INSTALL BEST UNCENSORED Roleplay TextGen UI LOCALLY in 1 CLICK!
Вставка
- Опубліковано 28 жов 2024
- The SillyTavern WebUI is by far the best UI for roleplay using local LLMs currently available! It makes it really easy to run your favorite UNCENSORED open-source AI LLM models on your local computer for absolutely free and have fun doing it! With custom backgrounds, animated characters, automatic voice recognition, text to speech, Group chat, dynamic audio and MORE! An incredible UI for your Ai!
In this video, I'll show you how to install the SillyTavern webui on your computer in 1-CLICK! Plus, I'll showcase the most common and fun use cases of the webui so that you can start having fun with it right now!
Have you managed to install the SIllyTavern WebUI? Let me know in the comments!
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
SOCIAL MEDIA LINKS!
✨ Support my work on Patreon: / aitrepreneur
⚔️ Join the Discord server: bit.ly/aitdiscord
🧠 My Second Channel THE MAKER LAIR: bit.ly/themake...
📧 Business Contact: theaitrepreneur@gmail.com
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
✨ PATREON LINK: / aitrepreneur
SillyTavern: github.com/Sil...
SillyTavern Xtts: github.com/das...
SillyTavern-Extras: github.com/Sil...
Background remover: www.photoroom....
TextGen Install Video: • INSTALL UNCENSORED Tex...
Coqui TTs install video: • CREATE UNCENSORED AI V...
SD install Video: • INSTALL Stable Diffusi...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
►► My PC & Favorite Gear:
i9-12900K: amzn.to/3L03tLG
RTX 3090 Gigabyte Vision OC : amzn.to/40ANaue
SAMSUNG 980 PRO SSD 2TB PCIe NVMe: amzn.to/3oBR0WO
Kingston FURY Beast 64GB 3200MHz DDR4 : amzn.to/3osdZ6z
iCUE 4000X - White: amzn.to/40y9BAk
ASRock Z690 DDR4 : amzn.to/3Amcxph
Corsair RM850 - White : amzn.to/3NbXlm2
Corsair iCUE SP120 : amzn.to/43WR9nW
Noctua NH-D15 chromax.Black : amzn.to/3H7qQSa
EDUP PCIe WiFi 6E Card Bluetooth : amzn.to/40t5Lsk
Recording Gear:
Rode PodMic : amzn.to/43ZvYlm
Rode AI-1 USB Audio Interface : amzn.to/3N6ybFk
Rode WS2 Microphone Pop Filter : amzn.to/3oIo9Qw
Elgato Wave Mic Arm : amzn.to/3LosH7D
Stagg XLR Cable - Black - 6M : amzn.to/3L5Fuue
FetHead Microphone Preamp : amzn.to/41TWQ4o
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Totoro
TNSEE
RG
Judy Godvliet
Gluthoric
Thank you so much for your support on Patreon! You are truly a glory to behold! Your generosity is immense, and it means the world to me. Thank you for helping me keep the lights on and the content flowing. Thank you very much!
#GPT4 #GPT3 #ChatGPT #gpt4xalpaca
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
WATCH MY MOST POPULAR VIDEOS:
RECOMMENDED WATCHING - All LLM & ChatGPT Video:
►► • CHATGPT
RECOMMENDED WATCHING - My "Tutorial" Playlist:
►► bit.ly/TuTPlay...
Disclosure: Bear in mind that some of the links in this post are affiliate links and if you go through them to make a purchase I will earn a commission. Keep in mind that I link these companies and their products because of their quality and not because of the commission I receive from your purchases. The decision is yours, and whether or not you decide to buy something is completely up to you.
HELLO HUMANS! Thank you for watching & do NOT forget to LIKE and SUBSCRIBE For More Ai Updates. Thx
Ok, but not because you told me to... Well mostly because you told me to
Morning! Thanks for your videos. I've watched them for a while now, and installing all this stuff is a lot easier now.
I use TabbyAPI with SillyTavern nowadays.
