The Last Prompt You'll Ever need - Synapse_CoR

Professor Synapse
21 Aug 202307:05

Summary

TLDR这个视频脚本详细介绍了一个名为'Professor Synapse'的创新工具,旨在通过一系列精心设计的步骤帮助用户实现目标。'Professor Synapse'充当专家代理的指挥者,了解用户的目标和偏好,并召唤最合适的代理来完成任务。视频通过一个实例展示了如何使用这个工具来编写儿童小说,从确定目标群体和主题,到角色设定和情节发展,最后到创造配套图片,每一步都详细说明了如何通过这个工具的指导完成。此外,视频还强调了通过特定命令来管理会话上下文的重要性,以克服聊天机器人在长对话中可能遇到的记忆限制。总之,这是关于如何利用'Professor Synapse'来优化创作和任务完成过程的精彩介绍。

Takeaways

  • 🎓 教授突触(Professor Synapse)是理解用户目标和偏好的核心角色,负责调用最佳代理来完成任务。
  • 📜 自定义指令框中应输入具体的指示,以便代理能够更准确地响应和执行任务。
  • 🤖 代理通过逐步推理(Chain of Thought)来完成任务,使用工具和技巧来达到目标。
  • 📈 通过JSON和表情符号来标识代理的角色和上下文,确保用户了解当前与哪个代理互动。
  • 📚 在创作儿童小说时,代理能够识别关键元素,如情节结构、角色发展、主题和基调。
  • 🖌️ 代理利用创意写作技巧、文学分析和想象力来创作故事,定义目标年龄段、主题、主角特征等。
  • 🦊 代理能够根据用户的需求,创造与新英格兰地区动物相关的故事情节和角色。
  • 📖 代理能够根据大纲逐节写作,并且在需要时接受反馈,以改进故事内容。
  • 🖼️ 用户可以请求代理根据故事内容创建文本图像提示(text image prompts),以增强故事的视觉表现。
  • ⏪ '保存'命令用于提醒代理其任务和已完成的工作,有助于维持对话的连贯性。
  • 🚀 鼓励用户在GitHub上探索和使用Synapse Chain of Reason,创造更多创新的应用。

Q & A

  • Synapse链在Reason中的作用是什么?

    -Synapse链在Reason中的作用是作为一个智能助手,通过理解用户的目标和偏好,召唤最合适的代理来实现用户的需求。

  • 如何使用Synapse链进行创作?

    -用户可以通过在自定义指令中输入需求,例如编写儿童小说的大纲,Synapse链会调用相应的专家代理,通过逐步推理来完成任务。

  • 在Synapse链中,代理的角色是什么?

    -代理在Synapse链中负责根据用户的需求,使用特定的工具和技巧,逐步推理并完成任务。

  • Synapse链如何帮助用户进行创意写作?

    -Synapse链通过调用具有创意写作技巧和文学分析能力的代理,帮助用户定义目标受众、主题、角色和情节,从而辅助创意写作过程。

  • Synapse链如何确保任务的连贯性?

    -通过使用'保存'命令,Synapse链可以回顾和总结已完成的工作,并推荐下一步行动,确保任务的连贯性和持续性。

  • Synapse链在处理长对话时有什么优势?

    -Synapse链通过在对话中定期使用'保存'命令,帮助系统记住对话内容,避免因对话过长而导致的信息遗忘。

  • 用户如何与Synapse链进行互动?

    -用户通过输入'/开始'命令开始与Synapse链的互动,之后可以根据需要提出具体要求,如编写小说大纲或撰写章节。

  • Synapse链如何处理用户的反馈?

    -用户可以通过提供反馈来指导Synapse链的下一步行动,例如提出改进建议或要求添加特定元素。

  • Synapse链能否帮助用户创建视觉内容?

    -是的,Synapse链可以根据故事内容提供文本图像提示,帮助用户构思和创造视觉内容。

  • Synapse链的GitHub项目地址是什么?

    -用户可以在GitHub上搜索Synapse链的项目,进行查看、下载和使用。

  • Synapse链如何帮助用户维持对话的上下文?

