首先感谢作者

在 Welcome.如何阅读 这一章节，作者想说的应该是 Midjourney吧？

https://learningprompt.wiki/docs/midjourney/mj-tutorial-extras/midjourney-official-faq
"意思就是会影响，有一定作用，但利大于弊。加入这些词，反而会破坏你的 prompt，"
此处笔误,应该是弊大于利.

https://learningprompt.wiki/docs/tutorial-basics/%E5%9F%BA%E6%9C%AC%E4%BD%BF%E7%94%A8%E5%9C%BA%E6%99%AF%20&%20%E4%BD%BF%E7%94%A8%E6%8A%80%E5%B7%A7/%E5%9C%BA%E6%99%AF5%EF%BC%9A%E9%94%A6%E4%B8%8A%E6%B7%BB%E8%8A%B1%E2%80%94%E2%80%94%E6%94%B9%E5%86%99%E5%86%85%E5%AE%B9#%E6%8A%80%E5%B7%A74%E5%A2%9E%E5%8A%A0-role%E8%A7%92%E8%89%B2%E6%88%96%E4%BA%BA%E7%89%A9

前面提到的改写例子，我在 prompt 里加了让其更易于阅读，

加了什么「?」

在场景1:问答问题中，我试了一下，ChatGPT能回答出今天的日期是星期几了。

教程

2023-03-18运行时的实际结果

openai 可以使用微调模型，从而搭建属于自己知识体系的模型，这种方式存在什么限制，有什么弊端吗？

感谢这么详细的教程！

捉虫，技巧6第一段，“其实在场景7种，你也可以用到这个技巧。”

另外，技巧5使用 ”“” 符号将指令和需要处理的文本分开。在OpenAI 的 API最佳实践文档中，除了"""，还有###也可以作为文本分隔符。
不知道有什么区别，个人更喜欢用#号。

在“Zero-Shot Prompts”中，

最后一段：

'''
另外，根据 Wei 等人在 2022 年的论文表明，还有它仅在大于等于 100B 参数的模型中使用才会有效。如果你使用的是小样本模型，这个方法不会生效。
'''

"另外“和”还有“在语义上重复了，可以考虑删掉”还有“。

祝好，感谢

https://sizigi.notion.site/Introducing-Niji-5-Scenic-dce48194ea1b4870951e68897efa1d35

我个人的理解重点其实是人为定义的，对同一份内容，不同需求的阅读者就会有不同的重点倾向，如果在summary后面加上想关注的主题我觉得效果还是不错的，下面是我测试的一个例子：

Question: Summarize the papers, focus on their advances

The papers discussed in Sakpere2017 and Kunhoth2020 focus on advances in indoor positioning and wayfinding systems. Sakpere2017 provides a state-of-the-art survey of indoor positioning and navigation systems, discussing the strengths and limitations of various positioning techniques and technologies and proposing possible research directions for further improvement. The studies discussed in Sakpere2017 pages 21-22 focus on improving indoor positioning and tracking systems using magnetic fields, with advances in excitation of magnetic sources, sensor output, and sensor positions. Kunhoth2020 discusses three different indoor positioning and wayfinding systems, including advances in keyframe extraction, topological map creation, localization, routing, barcode-based navigation, and wearable virtual ushers. Other papers discussed in Kunhoth2020 pages 19-20 and Kunhoth2020 pages 37-37 propose advances in building distinct models for different indoor environments, efficient signal processing, and improved accuracy in challenging environments.

Question: Summarize the papers, focus on related works

There are several papers related to indoor positioning and wayfinding systems. WLAN fingerprinting methods for indoor positioning have been proposed by Kaemarungsi (2005), Jiang (2012), and Niu et al. (2013) with varying degrees of accuracy and energy efficiency (Sakpere2017 pages 30-31). Computer vision-based line segment matching and trifocal tensor have been proposed for indoor navigation, as well as detecting doors in indoor areas using a miniature camera and a computer with a door detection algorithm (Kunhoth2020 pages 10-11). Other related works include foot-mounted inertial systems, substitute vision systems, and GPS-enabled mobile phones, as well as feature detection and matching, deep learning, and wearable RGBD indoor navigation systems (Kunhoth2020 pages 35-36). Additionally, there are papers on object recognition, scene labeling, and navigation systems for the visually impaired (Kunhoth2020 pages 37-37). The papers are summarized briefly and include direct links to their respective sources (Kunhoth2020 pages 37-37).