SillyTavern is nice. Kobold is cool too. Excited for this video though. You edit your videos so nicely! 🎉
@@brianheskin1654 you can use kobold as the API instead of textgenwebui silly tavern supports both. I find that kobold is better for story creation in silly tavern.
Is there a better updated model than the stable diffusion XL model you used this summer ?
Wow im impressed how this technology is improving
Can't wait to see where this all leads in the end.
Dude, I was looking for a good tutorial on this yesterday but then decided I need sleep like a normal person. Always good to know that the AI overlord has my back. :D
I am glad you are back making videos. missed you
Honestly considering the utter speed at we're moving I won't be surprised if by this time next year We'll be using Sillytavern to convert our chats into comic strips.
And considering the recent Text to 3D advancements Creating Vtuber avatar level meshes, will like be as easy as using Stable Diffusion.
Total unrestricted acceleration!!!
All we need now is a reciprocating linear actuator, a speed controller that silly tavern can control and a vr headset. and a fleshlight and some duct tape. and an industrial drum full of the good old water based lubrication. It's for a science experiment is what i'll tell my wife.
@@Smashachu I CAN now...
3d in vr ... in 2weeks i will demostrate this
@@contentfreeGPT5-py6uv You really want me to buy a VR headset, do you? :D
Do you have some more info on the vtuber thing? Personally not a fan of vroid and its restrictions while picking up blender as the alternative is a giant skill crater of its own. So even if I can't get exactly what I'd want out of AI, landing close enough to edit (or find someone willing) would go a long way.
There are AI voice models that can clone the voices of anime characters. I would like a tutorial on connecting that to Silly Tavern, so we get Text to speech voices that match the characters we talk to.
I like how you think
This is already a feature, check out the RVC extension in SillyTavern
Also we could create op super awesome npc fighters and cookers to do expeditions ect if we had dmmo nerve gear ect lol 😅
Simple coding. You can figure it out fairly easily with chatgpt.
Lol this already exists. Just look for selena1 by hobl games
Had a Python Code with TTS for Chat GPT or Local LLM in Combination with V-Tube Studio and Voice Meter with costum BG. But that is so much better. Totaly missed the live2d feature in Silly Tavern and Group Conversation is awesome. Your Channel is the best out there for such content.
HOly Moly!!! Best $5 I've spent in a while! Thank YOU!!
Now that's the stuff! I'll sure be using this to get my daily dose of open source LLM roleplay.
I’ll admit, this is pretty amazing and definitely would facilitate a more immersive dialogue with your AI companion. Neat, A bit too many moving parts for me though.
what is currently a good simple solution for uncensored story writing?
I only do role-play So I’m not able to give you any suggestions on storytelling sorry.
@@mygamecomputer1691 are the ones you use not able to write stories? As opposed to 1 on 1 roleplay
I’ve never bothered doing story generation because I always feel like if given an opportunity to type a lot of text it never does as good a job as when it’s nudged along as it is via a back-and-forth of chat.
@@mygamecomputer1691 well I mean you still do you own input every few paragraphs or correct some things to go the direction you wanna go.
Dude, You just gave me so many ideas. I already have spirits, 2d, an voice for a character I've been working on for some time... Thank you , thank you
Happy New Year 🎉 Thanks for all your video contributions! Wish you a even better 2024
Great that you show what's actually possible with sillytavern. I have an addition, there's an extension for automatic1111 called Rembg that let's you automatically cut out your generated characters. Then you can do everything locally and don't have to fiddle with the website. Live2d models are a dream, but as you stated correctly are hard to create. Another solution I had in mind would be to create gif files with the animatediff extension, then extract the character for your gif sequenze with batch process of Rembg and integrate it as expressions gif. Ah and for the xtts extension: still no push to talk button? Little function with lots of convenience. Thanks for the tutorial and keep it up!
I can't belive this released today when yesterday I was pulling my hair out from frustration with AI repeating itself!!!!!!!!!
