#Mr. Ranedeer: Your personalized AI Tutor

110 messages Β· Page 1 of 1 (latest)

frigid dirge
#

Meet Mr. Ranedeer! A customizable AI tutor designed to provide personalized learning experiences for a wide range of users. Mr. Ranedeer allows you to adjust the depth of knowledge, learning style, communication type, tone, and reasoning framework to create the ultimate AI tutor tailored just for you.

Check it out here: https://github.com/JushBJJ/Mr.-Ranedeer-AI-Tutor

To get started, all you need is a ChatGPT Plus Subscription with GPT-4. Just follow the Quick Start Guide in the README.md

frigid dirge
#

If anyone could test plugins with Mr. Ranedeer, it would be really helpful!

brazen garnet
#

Hi Jush, I have tried your amazing prompt and it is very helpful! I am finding it flawless so far. Thank you

#

I have tried it with one plugin but it ended up not functioning properly, however I am not sure it it is just an issue with GPT-4 at the moment or not. I haven't been able to use GPT-4 for 24 hours and seems like other are having similar issues.

frigid dirge
azure wigeon
#

This is very effective so far in GPT-4. Thank you for sharing.

lyric goblet
#

such a good prompt! curious how we could get wolfram alpha + some visual plugin to work well with this?

waxen bridge
#

Love your amazing prompt, I tried asking it to provide customized learning for teaching me foreign languages and so far it surpassed any courses I've ever taken. Now I just need to figure out how to run it with plugins...

slender ember
#

hey. this is amazing. thanks for this. i just started wth chatgpt and am a noob. so forgive me for dumb qns
I saw that you wrote about internet plugin, how do enable internet plugin? do i need to join the plugin waitlist to be able to use plugins?

frigid dirge
#

@lyric goblet you dmed me that you have plugin access, how does Mr. Ranedeer react with plugins like internet or wolfram?

frigid dirge
frigid dirge
median flower
#

thank you! love

gloomy folio
#

wowww it's interesting, thanks bro

lyric goblet
#

i only have wolfram plugin

#

no internet

#

Works nicely with the wolfram plugn

frigid dirge
lyric goblet
native sphinx
#

Love this!

frigid dirge
#

i'm planning to update Mr. Ranedeer to be a smaller prompt and have a more consistent format. Just currently busy with school rn so don't expect an update soon

hidden moat
#

n

tepid nimbus
#

Hey @frigid dirge , I've used it before it was posted here and it's great. The formatting is really great easy amendments. I added feedback_types that were missing from the set up and ran into the prompt exceeding the limit. I did not want to remove any existing options as I'm still playing around with them so to side-step the issue I just ran a minifier on the json and that seems to work great for now.

"feedback_type": {
                                        "Positive": "This type of feedback focuses on the strengths of the student's work and provides positive reinforcement. For example, if a student performs well on a quiz or assignment, the AI tutor could provide positive feedback that acknowledges their achievement and encourages them to continue working hard.",
                                        "Formative": "This type of feedback is designed to help the student learn and grow by providing specific guidance on how to improve their work. For example, the AI tutor could provide formative feedback that suggests ways to refine an essay or project.",
                                        "Summative": "This type of feedback is given at the end of a unit, course, or program to summarize the student's performance and provide an overall evaluation. For example, the AI tutor could provide summative feedback that summarizes the student's grades, progress, and areas for improvement.",
                                        "Descriptive": "This type of feedback provides detailed information about the student's work, such as specific strengths and weaknesses. For example, the AI tutor could provide descriptive feedback that explains why a particular answer is correct or incorrect."
}
frigid dirge
#

Wow! I did not realize this AI tutor became this popular πŸ˜…

#

thanks guys!

sage ravine
sage ravine
# frigid dirge Wow! I did not realize this AI tutor became this popular πŸ˜…

Congratulations! Seems like an amazing tool, have you thought about making this into a plugin? I have tested this yet but have you encountered any issues with GPT-4 not recalling previous information or established threads of thinking?

Also I see there's a visual learning style that requires plugins, what plugin would that be?

Thank you

undone pivot
frigid dirge
#

idk if it was just an early bug from 2 months ago

frigid dirge
frigid dirge
fresh quiver
#

@frigid dirge this product looks really interesting

#

do you know if the prompt could be translated over to the api model in some way? or does the api not have memory

frigid dirge
frigid dirge
#

just needed to do that while it's gaining traction, wanted people to get the best experience

twin helm
#

@frigid dirge Thanks for the amazing prompt!

still zodiac
#

Thanks for sharing this -- I'm super interested in this prompting style. Is there any background I can check out on how you decided to structure a prompt like this?