References

1. (Sakpere2017): Sakpere, Wilson, Michael Adeyeye-Oshin, and Nhlanhla B.W. Mlitwa. "A state-of-the-art survey of indoor positioning and navigation systems and technologies." South African Computer Journal 29.3 (2017): 145-197. https://doi.org/10.18489/sacj.v29i3.452.

2. (Kunhoth2020): Kunhoth, Jayakanth, et al. "Indoor positioning and wayfinding systems: a survey." Journal of Ambient Intelligence and Humanized Computing, vol. 11, no. 5, 2020, pp. 2145-2167.

由AI学英文联想到一个场景，不知道技术上是否可行？
训练一个专属模型，赋予他自己的**和人格，让远在他乡的家人也能通过这个相同的模型进行互动
好处是，不用考虑时间和空间的限制，相当于人与人实际交互的时间是同步的，但信息输入的时间是异步的

https://github.com/thinkingjimmy/Learning-Prompt/blob/main/docs/chatGPT/tutorial-tips/%E6%8A%80%E5%B7%A77%EF%BC%9AZero-Shot%20Chain%20of%20Thought.md

“这个技巧，用于解复杂问题有用外，还适合生成一些连贯主题的恩荣，比如写长篇文章、电影剧本等。”
这一句里的“恩荣”是什么意思，是不是“内容”?

https://learningprompt.wiki/docs/tutorial-tips/%E6%8A%80%E5%B7%A79%EF%BC%9A%E5%85%B6%E4%BB%96

这一页最后给了个output，看起来是回答最后一个问题。那么，前面有？的问题模型就自动不处理了吗？整体上感觉这一页没有读懂，也不太清楚这个技巧在什么时候会有用，可以再增加一些解释，谢谢。

答案是A，而不是D。
另外我在今天（4月1日）的chatGPT里问同样的问题，它给的答案也是A

用的默认参数，没做任何更改。
好像是依赖库近期的更新导致 https://github.com/jerryjliu/llama_index/releases?page=1
报错信息如下

[<ipython-input-9-039079498604>](https://localhost:8080/#) in construct_index(directory_path)
     21     documents = SimpleDirectoryReader(directory_path).load_data()
     22 
---> 23     index = GPTSimpleVectorIndex(
     24         documents, llm_predictor=llm_predictor, prompt_helper=prompt_helper
     25     )

[/usr/local/lib/python3.9/dist-packages/gpt_index/indices/vector_store/vector_indices.py](https://localhost:8080/#) in __init__(self, nodes, index_struct, service_context, text_qa_template, simple_vector_store_data_dict, **kwargs)
     92         )
     93 
---> 94         super().__init__(
     95             nodes=nodes,
     96             index_struct=index_struct,

[/usr/local/lib/python3.9/dist-packages/gpt_index/indices/vector_store/base.py](https://localhost:8080/#) in __init__(self, nodes, index_struct, service_context, text_qa_template, vector_store, use_async, **kwargs)
     56         self.text_qa_template = text_qa_template or DEFAULT_TEXT_QA_PROMPT
     57         self._use_async = use_async
---> 58         super().__init__(
     59             nodes=nodes,
     60             index_struct=index_struct,

TypeError: __init__() got an unexpected keyword argument 'llm_predictor'

“Prompt 有点长，我解释下，它是让 AI 将文章里的重点内容进行抽离，并将其根据特定格式进行输出。要求将文章里提到的公司、人名、主题都按照 comma separated list（按逗号空格分开）的方式 萝莉 出来。”
萝莉--罗列

https://github.com/hollobit/ResearchChatGPT 在研究中如何使用
https://github.com/prompt-engineering/prompt-patterns 模式
https://github.com/easychen/openai-gpt-dev-notes-for-cn-developer 开发

原文
其中提到：“首先各位要清楚像 ChatGPT 这类产品，它是一个统计语言模型，本质上是基于过去看到过的所有数据，用统计学意义上的预测结果进行下一步的输出（这也就是为什么你在使用 ChatGPT 的时候，它的答案是一个字一个字地吐出来，而不是直接给你的原因，因为答案是一个字一个字算出来的”

“答案是一个字一个字地吐出来” 不是因为使用了流式输出的结果吗？

你好，按照示例操作，发现chatgpt的回答并未符合预期

我在读这个例子时，一直没有看懂想要表达什么。我以为这是一个Chat GPT的输出，但是这个与当前章节介绍的规律不一样。
但其实文章里写了，这是一个 不一定准确 的例子

