!259 Add slides and notes of May 28th tech session of usability sig.

Merge pull request !259 from Zhang Tong/master

sigs/usability/docs/tech-sessions/20220528/README_CN.md

@@ -0,0 +1,61 @@
+# 5月28日易用性SIG技术分享活动精彩回顾
+
+![usability](./images/usability.png)
+
+2022年5月28日，MindSpore易用性SIG举行了第2次线上技术分享活动，主题是**《从零开始，“易”起上手》**，共有265位开发者朋友通过腾讯会议和B站直播参与了本次活动。
+
+此次活动新吸纳了87位开发者加入易用性SIG交流群。感谢开发者们的热情参与和各位老师的精彩分享，在此我们整理了此次活动的内容摘要以及提问和回答，希望大家满意！
+
+## 分享内容回顾
+
+本次活动的主题是《从零开始，“易”起上手》，5位演讲嘉宾主要围绕MindSpore初学者在初学和上手MindSpore时可能遇到的主要问题，从**安装**、**编程辅助**、**调试调优**、**模型迁移**等角度进行了精彩的分享。
+
+感谢张辉、Chissica、Miao、Louie、CQU弟中弟为大家带来的精彩分享！
+
+**1、《MindSpore漫游世界》**：MindSpore资深开发者张辉（在社区中被大家所熟知的张小白老师）通过生动有趣的讲解，为大家介绍了MindSpore各个平行世界（各组件）的奇异景色（功能特点）以及MindSpore在多平台上的安装方法，为大家打开了一个新的视角。
+
+![lecturer1](./images/lecturer1.jpg)
+
+**2、《智能辅助编程》**：MindSpore Dev Toolkit技术负责人、MindSpore布道师Chissica基于自己在代码推荐、代码搜索、注释生成领域的专长，为大家介绍了软件智能化开发中的关键技术——智能辅助编程的核心技术与使用场景，这些技术也已经落地到了MindSpore Dev Toolkit插件中，欢迎各位开发者去体验。
+
+![lecturer2](./images/lecturer2.jpg)
+
+**3、《AI框架功能调试的思考与MindSpore实践》**：两位MindSpore DFx专家Miao与Louie，对他们在AI框架功能调试领域的系统性思考进行了介绍，并基于MindSpore分享了在机器学习开发功能调试中常见的问题及其解决方法。
+
+![lecturer3](./images/lecturer3.jpg)
+![lecturer4](./images/lecturer4.jpg)
+
+**4、《MindSpore模型迁移经验分享》**：MindSpore易用性专家CQU弟中弟，以BERT为例，分享了从PyTorch/Tensorflow将模型迁移至MindSpore的相关方法和经验。
+
+![lecturer5](./images/lecturer5.jpg)
+
+## Q&A环节
+
+**Question（JeffDing）：炼丹师是不是只要会训练代码就可以（胜任）？**
+
+Answer（张辉）：AI开发中涉及很多知识技能点，每一个人根据自己的特长在AI框架的学习和使用上找到适合自己的路就可以了。比如我个人从安装入手来学习框架，可能别人在模型调参、AI for Science等方向入手会更合适。等学会一些基本的知识技能点以后，才能更近一步。
+
+**Question（B站听众）：MindSpore Dev Toolkit可以在哪里下载试用？**
+
+Answer（Chissica）：MindSpore代码仓中有一个子仓：https://gitee.com/mindspore/ide-plugin，可以在这里下载试用。
+
+**Question（B站听众）：将来是不是程序员说一个需求，AI就能直接实现了？**
+
+Answer（Chissica）：这是一个很远大和宏伟的目标，但目前仅能做到一部分，例如业界DeepMind做的AlphaCode，能够解一部分算法题，给予算法题的描述后就能给出可运行的代码。AI辅助编程目前仅能做到单领域的某些场景，还做不到那么理想的地步。个人理解，将来不一定能完全做到那么理想，因为在软件开发过程中，有一些创造性的活动，如需求分解等，目前让人工智能来做还达不到，起码还有5-10年发展的过程，但这仍然是一块非常值得探索的技术。
+
+**Question（Tong）：怎样去评估一个智能辅助编程系统的优劣/好坏，或说是成熟程度？**
+
+Answer（Chissica）：一方面是通过减少程序员敲击键盘的次数来衡量，程序员敲击键盘次数越少，说明AI的辅助编程能力越全。另一方面，程序员的修改越少，也说明推荐的越准确、效果越好。引申一下，未来如果有较强的AI辅助编程工具，能够直接推荐出来可执行的程序，可以说是达到了一个巅峰的状态。
+
+## 资料获取
+
+1、本次活动的演讲材料将发布到MindSpore社区，请点击[链接](https://gitee.com/mindspore/community/tree/master/sigs/usability)获取。
+2、本次活动的视频将上传到B站MindSpore官方账号，敬请关注。
+
+---
+
+加入易用性SIG交流群，共同打造易学易用、灵活高效的AI框架！
+
+加入方式：添加小助手的微信（vx: mindspore0328），添加时请备注：易用性。
+
+小助手拉你进群哦！