Woo lets goooo
Hey, I'm a model creator and work with undi on a new model, could you maybe include it? It's called noromaid 0.2, there is a 7b, 13b and soon also a 20b.
Could you help me setup that text generation web UI? I'm lost and dont even knwo what or where to google
what's the website for ai characters again?
Did you find it ?
I missed it too
This makes me want to make a Konosuba D&D RPG adventure xD
I don't want to roleplay. I just want a bot I can talk about true crime cases, conspiracy theories, and other weird shit with that isn't going to go "I'm sorry Lem I'm afraid i can't do that" or "I can't in good consciousness talk about this subject as my objective is only to help and enlighten others"
create one that will roleplay as someone who talks about those things
Use mixtral
I know how to do this but I'm going to subscribe to your patron because you put so much work into keeping us updated on new features. Thank you for your hard work 👍
If there would be a pen&paper style adventure generator based on AI as a GM that would be crazy
Definitely doable with taskweaver/autogen. It's just a matter of actually doing it and making it work nicely.
I tested a few language models for this, and to be honest, GPT-4 was 'playable' in the ChatGPT environment. However, we had to remind the Game Master (GM) of the concept of rolling the dice now and then. Also, make sure to physically throw your dice because... let's say GPT-4 threw it a bit too often off the odds, leading to too many situations where 'this went pretty well for all of us.' Remember when the architect said people didn't accept the first matrix? That's why; the pen-and-paper felt too easy.
I GUESS it's time for me to train a Stable diffusion model to create Live2D textures to apply mmmmmh...
(it would not be great but still...)
Is there any easy way to give these characters a memory? I know there were several extensions for Oobabooga in the past that tried to do that for example by injecting auto-summarizations in the context and there's also MemGPT which I still couldn't get running. But is as of now, is there something that seemlessly integerates into Silly Tavern or in the backend? 4096 context size isn't very much and with all these RPG stuff and emotes and even multiple characters, it gets filled up quickly.
you try 'long_term_memory'?
@@SlickSonicTitan What do you mean? I only found wawawario2/long_term_memory on Github which is for OB and long dead as it seems.
Can we make them use our voice from microphone or custom voices ? I mean i want to make an anim character that talk in real time when i talk into my mic, could be a cool overlay
How about making a video about AI Roguelite, the AI game that can generate most everything! It can use custom models so it looks better then on the storepage, can be connected to a1111 and use GPT3.5 or connect to a local kobold model.
An interesting thought since we're on topic. So far most locally run models have captain out a 30 Billion Parameters for the RTX 4090. If the Specs rumored are truth the RTX 5090 Will likely be able to handle 70Bilion Parametes.
I wonder when this larger AI models will be given more prominence, especially with AI first Hardware promising quite the quantum leap over the next few years.
Ot likely. 5090 will have 32 gb vram maximum. There is a probability it will atill have 24
@@Airbender131090 Have you seen how efficiency in running AIs has increased just this year. The 5000 series are still quite a few months aways. At this point (assuming 32Gb) It would be weirder if it couldn't handle a 70 Billion parameter model, either when or shortly after it comes out.
Nvidia won't release consumer gpu with more than 24 gb vram anytime soon. They need to sell A100/H100 and if you are able to buy 3x5090 and get 96 gb vram for say 6k usd then there is no point for you to buy A100 with 80gb vram for 20k usd
I found out by creating a small easy setup online custom ai platform from any available website is the limitation. But after research online in ai blogs i found anyone hypothetically can create their own code language based on new ai coding am/ai conversion though stuff so, it's like unlimited thinking space for that thing, but less focused. And just to give an idea anyone could youtube videos of attempts at massive info testing/complex testing or other testing finding the results of the error or non error data
16:53 Where do we find the torch string to copy from?
If anyone has issues installing xtts make sure you have a supported python version (ver3.8-3.10)
xtts is installed, the server is running via the launchXTTS. Sillytavern UI TTS says provider loaded and I selected XTTSv2 and enabled it, reload it, selected a voice .wav, but the text is not read out load :/
EDIT: Okay I got it to work. I need to enable the TTS option "auto generation" else it won't do audio for me.