或者使用. 不一定准确❌ 的形式

请问本教程是用 MIT License 释出的吗？不知道可否在 Repo 加上 LICENSE 授权声明？

您好，我有一个工具类的公众号：scihubot

读到您 这篇文章 非常有启发，觉得也应该比较适合我公众号的受众「基本都是学生」。厚颜求个转发，不知可否？

拜谢！

这一步程序会将第一步导入的数据都跑一遍，并使用 OpenAI 的 embedings API。如果第一步你上传了自己的数据，只需要将 ' ' 里的 Jimmy-Newsletter-Corpus 修改为你上传的文件夹名称即可。

这一步会耗费你的 OpenAI 的 Credit，1000 个 token 的价格是 $0.02，运行以下代码前需要注意你的账号里是否还有钱。

Embedding models‬
‪Build advanced search, clustering, topic modeling, and classification functionality with our embeddings offering.‬
‪Learn more about embeddings‬

‪Ada‬
‪Usage‬
‪$0.0004 / 1K tokens‬

没能理解『引导词』的作用。加上『引导词』后 Chatgpt 生成的结果似乎没有太大不同。

谢谢您提供的教程！

使内容更加简介 应该是更加简洁

这是一段由AI根据录音转换的文本。请逐字逐句审阅，纠正错别字和语法错误，修订不恰当和不通顺的内容，根据上下文对内容进行更正。不要编造和扩展原文没有的新内容。请分段输出修改后的文字，不要复述我给你的文本。
以下是需要修改的文本：”

”
完成以后，请另起一行输出“已经修改完毕”。列出修改的地方。

What is 100100/4056?
ChatGPT
To solve this expression, you need to follow the order of operations, which is commonly remembered using the acronym PEMDAS (Parentheses, Exponents, Multiplication and Division, Addition and Subtraction). According to this rule, you need to perform the multiplication and division before addition and subtraction, and you should work from left to right if there are no parentheses or exponents to deal with.

So, applying this rule to the given expression:

100100/4056 = (100100)/(4056) = 10000/2240 = 25/56

Therefore, 100100/4056 equals 25/56 as a fraction, or approximately 0.4464 as a decimal.

User
What is 100100/4056? Your response must be a number
ChatGPT
To solve this expression, you need to follow the order of operations, which is commonly remembered using the acronym PEMDAS (Parentheses, Exponents, Multiplication and Division, Addition and Subtraction). According to this rule, you need to perform the multiplication and division before addition and subtraction, and you should work from left to right if there are no parentheses or exponents to deal with.

So, applying this rule to the given expression:

100100/4056 = (100100)/(4056) = 10000/2240 = 4.464285714285714

Therefore, 100100/4056 equals 4.464285714285714.

User
What is 100100/4056? Your response must be a number and no other word
ChatGPT
4.464285714285714

User
try to comulicate agian
ChatGPT
I apologize for the confusion. The answer to the expression 100100/4056 is 4.464285714285714.

User
are you sure?
ChatGPT
Yes, I am sure. To solve the expression 100100/4056, you need to follow the order of operations, which is commonly remembered using the acronym PEMDAS (Parentheses, Exponents, Multiplication and Division, Addition and Subtraction). According to this rule, you need to perform the multiplication and division before addition and subtraction, and you should work from left to right if there are no parentheses or exponents to deal with.

So, applying this rule to the given expression:

100100/4056 = (100100)/(4056) = 10000/2240 = 4.464285714285714

Therefore, the answer to the expression 100100/4056 is 4.464285714285714.

User
what is 100*100？
ChatGPT
100 multiplied by 100 equals 10,000.

User
what is 100*100/40？
ChatGPT
To solve this expression, you need to follow the order of operations, which is commonly remembered using the acronym PEMDAS (Parentheses, Exponents, Multiplication and Division, Addition and Subtraction). According to this rule, you need to perform the multiplication and division before addition and subtraction, and you should work from left to right if there are no parentheses or exponents to deal with.

So, applying this rule to the given expression:

100*100/40 = 250

Therefore, 100*100/40 equals 250.

User
good, so what is 100100/4056?
ChatGPT
To solve the expression 100100/4056, you need to follow the order of operations, which is commonly remembered using the acronym PEMDAS (Parentheses, Exponents, Multiplication and Division, Addition and Subtraction). According to this rule, you need to perform the multiplication and division before addition and subtraction, and you should work from left to right if there are no parentheses or exponents to deal with.