sigs/usability/docs/tech-sessions/20220528/README_EN.md

@@ -0,0 +1,68 @@
+# Review of Usability SIG Tech Session on May 28th
+
+![usability](./images/usability.png)
+
+On May 28th, 2022, MindSpore Usability SIG held its second online technology sharing event, with the theme of "Start from Scratch, Start with Easy". A total of 265 developers participated in the event through Tencent conference and B station live broadcast.
+
+This event brought 87 new developers to the Usability SIG Exchange Group. Thank you for your enthusiastic participation and wonderful sharing. Here we have compiled the summary of the activity as well as the questions and answers. Hope you are satisfied!
+
+## Review of Lectures
+
+The theme of this activity is 'Starting from scratch, Easy to Get started', 5 speakers mainly focused on the main problems that MindSpore beginners may encounter when learning and starting MindSpore, from the perspectives of installation, programming assistance, debugging and tuning, model migration and so on.
+
+Thanks to Zhang Hui, Chissica, Miao, Louie and CQU dizhongdi for your wonderful sharing!
+
+**1. MindSpore Roaming the World**
+
+Senior MindSpore developer Zhang Hui (known as Teacher Zhang Xiaobai in the community) opened a new perspective through vivid and interesting explanations, introducing the strange scenery (functional features) of various parallel worlds (components) of MindSpore as well as the installation method of MindSpore on multiple platforms.
+
+![lecturer1](./images/lecturer1.jpg)
+
+**2. Intelligent Assisted Programming**
+
+Chissica, technical director of MindSpore Dev Toolkit and MindSpore sermon, based on his expertise in the fields of code recommendation, code search and annotation generation, introduced the key technology in intelligent software development -- the core technology and application scenarios of intelligent aided programming. These technologies have also been implemented in the MindSpore Dev Toolkit plug-in, and you are welcome to experience them.
+
+![lecturer2](./images/lecturer2.jpg)
+
+**3. Thoughts on AI Framework Function Debugging and MindSpore Practice**
+
+Miao and Louie, two MindSpore DFx experts, introduced their systematic thinking in the field of AI framework function debugging, and shared common problems and solutions in machine learning development function debugging based on MindSpore.
+
+![lecturer3](./images/lecturer3.jpg)
+![lecturer4](./images/lecturer4.jpg)
+
+**4. MindSpore Model Migration Experience Sharing**
+
+Cqudi Zhongdi, a MindSpore ease-of-use expert, takes BERT as an example to share the methods and experience related to model migration from PyTorch/Tensorflow to MindSpore.
+
+![lecturer5](./images/lecturer5.jpg)
+
+## Q&A
+
+**Question (JeffDing) : Is an alchemist just able to train code? **
+
+Answer (Zhang Hui) : AI development involves a lot of knowledge and skills. Everyone should find their own way to learn and use AI framework according to their own strengths. For example, if I personally learn framework from installation, maybe someone else would be more appropriate to start with model tuning, AI for Science, etc. After learning some basic knowledge and skills, you can take a step closer.
+
+**Question: Where can I download the MindSpore Dev Toolkit for trial? **
+
+Answer: (Chissica) MindSpore code warehouse with a child in the warehouse: https://gitee.com/mindspore/ide-plugin, you can download the trial here.
+
+**Question (audience at STATION B) : In the future, will a programmer say a requirement and the AI implement it directly? **
+
+Answer (Chissica) : This is a very big and ambitious goal, but only part of it can be achieved at present. For example, AlphaCode made by DeepMind in the industry can solve part of the algorithm problems, and give the description of the algorithm problem to give the executable code. Ai-assisted programming is currently only able to achieve certain scenarios in a single field, but not so ideal. Personally, in the future, it may not be completely ideal, because in the process of software development, there are some creative activities, such as demand decomposition, which cannot be done by artificial intelligence at present. At least it will take 5-10 years to develop, but it is still a technology worth exploring.
+
+**Question (Tong) : How to evaluate the merits/demerits/maturity of an intelligent auxiliary programming system? **
+
+Answer (Chissica) : On the one hand, it can be measured by reducing the number of keyboard strokes by programmers. The less keyboard strokes by programmers, the more complete the auxiliary programming ability of AI is. On the other hand, the fewer changes programmers make, the more accurate and effective the recommendations are. By extension, in the future, if there are strong AI-aided programming tools that can directly recommend executable programs, it can be said that they have reached a peak state.
+
+## Related Materials
+
+1. The activity of speech materials will be released to the MindSpore community, please click on the [link](https://gitee.com/mindspore/community/tree/master/sigs/usability).
+2. The video of this activity will be uploaded to the official account of MindSpore at site B, please pay attention.
+
+---
+Join the SIG communication group to build an easy-to-learn, flexible and efficient AI framework together!
+
+Way to join: add the wechat of the assistant (wechat: mindspore0328). Please note 'Usability SIG' when you are adding.
+
+Assistant will invite you into the group!