Thank you, Torch and more dependencies required python 3.10 and the 3.12 was throwing errors
I think something changed with Silly Tavern between 1.1.0 and 1.1.2 because Silly Tavern can't connect.
What model are you usign? You missed that completely no? Thats rhe most essential part
Hey Aitrepreneur, I've just subscribed to your channel as I have recently started playing with stable diffusion and find your videos really helpful. There is one thing I'm confused about and having trouble with, research doesn't seem to answer the question.
If I'm training a big model in Dreambooth , it could take 40 hours or more. I'd like to be able to stop training and resume later, but it always goes back to epoch 1 , step 1. Could you do a guide on how to resume interrupted training?
Many thanks :)
tu es au top, mec!!!
Nice! Thanks.
You deserve million sub for showing us this impressive stuff❤
Thanks man I start playing around with sillytavern because I saw you vids and then It's good, thanks for the guide vid.
PS. It's seem I can't made coqui-TTS to work. even with the other like alltalk TTS that seem easy enough to install.
Bro my boy. Make a series on it
Can we use silly tavern in place of the interface for the locollized llm's like Ollama 3.0? Better yet can it be used in tandem with the LLM interface?😊
The commands are not in the description :(
I appreciate the video so far but I have one major issue. I can get everything up and running but that's about it. The characters give me extremely short answers. about a 5 word sentence with little to no descriptive language like what they are in greyed text. The default character does this and any character I add does this as well. Am I setting up something wrong? Or am i possibly missing a whole bunch of steps not in this video?
hardware - 3060 laptop gpu, i7-10750h, 16gb ram
Do you still need help with your characters? I've found there are a bunch of extra settings he doesn't go over that help understand character responses better.
This is really good. Though both chat models (Tiefighter and tavern something) are slow for me with a rtx 3050 ti. Had to switch to the version that runs on CPU as well as GPU. Still, really amazing for something running locally on my own PC.
wait I'm a bit confused, it's running whatever llm model we're using in oobabooga, right? Silly tavern's performances will vary on how good the installed LLM is, correct?
Me: oh yeah talk dirty to me.
Her: 911 was an inside job
Me: Oh keep going.
Her: We never landed on the moon.
Me: Uggghh almost there.
Her: The NSA is logging your keystrokes
Me: AAAAWYOOOOOO
That conversation sounds non coherent at all, (if it's true and not just sarcasm). There are way better models than that. We had non coherent models like that 3 months ago, but most newer models are OK.
Dont kink shame me bro @@Slav4o911
Your model did not load. You are hearing samples from the tts engine.
It's insane what you can do with AI...
Wow just stumbled across this. This is absolutely amazing and will be perfect for my needs on a project I'm working on.
this gun be good
It will :P
ill wait 2 more years
*months
Weeks
Nah he was right, years. All these ui things are improving amazingly, but every local LLM is trash, and have barely gotten any better in the past months, if at all. They always spit out nonsense that is only loosely related to the prompt and have low context retention.
Seconds*
@@gotenks157 yeah, you're right. This is going pretty fast but it's all far from perfect. In a couple of years this is gonna work flawlessly as some finished products just like Photoshop or Excel do.
Failed to build tts mojimoji
ERROR: Could not build wheels for tts, mojimoji, which is required to install pyproject.toml-based projects. What does this mean in particular?
In my case, it was a problem with the "app installer" program through the windows store. I also had issues with the store, so I searched how to reinstall the store, did that, then updated the app installer and it worked. I still had to relaunch the file a few times before the majority started to install, but it works fine now.
@@NightSwift271 that did the trick windows found my UAC was off for the store which was blocking part of the installation process.
@@vidmaniacz Glad to hear it!