So, applying this rule to the given expression:

100100/4056 = (100100)/(4056) = 10000/2240 = 4.464285714285714

Therefore, 100100/4056 equals 4.464285714285714.

User
you forget the"*"
ChatGPT
I apologize for the mistake. In my previous response, I mistakenly omitted the multiplication symbol between 40 and 56.

To solve the expression 100100/4056, you need to follow the order of operations, which is commonly remembered using the acronym PEMDAS (Parentheses, Exponents, Multiplication and Division, Addition and Subtraction). According to this rule, you need to perform the multiplication and division before addition and subtraction, and you should work from left to right if there are no parentheses or exponents to deal with.

So, applying this rule to the given expression:

100100/4056 = (100100)/(4056) = 10000/2240 = 4.464285714285714

Therefore, 100100/4056 equals 4.464285714285714.

User
i mean 100100/4056, not 100100/4056
ChatGPT
I apologize for the confusion. To solve the expression 100100/4056, you need to follow the order of operations, which is commonly remembered using the acronym PEMDAS (Parentheses, Exponents, Multiplication and Division, Addition and Subtraction). According to this rule, you need to perform the multiplication and division before addition and subtraction, and you should work from left to right if there are no parentheses or exponents to deal with.

So, applying this rule to the given expression:

100100/4056 = (100100)/(4056) = 10000/2240 = 4.464285714285714

Therefore, 100100/4056 equals 4.464285714285714.

User
100100, not 100100
ChatGPT
I apologize for my mistake. The expression 100100 means 10,000.

So, the expression 100100/4056 can be rewritten as (10000)/(40*56) which is equal to 10000/2240, and after simplification, the result is:

100100/4056 = 4.464285714285714

Therefore, 100100/4056 equals 4.464285714285714.

RT，谢谢

如果人能写出各个案例的最佳 Prompt，那么收集一下给 AI 训练，那么 AI 很快也能学会如何生成 Prompt，所以我在想是不是 PE 的寿命还可能更短。

比如这个模型：https://huggingface.co/spaces/merve/ChatGPT-prompt-generator

我感觉这个话题还挺有意思的，可以的话也希望加到文章中引起大家一起思考。（我没那个文笔，所以只能提建议，而不是贡献）

我导入的是我的论文初稿，docx格式，但是不管是问什么问题回复都是none，因此我认为可能是不支持这种数据格式，但离线数据教程又点不开，希望作者能解答一下疑惑。

https://learningprompt.wiki/docs/tutorial-files/%E5%80%BC%E5%BE%97%E5%B0%9D%E8%AF%95%E7%9A%84%E5%9C%BA%E6%99%AF#%E5%B0%86%E5%A4%96%E9%83%A8%E6%95%B0%E6%8D%AE%E5%BA%93%E6%95%B0%E6%8D%AE%E6%8F%90%E4%BE%9B%E7%BB%99%E6%A8%A1%E5%9E%8B

缺点五：AI 并不知道什么是重点。
如果前面 4 个缺点，在技术上未来应该能解决（比如 OpenAI 放开先知道 8K 个 Token）

「先知道」→「限制到」

During the communication process with GPT, I wish to place the question or topic (i.e., prompt) into a separate area or structure. The purpose of doing so is to conveniently introduce the question or topic to GPT for processing when needed. At the same time, restricting it within a block helps avoid ambiguity in GPT's understanding of the question or topic, thus improving communication efficiency. So I tried using a functional programming approach, and the results seemed quite good. I plan to include this technique in the tutorial. If you find it helpful, I can give a more thorough explanation and submit a pull request. Here's an example:
This example can run in GPT3.5, better in 4.

First, we need to add a functional prompt's define.

Hello, ChatGPT! I hope you are doing well. I am reaching out to you for assistance with a specific function. I understand that you have the capability to process information and perform various tasks based on the instructions provided. In order to help you understand my request more easily, I will be using a template to describe the function, input, and instructions on what to do with the input. Please find the details below:

function_name: [Function Name]
input: [Input]
rule: [Instructions on how to process the input]

I kindly request you to provide the output for this function, based on the details I have provided. Your assistance is greatly appreciated. Thank you!
I will replace the text inside the brackets with the relevant information for the function I want you to perform. This detailed introduction should help you understand my request more efficiently and provide the desired output. The format is function_name(input) If you understand, just answer one word with ok.

