NeuHelium > Information Insight > Events > [4.4 LIVE] NH Tech Salon 27- Creating ChatGPT Solutions for Applications 2023/4/4
[4.4 LIVE] NH Tech Salon 27- Creating ChatGPT Solutions for Applications

Microsoft Senior Cloud Technology Evangelist, former Microsoft's Most Valuable Expert, former Microsoft Technology Community Regional Director, specializing in artificial intelligence, cloud native, and front-end application development technologies, providing different artificial intelligence technology solutions for education, finance, and telecommunications. Microsoft Artificial Intelligence Hacker Song Coach, lecturer at multiple Microsoft Ignite conferences, and guest speaker at multiple Python and. NET, AI related conferences. Release the "Artificial Intelligence Introduction to Practice Training Camp" and the "OpenAI Technology Application Course" for education in NetEase Cloud Classroom.



We live in an era of AI explosion and ChatGPT has recently become a hot topic.


The current popular GPT-3 is the third generation generative pre-trained Transformer model by Open AI. With deep learning, it predicts the next token using a large language prediction model.  Nowadays, many people use GPT3 to create poetry, images, and more. When we thought GPT3 was powerful, GPT4 had already been released. It not only improved the grammar algorithm for training, but also achieved good results on deep text.



GPT brings a new paradigm. Although it has a pre-trained model, it does some manual tasks when processing. When it comes to understanding tasks, it doesn't require us to train anymore, but can directly use this model.



How to do the Prompt project: apply the latest models to achieve the best results; Place the explanation at the beginning of the prompt, And use # # # or "" to separate the description and context; be as specific, descriptive, and detailed as possible for the desired background, results, length, format, style, etc.; clarify the desired output format through examples; start with zero shot, then feed shot, both of which do not work, and then fine tune; reduce "fluffy" and imprecise descriptions; rather than saying what should not be done, say what should be done; specific to code generation - use The 'guiding words' push the model towards a specific pattern.



Speaker Introduction

Kinfey Lo        Microsoft Senior Cloud Advocate

Microsoft Senior Cloud Technology Evangelist, former Microsoft's Most Valuable Expert, former Microsoft Technology Community Regional Director, specializing in artificial intelligence, cloud native, and front-end application development technologies, providing different artificial intelligence technology solutions for education, finance, and telecommunications. Microsoft Artificial Intelligence Hacker Song Coach, lecturer at multiple Microsoft Ignite conferences, and guest speaker at multiple Python and. NET, AI related conferences. Release the "Artificial Intelligence Introduction to Practice Training Camp" and the "OpenAI Technology Application Course" for education in NetEase Cloud Classroom.

 

 

 

Live Review:https://www.bilibili.com/video/BV1CL41127My/