(FAQ) Frequently Asked Questions - VBPXKSMI/Wiki-Test GitHub Wiki
AI Dungeon Frequently Asked Questions (Actively Updated)
https://pastebin.com/PP3FQm4R
If you prefer pastebin to github the FAQ is also located here:A.What is AI Dungeon?
A text adventure game where you can do literally whatever you want and the AI will respond and play Dungeon Master for you.
B.Can I do lewd stuff with it?
Yes, including that specific fetish you have that no one else has. It may take a bit more effort to coax the AI doing it but it's worth it.
C.Can I go on fantasy adventures or is it only good for smut?
What part of literally anything don't you understand? You can fly into space and become a dragon and fight the Starship Enterprise. Get creative with it and make your own batshit insane stories!
D.How do I start playing?
Go here and pick a Colab of your choosing.
If you can't choose, just play this one.
If you want to play the official version that's inferior but just works out of the box go here.
E.Can I play it on my phone?
Yep, passing by strangers can see just how much you wanted to drag a slime girl into your basement and do very naughty things to her. Just pick one of two links above. There's a section in the wiki for making the Colab verison look nice on mobile: https://github.com/VBPXKSMI/Open-CYOAI-Project/wiki/How-to-make-the-Colab-version-look-nice-on-mobile
F.How do I play it locally on my PC?
Check out this page for info on how to download it. You can do it manually or use the installers. We have two different kinds available currently.
G.What is the difference between the official version and 4chan's versions?
Here's a list showing some differences between the official version and 4chan modded versions:
- Official version has no custom AI models.
- Official version runs worse locally as it doesn't use PyTorch, so local running is less possible if possible at all (difference is huge, 6-8GB of VRAM for Clover Edition to run vs 12-16GB of VRAM needed for the official version).
- Official version has no temperature or vocabulary control.
- Official version has no repetition penalty system.
- Official version saves all your stories to a public folder that anyone can access, or saves directly in the servers of the original author.
- Official version has no prompt retrying or altering.
- Official version has no /remember command to add things for the AI to remember after the 8-12 response limit.
- Official version doesn't have the hundred+ custom prompts that we have compiled.
- Official version doesn't have an ability for you to save your own custom prompts to use again at start.
- Official version can't alter AI responses to keep it on track and provide more accurate future responses.
- Official version is trying to remove sex and any other related lewd, or politically incorrect things.
Needless to say Clover and Thadunge2's versions are superior to the original official version when it comes to quality of life and running smoothly. The web-based official version (where is currently hosted) suffers from all the previously stated points, except that it runs right out off the box.
H.What is the difference between Clover and Thadunge?
Currently they pretty much have the same commands, Clover has a little more optional functionality, like the suggested actions and d20 mode(which makes it so you attempt every action and can possibly fail them instead of just succeeding outright). Apparently the models are different enough between Clover and Thadunge so the AI gives different responses based on which one you're using. If you find you're dissatisfied with one version then try out the other one for a change of pace. Locally Clover is superior for only needing 8GB of VRAM to run locally vs 16GB of VRAM for Thadunge. It's also possible to only need 4GB of VRAM to run Clover locally depending on which custom model you use.
I.The Colab disconnected or I got some weird bug or error what do I do!?
The Colab disconnecting is on Google not us, there's nothing we can do about it. If you want no disconnects then play locally. For any other errors or bugs check these links:
For Colab Troubleshooting: https://github.com/VBPXKSMI/Open-CYOAI-Project/wiki/Colab-Troubleshooting
For Local Troubleshooting(check the quick troubleshooting tab): https://github.com/VBPXKSMI/Open-CYOAI-Project/wiki/How-To-Install-AI-Dungeon-2-Locally-On-Windows-(Manual-and-Automatic-installations)
If you don't find your error on the pages or the answer you're looking for post a screenshot of the error in the thread and ask for help.
J.The AI is being stupid and won't cooperate with me in playing out my smut fantasy what do I do?
The AI tends to base its style and pacing on the most recent couple of prompts. The easiest way to get your AI to do better pacing and descriptions and stuff is to manually /alter the first few responses you get to be more fleshed out, have slower pacing, etc. Once the AI has enough to go off of it'll usually stick to that level for future responses and may only need a little prodding and editing afterwards.
ex. If you start out stripping a character and get
You take off Anonette's blouse and touch her boobs.