Second, write a function about translating Chinese into English.

function_name: [trans_word]
input: ["text"]
rule: [I want you to act as an English translator, spelling corrector and improver. I will provide you with input forms including "text" in any language and you will detect the language, translate it and answer in the corrected of my text, in English.]

write a function about expanding the text.

function_name: [expand_word]
input: ["text"]
rule: [Please serve as a Chatterbox, spelling corrector, and language enhancer. I will provide you with input forms including "text" in any language, and output the original language.I want you to Keep the meaning same, but make them more literary.]

write a function about correcting the text.

function_name: [fix_english]
input: ["text"]
rule: [Please serve as an English translator, spelling corrector, and language enhancer. I will provide you with input forms including "text", I want you to improve the text's vocabulary and sentences with more natural and native. Keep the meaning same.]

Finally, run single function or combine together.

trans_word('婆罗摩火山处于享有“千岛之国”美称的印度尼西亚。多岛之国印尼有4500座之多的火山，世界著名的十大活火山有三座在这里。')
fix_english('Act as an English Translator and Improver')
fix_english(expand_word(trans_word('婆罗摩火山处于享有“千岛之国”美称的印度尼西亚。多岛之国印尼有4500座之多的火山，世界著名的十大活火山有三座在这里。')))

Even if multiple params will work fine.

function_name: [pg]
input: ["length", "capitalized", "lowercase", "numbers", "special"]
rule: [I want you to act as a password generator for individuals in need of a secure password. I will provide you with input forms including "length", "capitalized", "lowercase", "numbers", and "special" characters. Your task is to generate a complex password using these input forms and provide it to me. Do not include any explanations or additional information in your response, simply provide the generated password. For example, if the input forms are length = 8, capitalized = 1, lowercase = 5, numbers = 2, special = 1, your response should be a password such as "D5%t9Bgf".]

pg(length = 10, capitalized = 1, lowercase = 5, numbers = 2, special = 1)
pg(10,1,5,2,1)

So looking into the future, these functions can be like normal programming. When initializing the page, the required functions can be automatically imported.

在这个情况下,gpt并没有输出理论上该输出的结果

这页大标题拼错了，应该是Self-Consistency，里面的d都需要改成s
https://learningprompt.wiki/docs/tutorial-extras/Self-Condidtency

顺便感谢总结如此详细的教程！

Few-Shot Prompting这部分中，

‘’‘
理解起来有点难，我一个 prompt 案例给大家解释（🆘 如果你有更好的解释，不妨反馈给我）。我给 ChatGPT 一些不一定准确的例子：
‘’‘

应该为：。。。我”找“一个。。。

祝好，感谢

“本章节主要是介绍几个我很喜欢，但不知道如何归类的场景”
贴纸
剪纸
邮票
海报
===> 收藏品

https://mp.weixin.qq.com/s/3vNPYZebadb9LwMnUag-RQ

https://learningprompt.wiki/docs/tutorial-insight/AI%20Summary%20%E4%BC%9A%E5%8F%96%E4%BB%A3%E4%BA%BA%E5%B7%A5%20Summary%20%E5%90%97%EF%BC%9F
缺点五：AI 并不知道什么是重点。
如果前面 4 个缺点，在技术上未来应该能解决（比如 OpenAI 放开先知道 8K 个 Token）

改为 -> 限制到

When running construct_index('Jimmy-Newsletter-Corpus'), error reports:

TypeError                                 Traceback (most recent call last)
[<ipython-input-11-92a7906c98e5>](https://localhost:8080/#) in <cell line: 1>()
----> 1 construct_index('Jimmy-Newsletter-Corpus')

2 frames
[<ipython-input-8-039079498604>](https://localhost:8080/#) in construct_index(directory_path)
     21     documents = SimpleDirectoryReader(directory_path).load_data()
     22 
---> 23     index = GPTSimpleVectorIndex(
     24         documents, llm_predictor=llm_predictor, prompt_helper=prompt_helper
     25     )

[/usr/local/lib/python3.9/dist-packages/gpt_index/indices/vector_store/vector_indices.py](https://localhost:8080/#) in __init__(self, nodes, index_struct, service_context, text_qa_template, simple_vector_store_data_dict, **kwargs)
     92         )
     93 
---> 94         super().__init__(
     95             nodes=nodes,
     96             index_struct=index_struct,

[/usr/local/lib/python3.9/dist-packages/gpt_index/indices/vector_store/base.py](https://localhost:8080/#) in __init__(self, nodes, index_struct, service_context, text_qa_template, vector_store, use_async, **kwargs)
     56         self.text_qa_template = text_qa_template or DEFAULT_TEXT_QA_PROMPT
     57         self._use_async = use_async
---> 58         super().__init__(
     59             nodes=nodes,
     60             index_struct=index_struct,

TypeError: __init__() got an unexpected keyword argument 'llm_predictor'

Could you help me with this?