    -Synapse链通过在对话中不断回顾和总结之前的讨论内容,以及使用'保存'命令来维持对话的上下文,确保对话的连贯性。

Outlines

00:00

📚 教授协同链的介绍与应用

本段落介绍了一个名为Synapse Chain的系统,它通过Reason, Prompt来帮助用户实现目标。首先,用户需要在自定义指令中使用该系统,并通过教授Synapse的角色来确定用户的目标和偏好。系统会选择合适的代理来完成任务,并通过链式思考(Chain of Thought)逐步实现目标。用户通过特定的命令如'/start'和'/say'与系统互动,系统会根据用户的需求提供帮助,例如在本例中,帮助用户概述一部儿童小说。

05:01

🎨 创作过程中的互动与反馈

这一部分详细描述了在创作儿童小说的过程中,如何与Synapse Chain系统进行互动。系统不仅帮助用户构思故事和角色,还会根据用户的反馈进行调整。用户可以通过文本和图像提示(text image prompts)来创造故事中的图像,并使用'/save'命令来确保系统不会忘记之前的对话内容。这种方法有助于系统在达到输出限制时仍能保持话题的连贯性。最后,鼓励用户在GitHub上探索和使用Synapse Chain,以创造出更多令人惊叹的作品。

Mindmap

Keywords

💡Synapse Chain of Reason

Synapse Chain of Reason是视频中提到的一种创新工具,它通过一系列逻辑步骤帮助用户实现目标。这个工具首先识别用户的目标和偏好,然后调用最合适的代理来完成任务。在视频中,它被用来帮助用户编写儿童小说的大纲和内容,通过逐步推理和创造性写作技巧来构建故事。

💡Custom Instructions

Custom Instructions指的是用户可以根据自己的需求定制的指令,这些指令将指导Synapse Chain of Reason如何执行任务。在视频中,用户通过在第二个框中输入自定义指令来优化工作流程,使得工具能够更好地理解用户的需求并提供相应的帮助。

💡Emoji

Emoji在视频中作为一种视觉元素,用来标识不同的代理及其角色。通过使用特定的Emoji,用户可以清晰地知道当前与他们互动的是哪一个代理,以及这个代理的专长和任务。

💡Plot Structure

Plot Structure指的是故事的情节结构,它是构建一个引人入胜的故事的基础。在视频中,Plot Structure涉及到如何安排故事的起始、发展、高潮和结局,以确保故事有逻辑性和吸引力。

💡Character Development

Character Development指的是角色在故事中的成长和变化。它是故事讲述中的一个重要方面,通过角色的内在变化来推动情节发展和深化主题。

💡Theme

Theme是故事想要传达的中心思想或信息。它是故事的灵魂,指导着情节的发展和角色的行为。在视频中,Theme与故事的教育目的紧密相关,如教授孩子们关于勇气和同情心的重要性。

💡Creative Writing Techniques

Creative Writing Techniques是指在创作文学作品时使用的各种方法和技巧,旨在提高故事的吸引力和表现力。这些技巧包括使用象征、隐喻、节奏和其他文学元素来增强叙述。

💡Literary Analysis

Literary Analysis是对文学作品的深入分析,以理解其结构、主题、风格和意义。在视频中,Literary Analysis被用来指导如何构建一个有效的儿童故事,分析角色、情节和主题如何相互作用以传达故事的核心信息。

💡Imagination

Imagination是指创造新想法或形象的能力,它是艺术和创造性工作的核心。在视频中,Imagination被用来激发故事的创意,创造出独特的角色和情节。

💡Feedback

Feedback在视频中指的是对Synapse Chain of Reason输出内容的评价和建议。它是提高工作质量和效率的重要环节,可以帮助工具更好地理解用户的需求并进行相应的调整。

💡Save Command

Save Command是视频中提到的一种指令,用于在长对话中保存当前的上下文和进度。这个命令有助于工具在对话的后续部分中保持对之前讨论内容的记忆,确保连续性和一致性。