Manually rewrite it to
You reach your hands out and begin to slowly unbutton her blouse. She looks at you with her big doe-eyes as you fiddle with each button. You snap the last button and watch as her supple breasts spill out of her open blouse. You lick your lips as you take the sight in. You timidly slide your hands up Anonette's slim waste and cup her soft breasts with your hands. You hear a tiny gasp escape her mouth as you flick your nipples with her hands and start to squeeze her tits.
A couple of those and the AI will take a hint. Use lots of different adjectives and nouns so it doesn't get caught in a rut (the thesaurus is your friend)
K.Do you have any advice on how to make the AI not stupid? It just won't listen to me!
Check out this page. It has a lot of helpful information in general on this: https://github.com/VBPXKSMI/Open-CYOAI-Project/wiki/How-to-save-your-AI-from-becoming-retarded
L.I can't think of anything to write! Is there any custom prompts other people have posted that I can use?
There are so many custom prompts. There is a smut section labeled just porn for those who only want smut ones. Check out this page: https://github.com/VBPXKSMI/Open-CYOAI-Project/wiki/Custom-Prompts-Index
If you're too lazy to use the wiki here's the pastebin: https://pastebin.com/KD4yN2Gc
M.What Wiki do you keep talking about? Do you have a link to it? (This is redundant if you're reading it on the wiki)
How the heck did you get this far without having read the thread OP or seeing the wiki link once? Okay whatever here's the link to it: https://github.com/VBPXKSMI/Open-CYOAI-Project/wiki
N. Is there a discord server?
Not now and hopefully never. Please don't make one. We really don't need one. I promise you the threads are enough for us. We do not need one under any circumstances.
O.Which model is the best for smut?
Absolutely not the pure smut trained one. No I am not joking or being sarcastic ignore that one it spews out pure nonsense. So far it seems the consensus is that the one trained on novels is the best.
Download link here: https://mega.nz/#F!51ED1CSA!YbifAmboqSLCrr3Y78axqw!gttGFYqS
Pytorch version for Clover here: https://mega.nz/#F!4e5kRCIB!v7q0ItVjhhGcIqfZOZy9yA!1XYG3C7S
P.What is a model?
The most basic way of explaining it would be to say it's the text that the AI is fed to be trained on. It's the information the AI is drawing from to give you a response and get context for what you're telling it. As an example, the original is trained on a lot of choose your adventure text. So it knows what a dragon is and a wizard and a fireball ect.
Q.How do I install a model?
Place the model you want into your models folder in your G-Drive or in your models folder in your local version and Thadunge or Clover will ask you when you start up if you want to use a different model. For Thadunge it's AI Dungeon>Generator>gpt2>models For Clover it's Clover-Edition>Models If it still doesn't work after you place it in the folder and try to run it when prompted then do like pointed out earlier and take a screenshot and post it in the thread don't just show up whining about how it doesn't work for some reason.
R.Where can I download all of the custom models/where are they uploaded to?
Regular Models: https://mega.nz/#F!51ED1CSA!YbifAmboqSLCrr3Y78axqw
Pytorch Converted Models: https://mega.nz/#F!4e5kRCIB!v7q0ItVjhhGcIqfZOZy9yA
Pytorch DF: https://mega.nz/#F!ww4A0IKK!mEAePh464lPfZcCYIB4Hkg DF is specifically for having sex with dragons. No idea what the difference between the two is but one anon said they liked the first one more.
S.How do I make the Clover local version use CPU and not CUDA(GPU)?
Open config.ini in notepad and change force-cpu = off to force-cpu = on
T.Can my computer even run Clover locally?
Does your computer have at least 8GB of VRAM available? Probably yes then. Otherwise test it and find out. Depending on the model you need less VRAM but you'll have to check the models yourself for that. Will it run smoothly though? Again, you test it and find out. We can't possibly just know without a shadow of a doubt your computer that barely meets the specs will run it quickly or slow as molasses.
U.If my computer can't run Clover locally can it run Thadunge locally instead?
If you can't run Clover then you definitely can't run Thadunge or the official version locally with GPU. However you can try running it with CPU but don't be surprised if the response times are a few minutes long at the minimum.
V.What is the best temp and top_p for smut?
There is no definitive answer for this yet but in my experience 0.2 temp and 0.9 top_p have led to better results than most settings I tried.
W.How do you turn off the d20 command in Clover?
/set action-d20 off
X.This FAQ is missing a question I have and I swear it's important enough to be added! What do I do!?
If you have a question and answer you seriously want added to this page then submit an issue to the github and I'll add it as soon as I possibly can.