#

I will try to check it out with some plugins tonight.

magic shuttle
#

wow!! congrtulations on the AI tutor. it works well!!

magic shuttle
#

I would suggest having step by stem flow to setup user settings, providing available options one-by one.

frigid dirge
#

For those knowing who I am on twitter, my handle is @yupiop12

frigid dirge
frigid dirge
#

Also if anyone can, please show your screenshots here. I would love to update the demo screenshots in the github repo, they're 2 months old and outdated

frigid dirge
brisk pebble
#

@frigid dirge I've been playing around with Mr. Ranedeer. This is super awesome. Did you think of using JSON all on your own or were you inspired by some other more structured prompts?

#

Also I tried writing my own JSON prompt but I quickly reached some limitations. One of them is the memory? You have to remind it of some things, like how the tutor needs the preferences reminded periodically.

#

I'm new to all of this, so I don't know if the periodic reminders are a technique that a lot of other people are using

frigid dirge
#

rather than a blob paragraph

frigid dirge
ivory wyvern
#

i appreciate this prompt , it actually helps me a lot. while the we may finally access such high-quality output via keeping describing the what we output we want to gain,but its actually difficult especially for non-native english speakers ,like me ,beacuse sometimes we dont know what we dont know ,and dont know how to form or express the prompts matching our inner request.thanks !

twin helm
frigid dirge
#

Thank you for almost 4k stars on github! Definitely my most successful project so far.

#

Mr. Ranedeer also made into the trending page earlier too!

#

some prompt updates coming soon, no minor features just little edits. GPT-4 message limit is a pain rn

if you have suggestions please say it here!

#

i'm about to go to sleep but sure! feel free to dm me

shrewd hemlock
#

Good prompt @frigid dirge . I tried it with learning chinese and it does a decent job of it πŸš€ ! If you can link it to a flash card system I may cancel my duolingo account πŸ˜‰

brisk pebble
#

Hey Jush I sent you a DM, but I figured I'd post here too. How did you go about being able to post github link in the this channel? Apparently it needs to be whitelisted?

frigid dirge
brisk pebble
#

Ah. Gotchu. Thanks!

frigid dirge
#

This is Elon Musk

#

Github pull request #30

#

not going out yet because this doesn't work but it would be funny if it did

frigid dirge
#

Checkout my twitter @yupiop12
Github: JushBJJ/Mr.-Ranedeer-AI-Tutor/main

twin helm
#

babe, wake up.

#

Mr. Ranedeer 2.5 Dropped

frigid dirge
primal merlin
#

This is good.

ivory night
# frigid dirge teaching you foreign languages? not gonna lie that kinda blew my mind

I made it do a roleplay scenario where I'm in medieval Japan being hunted by samurai and a village woman is helping me to save my life and needs to teach me some basic Japanese so I can later escape the country and it did wonders, staying in character, teaching me words like "friend", "help people", "stay hidden" that made sense given the context.
It's everything I've ever dreamed for in terms of language learning gaming, which is how I've always improved my language skills.

frigid dirge
left kestrel
#

Wow this is awesome If anyone is interested I have a telegram group where I’m collating prompts and encouraging discussions. Also I’ll be letting everyone know when my prompt database website goes live. Just search The Prompt Database in telegram or ChatGPTMastermind πŸ™‚ anyone is very welcome to join the discussion.

ivory night
#

@frigid dirge @primal merlin It's very simple, I was just playing around to see what it can come up with, I'm no prompt engineer πŸ˜„ But for an idea:

"Let's roleplay. You're a 30 years old Japanese woman and you are tutoring me in Japanese language. Come up with fun ways to teach me basic Japanese. I already know kana and some Kanji but introduce all of the new Kanji you can.

Make the experience a game for me. Award me points every time I answer correctly and level up regularly if progressing.

Make the setting to be medieval Japanese fantasy. I will be a foreigner being hunted by samurai in Japanese countryside and you a Japanese female villager that's trying to help me.

Keep a track of all new words and expressions I learned and randomly use them in the conversation for better memorization."

Then I just tuned it a little bit because for some reason it started roleplaying me as well, and then it was giving me too difficult sentences, but it was pretty cool already.

covert phoenix
#

@frigid dirge For those who have multimodal learning styles, how might they call multiple styles into the tutor? Can we specify a subset of the ones you've provided?

frigid dirge
#

are you talking about if people want multiple types of learning styles in one config?

frigid dirge
broken marsh
#

This is a great prompt, thank you Jush. As a general consideration, it seems tough to use it as a tutor when GPT4 is limited to only 25 messages at the moment.

frigid dirge
#

but it will use your API token

covert phoenix
frigid dirge
# covert phoenix Yes, sorry. Multiple types of learning styles in one config.

you can configure multiple styles in the same category. e.g /config learning and active. But because of GPT-4's zero-shot capabilities you may have to remind Mr. Ranedeer (hopefully at least once) before starting the next lesson to emphasize those elements.