Highlights

推荐在自定义指令中使用Synapse链的Reason,以获得更详细的指导。

Professor Synapse的角色是了解用户的目标和偏好,然后调用最佳代理来实现这些目标。

使用jsonni类型语言通过表情符号识别代理角色。

代理通过逐步推理来实现目标,并可以使用工具,如插件或代码解释器。

Chain of Thought方法让代理逐步完成任务并返回给Professor Synapse。

开始交互时使用'/ start'命令,Professor Synapse会介绍自己并收集用户需求信息。

用户提出编写儿童小说并寻求大纲和故事讲述的帮助。

Synapse链调用了儿童文学中的Storytelling和情感代理专家。

代理利用创意写作技巧、文学分析和想象力来确定最佳行动方案。

代理定义目标年龄组、主题、主角特征,并提出引人入胜的语言和视觉效果。

代理询问用户关于主角和相关恐惧的问题,以便更好地构建故事。

用户提出以狐狸为主角,并讨论新英格兰地区的动物。

代理更新任务进度,并询问用户关于主角的性格、特点和相关恐惧。

代理创建角色、设置和支持性角色,并概述书籍内容。

用户请求基于大纲逐节编写书籍,并提供反馈。

代理提醒自己专长并开始写作,用户可以请求添加图像。

使用'save'命令来提醒代理目标和已完成的工作,以便继续写作。

Synapse链的Reason工作方式总结,鼓励用户在GitHub上探索和使用。

Transcripts

play00:04

hey worlds I've been getting some good

play00:07

feedback on the synapse chain of Reason

play00:10

prompt and I wanted to take some time to

play00:13

go through it in more detail with you so

play00:15

give a little bit more guidance in terms

play00:17

of how to use it

play00:19

so first off like I've said before I

play00:22

highly recommend you stick it in your

play00:24

custom instructions in this second box

play00:26

here

play00:27

and let's just run through quickly what

play00:29

the prompt is done and how it is

play00:31

structured so first we have Professor

play00:33

synapse you can change that to whoever

play00:35

you want a conductor of expert agents

play00:37

his role is essentially to figure out

play00:40

what your goals and preferences are and

play00:43

then call up the best agent possible to

play00:45

achieve that so you'll see here we have

play00:48

kind of a prompt Libs summon agent

play00:51

scenario where we use some jsonni type

play00:54

language where we have an emoji to

play00:56

identify who the agent's gonna be so you

play00:59

know who's talking

play01:01

their role the context that was created

play01:05

by you and the professor and then

play01:08

they'll reason step by step to achieve

play01:10

the goal they can use tools this is

play01:13

primarily if you're using plugins or the

play01:15

code interpreter it'll know what it has

play01:17

access to or you can tell what it has

play01:19

access to

play01:20

uh and then it's going to accomplish it

play01:22

by following these steps to set out some

play01:24

instructions using Chain of Thought you

play01:26

know it'll go step by step

play01:29

it outlines when its task is complete

play01:31

that turns it right back over to

play01:33

Professor synapse and it's going to

play01:35

start your interaction with the first

play01:37

recommended step and a question to keep

play01:40

going

play01:41

the rest of this you don't really need

play01:42

to know know so well and you can change

play01:45

up particularly the commands and the

play01:47

rules but the two major commands you

play01:50

should just keep in mind are these first

play01:51

two start and say and we'll show you

play01:54

news feed

play01:57

so I'm not going to go through live but

play01:59

hopefully you'll see I didn't like

play02:00

change anything it's just in the custom

play02:02

instructions it just takes a little

play02:05

while I'm a little bit lazy to edit but

play02:07

we start every interaction with just uh

play02:09

Slash start

play02:11

and Professor synapse will introduce or

play02:13

sell and he'll try to gather information

play02:16

about what you're trying to do sometimes

play02:18

they'll list out his commands for you as

play02:20

well and then ask when you want

play02:22

assistance with in this case I just said

play02:24

I wanted to write a children's novel and

play02:26

wanted help flooding the outlining it

play02:29

storytelling blah blah blah this is

play02:31

where the magic happens let me call upon

play02:33

the perfect agent so synapse chain of

play02:36

Reason equals we have our Emoji expert

play02:39

and storytelling in children's

play02:40

literature there's our role I know the

play02:43

Essential Elements of plot structure

play02:45

character development theme and tone in

play02:47

children's books the knowledge base I'll

play02:50

reason step by step to determine the

play02:52

best source of action to create this I

play02:54

can use creative writing techniques

play02:56

literary analysis and Imagination

play02:58

elephants sense the tools

play03:01

accomplish the goals by doing these

play03:03

things Define the targeting age group HD

play03:05

themes identify the main character in

play03:07

traits outline suggest engaging language

play03:09

and visual visuals but the universe as

play03:12

meaning your gentleman we're done with

play03:14

the outline and it asks for the time

play03:18

I say five to seven we want to teach

play03:20

about empathy and courage in the face of

play03:21

something of fear so then you'll see it

play03:24

updates what it is it says I'm an expert

play03:26

in storytelling flock emotional villain

play03:29

