |
 | NAS | [CARS: Continuous Evolution for Efficient Neural Architecture Search](https://arxiv.org/abs/1909.04977) | Structure Search Method of Multi-objective Efficient Neural Network Based on Continuous Evolution | [ref](./docs/en/algorithms/cars.md) |
+| NAS | ModularNAS: Towards Modularized and Reusable Neural Architecture Search | A code library for various  neural architecture search methods including weight sharing and network morphism | [ref](./docs/en/algorithms/modnas.md) |
+| NAS | [MF-ASC](https://dl.acm.org/doi/10.1145/3292500.3330893) | Multi-Fidelity neural Architecture Search with Co-kriging | [ref](./docs/en/algorithms/mfasc.md) |
 | NAS | [NAGO: Neural Architecture Generator Optimization](https://arxiv.org/abs/2004.01395) | An Hierarchical Graph-based Neural Architecture Search Space | [ref](./docs/cn/algorithms/nago.md) |
 | NAS | SR-EA | An Automatic Network Architecture Search Method for Super Resolution | [ref](./docs/en/algorithms/sr_ea.md) |
 | NAS | [ESR-EA: Efficient Residual Dense Block Search for Image Super-resolution](https://arxiv.org/abs/1909.11409) | Multi-objective image super-resolution based on network architecture search | [ref](./docs/en/algorithms/esr_ea.md) |
@@ -48,53 +58,22 @@ Vega is an AutoML algorithm tool chain developed by Noah's Ark Laboratory, the m
 | Fully Train | [Beyond Dropout: Feature Map Distortion to Regularize Deep Neural Networks](https://arxiv.org/abs/2002.11022) | Neural network training (regularization) based on disturbance of feature map | [ref](./docs/en/algorithms/fmd.md) |
 | Fully Train | [Circumventing Outliers of AutoAugment with Knowledge Distillation](https://arxiv.org/abs/2003.11342v1) | Joint knowledge distillation and data augmentation for high performance classication model training, achieved 85.8% Top-1 accuracy on ImageNet 1k | Coming soon |
 
-## Obtaining and Installing
-
-Install Vega and the open source softwares that Vega depends on:
-
-```bash
-pip3 install --user noah-vega
-python3 -m vega.tools.install_pkgs
-```
-
-For more detail, please refer **[installation guide](./docs/en/user/install.md)**. If you want to deploy Vega in local cluster, see the **[deployment guide](./docs/en/user/deployment.md)** .
+## install
 
-## Usage Guide
-
-The Vega is highly modularized. You can configure the search space, search algorithm in a pipeline way. To run the Vega application is to load the configuration file and complete the AutoML process based on the configuration.
-Vega provides detailed operation examples for your reference. For details, see the **[examples](./docs/en/user/examples.md)** . Example of running CARS algorithm:
+Run the following commands to install Vega and related open-source software:
 
 ```bash
-cd examples
-python3 ./run_pipeline.py ./nas/cars/cars.yml -b pytorch
+pip3 install --user --upgrade noah-vega
 ```
 
-Therefore, before using the Vega, you need to fully understand the meaning of the configuration items. For details, see the **[Configuration Guide](./docs/en/user/config_reference.md)**.
-
-**Note:**
-
-Before running an example, you need to configure the directory where the dataset and pre-trained models are located in the algorithm configuration file. Please refer to **[Example Reference](./docs/en/user/examples.md)** .
-
-## Developer Guide
-
-The Vega framework components are decoupled, and each functional component is combined using the registration mechanism to facilitate function and algorithm extension. For details about the Vega architecture and main mechanisms, see the **[Developer Guide](./docs/en/developer/developer_guide.md)** .
-
-In addition, you can refer to the **[Quick Start Guide](./docs/en/developer/quick_start.md)** to implement a simple network search function and quickly enter the Vega application development through practice.
-
-During the development of the Vega application, the first problem is how to introduce the service data set to the Vega application. For details, see the **[Dataset Guide](./docs/en/developer/datasets.md)** .
-
-For different algorithms, you can refer doc **[Algorithm Development Guide](./docs/en/developer/new_algorithm.md)** . You can add the new algorithm to Vega step by step based on the example provided in this document.
-
-In most Automl algorithms, the search space is closely related to the network. We try to unify the definition of the search space so that the same search space can adapt to different search algorithms. This is called the **[Fine-Grained Search Space Guide](./docs/en/developer/fine_grained_space.md)** . Welcome to try it.
-
-Of course, this document cannot solve all the problems. If you have any questions, please feel free to provide feedback through the issue. We will reply to you and solve your problems in a timely manner.
+If you need to install the Ascend 910 training environment, please contact us.
 
 ## Reference List
 
 | object | refrence |
 | :--: | :-- |
-| User | [Install Guide](./docs/en/user/install.md), [Deployment Guide](./docs/en/user/deployment.md), [Configuration Guide](./docs/en/user/config_reference.md), [Examples](./docs/en/user/examples.md), [Evaluate Service](./docs/en/user/evaluate_service.md) |
-| Developer | [Developer Guide](./docs/en/developer/developer_guide.md), [Quick Start Guide](./docs/en/developer/quick_start.md), [Dataset Guide](./docs/en/developer/datasets.md), [Algorithm Development Guide](./docs/en/developer/new_algorithm.md), [Fine-Grained Search Space Guide](./docs/en/developer/fine_grained_space.md) |
+| [**User**<br>(User Guide)](./docs/en/user/README.md) | [Install Guide](./docs/en/user/install.md), [Deployment Guide](./docs/en/user/deployment.md), [Configuration Guide](./docs/en/user/config_reference.md), [Examples](./docs/en/user/examples.md), [Evaluate Service](./docs/en/user/evaluate_service.md) |
+| [**Developer**<br>(Developer Guide)](./docs/en/developer/README.md) | [Development Reference](./docs/en/developer/developer_guide.md), [Quick Start Guide](./docs/en/developer/quick_start.md), [Dataset Guide](./docs/en/developer/datasets.md), [Algorithm Development Guide](./docs/en/developer/new_algorithm.md), [Fine-Grained Search Space Guide](./docs/en/developer/fine_grained_space.md) |
 
 ## FAQ
 
@@ -113,11 +92,6 @@ For common problems and exception handling, please refer to [FAQ](./docs/en/user
 }
 ```
 
-## Video
-
-[video](https://box.saas.huaweicloud.com/p/fe9dfbd8b61a37280b371f73a90132c2)
-[Youtube](https://www.youtube.com/watch?v=Le5SGaW29TQ)
-
 ## Cooperation and contribution
 
 Welcome to use Vega. If you have any questions, ask for help, fix bugs, contribute algorithms, or improve documents, submit the issue in the community. We will reply to and communicate with you in a timely manner.

RELEASE.md

@@ -1,4 +1,4 @@
-**Vega ver1.2.0 released:**
+**Vega ver1.3.0 released:**
 
 **Introduction**
 
@@ -14,9 +14,7 @@ Vega is an AutoML algorithm tool chain developed by Noah's Ark Laboratory, the m
 
 Install Vega and the open source softwares that Vega depends on:
 
-`pip3 install noah-vega`
-
-`python3 -m vega.tools.install_pkgs`
+`pip3 install --user --upgrade noah-vega`
 
 **Cooperation and contribution**