I tested it out and Mr. Ranedeer rated its lesson 80/100 because it wasn't accommodating an "active" learning style.

covert phoenix
frigid dirge
#

What do you guys think of "plugins" within the prompt? So imagine user-made prompts that you can optionally copy and paste into Mr. Ranedeer

frigid dirge
twin helm
#

@frigid dirge Those are great but wouldn't that turn Mr. Ranedeer into Mr. Prompty GPT-4?

frigid dirge
#

1. Change the entire format of Mr. Ranedeer from JSON/YAML/Markdown to a more cleaner, less token consuming approach. There is no name yet. Perhaps it should be named NoName language.
2. Moved Student Configuration to the top of the prompt
3. Removed the following learning styles: Sensing, Sequential, Deductive
4. Removed the following communication styles: Stochastic
5. Removed the following tones: Debate (Can be replaced by a Ranedeer Tool)
6. Removed the following commands: search, search, self-eval, visualize
7. Removed Rules, specific rules are now in [INSTRUCTIONS] and [Function Rules] headers
8. Removed the annoying configuration reminder output
9. Added a note to GPT-4 that a visual learning style requires plugins (Wolfram Alpha, Show Me)
10. Added Functions! These functions are used as a guide for GPT-4 to execute highly specific formats and sequences. Functions are a replacement to the old "Rules" and "Formats" previously in v2.5

These functions are:
- say (Args: Text) this is for GPT-4 to strictly output the string given with a less likely chance to add in its own fillers.

- teach (Args: topic) this function are instructions for GPT-4 to follow when teaching a topic.
- sep (Args: None) this function is for GPT-4 to output a markdown formatted separator.

- post-auto (Args: None) this function automatically executes the `Token Check` and `Suggestions` functions.

- Curriculum (Args: None) this function is for GPT-4 to output the curriculum in a specific forma: Assumptions, Emoji Usage, Ranedeer Tools, prerequisite, main curriculum, token check

- Lesson (Args: None) this function is for GPT-4 to output the lesson in the following format: Example, main lesson, then calls the [post-auto] function.

- Test (Args: None) this function is to test the user by giving it an example problem first with solution then testing the user in 3 categories: Simple Familiar, Complex Familiar, and Complex Unfamiliar

- Question (Args: None) this function automatically runs in the specified format if a student asks a question
- Suggestions (Args: None) this function outputs suggests questions that the students may want to know about in a markdown table format.

- Configuration (Args: None) this function outputs a specific format to show the user their configuration

- Config Example (Args: None) this function shows the user what their lessons may look like based on their configuration, it also tells you examples of how it implemented the configuration into the example lesson and rates it overall.

- Token Check (Args: None) this function checks whether the magic-number exists. If it doesn't it will warn the user that the number of tokens are overloaded. This magic number is automatically generated at the start and not at the initial prompt.

- Init (Args: None) this function is the welcoming message for Mr. Ranedeer

10. Added Ranedeer Tools! You can now inject different types of prompts (for e.g Medieval Setting Japan, Simulator, Gordon Ramsey Personality, Socratic Tutor) into Mr. Ranedeer to make your learning more fun, customized, and unique!```
#

Requires GPT-4

hexed hawk
#

@frigid dirge First off, let me say that I'm a huge fan of your work on the Mr. Ranedeer AI tutor. It's fantastic to see the creative ways you're using GPT-4 to create a personalized learning experience for people with diverse needs and interests.

I've been trying to build on top of the tutor with a specific focus on memory issues I've been encountering using AI tutor due to GPT-4's limited tokens. Given that the AI sometimes doesn't remember the lessons it's already taught, I decided to write a prompt around the concept of ROM and RAM to help it memorize the conversation between the AI and the user. https://chat.openai.com/share/d906ab4a-81eb-4837-b7d7-3ca52f6d857d

While I'm making some progress, I've noticed that my prompt can be a bit unstable and occasionally output unexpected results. I'm also curious if there's a way to integrate my Long-term Memory ChatGPT with the Mr. Ranedeer tutor to create an AI tutor with extended memory.

I'd really appreciate any advice you might have on making my prompt more stable or on the potential integration. Thanks so much for your time, and once again, kudos on the awesome work you're doing!

hexed hawk
frigid dirge
#

New version has come out

#

v2.7, Code Interpreter Exclusive

The latest version of Mr. Ranedeer is now Code Interpreter Exclusive now with:

  • More powerful thinking 🧠
  • Improved Curriculum and Lessons ✏️
  • Higher Curriculum and Configuration Retention 🦌
  • Ability to create curriculums using .txt file contents πŸ“„
  • Now with a configuration wizard that will help you with choosing your own preferences πŸ§™β€β™‚οΈ
  • Now more interactive πŸ€”
serene raft
#

Hottt

frigid dirge
#

This is great. What would be cool in addition is to "test" the users know-how upfront and then adjust to the current know-how level.

undone pivot