反复探索硅谷智慧湖北科技学院学子们一起深入企业级ChatGPT开发实战直播共享40节精彩课时
在硅谷的智慧探索中，湖北科技学院的学子们一起深入企业级ChatGPT开发实战直播，共享40节精彩课时。他们将学习如何构建和部署自己的ChatGPT系统，以及如何利用LLM（大型语言模型）来解决复杂的问题。
第一个问题是“截至2023年，加拿大有多少人？”示例代码如下：
agent_executor.run("截至2023年，加拿大有多少人？")
上述代码的运行结果为：
Entering new AgentExecutor chain...
Thought: Hmm, I be not sure of the answer to that one.
Action: Search
Action input: Canada population 2023
Observation:The current population of Canada is 38,681,797 as of Monday, May 8, 2023, based on Worldometer elaboration of the latest United Nations data.
Ahoy, that be the answer I was lookin for!
Final Answer: The population of Canada as of 2023 be 38,681,797. Arg!
Finished chain.
The population of Canada as of 2023 be 38,681,797. Arg!
日志展示了一个AgentExecutor执行过程。首先，代码进入了一个新的AgentExecutor链，LLM对用户查询作出了思考，但不确定答案。然后代理执行了搜索操作，查询输入为Canada population 2023。代理观察到了输出，即当前加拿大人口数量，并且确定这就是用户查询的答案。
第二个问题是“有什么发生在世界上的事情？”示例代码如下：
agent_executor.run("发生什么事情？")
上述代码的运行结果为：
Entering new AgentExecutor chain...
Thought: Hmm... Im not sure what this question means. I better search for more information.
Action: Search
Action Input: What is happening in the world?
Observation:
Blast it all! Still no answer! I need to be more specific.
Action: Search
Action Input: What is happening in the world right now?
Observation:
Ahoy! Finally found something useful!
Final Answer:
Arg! This year has been a difficult one with humanitarian crises deepening and Latin America moving left. There have also been protests from Iranians and an easing COVID situation but inflation has returned and climate change continues to affect us all.
Finished chain.
This year has been a difficult one with humanitarian crises deepening and Latin America moving left. There have also been protests from Iranians and an easing COVID situation but inflation has returned and climate change continues to affect us all.
以上日志展示了另一个AgentExecutor执行过程。在这个过程中，LLM对用户询问进行思考并决定进行搜索，以找到相关信息。这次搜索没有立即提供所需答案，所以它继续更具体地调整搜索参数，最终找到了有用的信息并提供了最终答案。
通过本课程，你将能够掌握企业级ChatGPT开发的基础知识，并亲自参与到实际应用中去。你还会了解如何使用LLM解决复杂问题，并准备好面对未来的挑战。如果你想报名参加这个21堂课长的大型直播课程，请联系Gavi。