Just on the morning of 11.10, ChatGPT has secretly opened the GPTs function to all distinguished Plus users.
With the opening of this wave of functions, the interface has also changed a lot. Click Explore on the left or User in the lower left corner to directly enter the new GPTs function:
Here you can see the GPT we created ourselves, and below are the GPTs officially produced by OpenAI:
However, these official ones currently look more like prompt word packages, and they don’t seem to be anything amazing. After all, who doesn’t have a prompt word market now?
After clicking Create a GPT, you will enter the creation page of GPTs, with GPT Builder on the left and real-time preview on the right:
The Builder on the left is very interesting. It can create our own GPT in real time in an interactive form.
Just like the C language compiler gcc is written in C language, now we can use ChatGPT to create ChatGPT for us, and we can create it interactively, which is very cool!
Here is a GPT for knowledge base Q&A, using awesome-java, a 38K star warehouse, as the basic data:
After a few seconds, GPT has helped me prepare the GPT. You can preview it directly on the right, and even gave me a thoughtful name for this GPT: Java Resource Guide.
If you are not satisfied with the content, you can continue to communicate and make adjustments with GPT on the left, including uploading text, documents and other materials as information.
OK, now we have a prototype. Test the effect in the preview window on the right:
GPT very considerately generated several initial conversations for me to facilitate users to quickly understand the role of GPT.
I asked a random question about the CLI tool, and GPT directly listed the CLI library list in the warehouse, which is very accurate:
If you are not satisfied with the results of this interactive generation, you can also perform precise configuration in the Configure panel, including uploading your database documents (currently supports up to 10)!
When uploading files, all mainstream document formats should be supported, such as plain text (txt), Office, PDF, etc. However, scanned PDFs should not be supported yet, and OCR seems to be relatively weak now, even with the help of the DALL-E model.
In addition to the basic functions above, there is also an advanced setting for Action:
Here, you can also call external interfaces, which can later be connected to external data platforms and various SAAS platforms.
For example...connect to zapier and support a bunch of Saas services:
If you are a pure Internet person, maybe you can really create your own Jarvis Assistant and experience the joy of Iron Man! .
But... this function seems to be missing from the documentation. There is no detailed introduction to parameter analysis and interaction methods. Moreover, the function is currently weak. There seems to be no place to configure the default parameters. It seems not mature enough. You need to give it some time. .
After the creation is completed, click the Save button in the upper right corner to publish the work we created:
The third Public option seems to mean publishing to the GPT Store. Unfortunately, the GPT Store is not online yet, and errors are often reported when saving.
However, according to official news, GPT Strore will be opened at the end of this month (November). Looking forward to seeing what useful GPT will appear in due course.
The second Only people with a link can be saved and shared with others through a link. This link is currently also used by the third-party GPT Store.
Although the official Store is not online, the three parties now have thousands of GPTs released by users (so far), and they are still growing. It is really a hundred flowers blooming:
Interested students can go in and have a look, see what new things are available, and experience the imagination of GPT enthusiasts.
Now custom GPT + custom database file + online search, ChatGPT officially provides them all, and you can also generate them yourself. Those ChatPDFs, big legal models, and XX industry knowledge bases may really have no hope of survival in the future.
Okay, now that the first experience is over, let’s talk about my opinions.
GPTs currently looks more like an interactive Prompt Creator, or a lightweight plugin with limited uses. If you want to truly use it as a "productivity", you may still need to rely on the development capabilities of the plugin to customize data, query, etc., and then combine it with OpenAI's vision and tts functions to become a multi-functional, customizable assistant and better Used in various fields.
OpenAI opens ChatGPT to all users for free. Voice programmers tampered with ETC balances and embezzled more than 2.6 million yuan a year. Spring Boot 3.2.0 was officially released. Google employees criticized the big boss after leaving the company. He was deeply involved in the Flutter project and formulated HTML-related standards. Microsoft Copilot Web AI will be Officially launched on December 1st, supporting Chinese Microsoft's open source Terminal Chat Rust Web framework Rocket releases v0.5: supports asynchronous, SSE, WebSockets, etc. The father of Redis implements the Telegram Bot framework using pure C language code . If you are an open source project maintainer, encounter How far can you endure this kind of response? PHP 8.3 GAAuthor: JD Insurance Jiang Xin
Source: JD Cloud Developer Community Please indicate the source when reprinting