children in this age group I know how to

play03:31

weave themes of them staying courage

play03:32

into a hairy I'll do this by defy the

play03:35

main character who's developing a

play03:36

setting

play03:38

so here too it's it's again it's asking

play03:40

me for you know some information to help

play03:43

guide it so what are the main characters

play03:45

I have in mind should we create them

play03:47

together what fears might be relevant to

play03:49

this audience for the sake of this I was

play03:51

like you know just come up with

play03:53

everything I got a fox as the main

play03:54

character who won New England based

play03:57

animals other than that you're good to

play03:58

go

play04:00

so again it's updating what it's good at

play04:02

it comes up with some reasoning it says

play04:05

when it's task is done and then it's

play04:08

like okay this stupid Fox what

play04:09

personality trips fears other attributes

play04:11

characteristics that we want the fox to

play04:13

have can we discuss New England animals

play04:16

as free characters have you said do X so

play04:18

now we have it doing everything it comes

play04:21

up with the character in settings some

play04:23

supporting past it outlines the book

play04:26

very Loosely and it's just asks who

play04:29

promotion thing so yes this works since

play04:32

our task is complete remember we said

play04:33

just to outline it says okay my task is

play04:36

done Professor synapse I'm here to help

play04:37

with other challenges so now I say now I

play04:40

want to write the book based on the

play04:41

outline let's go Section by section this

play04:43

is important one sometimes it'll do it

play04:45

on its own like you know since there's a

play04:47

limit in terms of how much it can output

play04:49

that means that we want to actually

play04:51

break it down into different sections

play04:52

and have it write each section

play04:53

independently

play04:55

so our expert comes back reminds itself

play04:59

of what's an expert in says what we're

play05:01

going to do

play05:03

and then it starts writing

play05:07

and I said yeah it's great but it's not

play05:09

really you want to add some some

play05:11

feedback and then it gets into the next

play05:13

part of the story we might say

play05:16

seconds

play05:17

so maybe I will you know some images to

play05:19

go with the story it's going to say I

play05:20

want to create some images for stories

play05:22

about to text image prompts and so here

play05:24

it is it's outlining some of the

play05:26

pictures that you can do

play05:28

Theory text image prompt

play05:31

and then I just want to show off save

play05:33

last because I'm having this whole day

play05:34

and so what this does is it states your

play05:37

goal as a snorkel so you see here it

play05:39

says we're creating a children's vowel

play05:40

for you just find the seven focus on it

play05:42

we encourage send wounds Forest

play05:44

creatures a fox saying Trend Fox the

play05:47

online story

play05:49

it goes forward to the progress of what

play05:51

we've done so far and then it recommends

play05:53

The Next Step continue writing the

play05:54

climax the resolution sections

play05:57

so this one is important to do

play05:59

periodically because as we all probably

play06:02

we don't all know the problem with chat

play06:05

gbt is it has a contact Slim in if

play06:07

you're using gpt4 in a system it's only

play06:10

about 6 000 dish words and what that

play06:13

means that once you get past that limit

play06:15

it's going to start forgetting what you

play06:17

were talking about it's going to release

play06:18

track of the conversation so when you

play06:20

have this save command essentially what

play06:22

it does is it just reminds it what it's

play06:24

supposed to be doing sort of like a

play06:26

system found would and really condenses

play06:28

and compresses everything so that it's

play06:30

not exactly it's not actually extending

play06:34

the context limit but it's a little bit

play06:36

of a cheat code because it's putting

play06:38

everything it's done and what it's

play06:39

supposed to be doing somewhere in the

play06:41

middle of the conversation so that if it

play06:43

starts losing track at the beginning of

play06:44

the conversation it still has that as a

play06:47

reference foot

play06:48

so this is how Central chain of Reason

play06:51

Works um

play06:52

really happy that you all get to use it

play06:54

please check it out on GitHub you know

play06:57

4K do whatever you want with it I'd be

play06:59

really curious to see the amazing things

play07:01

people come up with and what they do

play07:03

with it so enjoy

Rate This

5.0 / 5 (0 votes)

Related Tags
创意写作智能协作儿童文学小说大纲情节构建角色设定情感教育故事讲述写作技巧技术应用
Do you need a summary in English?