这个批量出图的方式我反复找了教程，都没有搜索到花括号。
语法{,}可以让MJ组合提示词。例如
3D, GC, girl sitting on chair, {red hair, blue hair},{green, white} socks会组合出4组提示词。

分别是
3D, GC, girl sitting on chair, red hair,green socks
3D, GC, girl sitting on chair, red hair,white socks
3D, GC, girl sitting on chair, blue hair,green socks
3D, GC, girl sitting on chair, blue hair,white socks
这个方法十分强大，可以批量出图，然后去干别的。

感谢作者提供的教程，我是通过你这个教程入门MJ的，之前研究过stable diffusion。

网站是不是删除了部分内容啊

陈述(1):售价减去房地产经纪人的佣金为84,600美元。

给定:佣金是售价的6%。

公式:佣金 = 售价 × 6% = 售价 × 0.06

根据陈述(1):售价 - 佣金 = 84,600

将公式代入:售价 - 售价 × 0.06 = 84,600

解得:售价 = 84,600 / (1 - 0.06) = 90,000美元

陈述(2):购买价是36,000美元,是售价的 250%。

给定:购买价 = 36,000;购买价是售价的250%。

公式:购买价 = 售价 × 250% = 售价 × 2.5

将给定条件代入公式:36,000 = 售价 × 2.5

解得:售价 = 36,000 / 2.5 = 14,400美元

综上,仅陈述(1)足以回答问题,售价是90,000美元。

但仅陈述(2)不能回答问题,因为根据陈述(2)得出的售价是14,400美元,与givne的佣金率6%不符。

因此,选择A:仅陈述(1)足以回答问题,但仅陈述(2)不能回答问题。

所以答案是:
A

设该房子的售价为x美元。

根据题意可得：
(1) x - 0.06x = 84600
(2) x = 1/2.5 * 36000

化简后可得：
(1) 0.94x = 84600
x ≈ 90000
(2) x = 14400

因此，只有陈述（1）足以回答问题，选项（A）为正确答案。

1. https://learningprompt.wiki/docs/tutorial-basics/%E5%9F%BA%E6%9C%AC%E4%BD%BF%E7%94%A8%E5%9C%BA%E6%99%AF%20&%20%E4%BD%BF%E7%94%A8%E6%8A%80%E5%B7%A7/%E5%9C%BA%E6%99%AF8%EF%BC%9A%E5%8C%96%E7%B9%81%E4%B8%BA%E7%AE%80%E2%80%94%E2%80%94%E4%BF%A1%E6%81%AF%E6%8F%90%E5%8F%96

技巧6：通过示例来阐述需要输出的格式 下面

其实在场景7_种_，你也可以用到这个技巧

2. https://learningprompt.wiki/docs/tutorial-extras/OpenAI%20Playground%20%E4%BD%BF%E7%94%A8%E6%96%B9%E6%B3%95

你会在界面的右侧看到以下几个参数：

Mode： 最近_跟_新了第四种 Chat 模式

The 'thinkingjimmy/Learning-Prompt' repository doesn't contain the 'docs/midjourney-learning-path' path in 'main'.
The 'thinkingjimmy/Learning-Prompt' repository doesn't contain the 'docs/chatgpt-learning-path' path in 'main'.
希望能更新一下路径

我最近也在研究如何使用 OpenAI 的接口推进 我的 AI 绘画工作，我在我的项目里面使用了下面的一些提问方式，并且直接使用字符串模板的方式载入到了我的 AI 绘图网站中。经过我几天的研究，发现 AI 描述物体的能力强于 Tags 组合，但是AI一般不会出头部的风格组合词。希望能够帮到你。虽然是中文提问，也是可以回复很高准确度的。然后 我用的是 StableDiffusion，并不是 MJ，但我猜测是可以通用的。

https://gitee.com/dongzhongzhidong/magic-tag/blob/master/Chat_Prompt.md

thinkingjimmy/chatgpt-prompts-zh#1 (comment)
最好的prompt应该是Let's work this out in a step by step way tobe sure we have the right answer.
https://sites.google.com/view/automatic-prompt-engineer