Try enabling "LongPathsEnabled" in the Windows Registry.
wow...the girl went straight to the point there. First thing she says is "worship me". Hard to fathom what this will do the psyche of millions of lonely men around the world. Technically it's wonderful. Sociologically it's a problem of a kind and magnitude as we've never seen before
Depends on who you are. The Middle East and East Asia are full of young men without much prospect for marriage or upgraded social circumstances. An AI 'girlfriend' that keeps their frustrations under control will be a tempting approach for some governments, I am sure. Because history shows that when young men have few prospects...history happens.
But I agree. It's a little creepy and a departure from what I'd call normal.
I just want to program Trump's voice to talk to me in my car, like KITT in the old Knight Rider series. I'd crack up every cop who pulls me over. Making them laugh sometimes gets me out of tickets when I get stopped every now and then.
Aight, time to learn Live2D, I reckon
there is no link in description to download git
Any idea why the expressions aren't changing? i made an expression pack and used local and extras and it never changes the expression
Aqua and Megumin?!! Ok I gotta try this xD
Got the uncensored ai working thanks to one of your vids. Ran the 13b version but it's really slow, so going to try the 7b version of this. Great for DnD games!
Can you please put the links in the description?
"I am not a goddess, I am the goddess of water"
Ah yes. I see.
Hello ! I want to know if you can do a vidéo about the Ai danse that we can see on Instagram or tiktok ? Maybe we can do some animation with it ? And conbine with other model 🤷♂️
how do this one works when compared to janitorai? I've been testing janitor, and while it is good, there are times where the character forgets pieces of the conversation...
so to make sure after installing all of it. you can then run everything locally and not connected to internet at all?
I'd love to try to use it for graphically described cosmic horror roleplay. You know something along the lines of Call of Cthulhu.
*Edit* I forgot to mention nonshrexkual content. It would be truly horrifying to fetishize Dead Space like creatures.....and now I realize that is a non zero number of people who have. That's just... 😳 wow.
let's gooo
You my sir, are a genius... I'm going to your patreon
A good idea is a video about LLAVA 😢
AI Dungeon already does a pretty good job at generating smut, though the quality can be about as rangy as real chatmates I've played with, at least I don't have to be polite when it's really terrible. This would interest me if it at least generated good images, which AI Dungeon absolutely cannot do.
When are they going to have posable options and interactions between characters and change of attire in game. That would raise the immersion even more
DO I need to have a PC that capable of running AI locally? Or I can just use a weak PC and assign API from other platforms?
This is awesome! , thx for chismas gif, if i want to use in spanish Wichita chances should i do?, i tried a few llms but don work fine on spanish
@Aitrepreneur question, its possible to train this model to execute to have automatic insertion of key entities, topics, and keywords creating sentiment analysis uploading data to the models DB?
What is the website called for presets?
Wish this video were came out a little bit earlier before i started yesterday... 😂 But thanks for the content ... This video is very useful...
Need some help or guidance ControlNet doesn't seem to have those option in SD now, or am I missing something?
Wow the Leyley family is just taking over all of December. The TTS is labeled as a "server" though is it running completely local?
a server can run local yes. once it's started you have a local IP you connect your sillytavern ui to.
bro I don't see any live2d folder in the public folder even though I have insallted the extension :(
I enabled the SAME character as per your video, but the character does not appear. I can chat fine, but no image/animation comes up...What do I do to work around this?
How big of a gpu do i need? And if its too big, could i just get extra gpu's or can the model only run on one gpu?
When I get to the step where I'm starting the UI, the cmd window stops on this line "To create a public link, set `share=True` in `launch()`."
cant do anything else.
Hi, thanks for the tutorial, there is a problem with the installation of Oobabooga, the cmd window pop up and disappear. Windows defender does not block anything based on history.
the problem was that I tried to install it in a hidden folder
Why are you not making more videos
For some reason I can't load the Kobold 13b model, I have a vicuna uncensored 13b that does load. Can I use that instead, or another besides kobold to try?
Did you do used to do Skyrim mod videos? Sloots?
19:00 After getting character to speak for the first time on the next reload I get
gpt_cond_latent, speaker_embedding = self.Model.get_conditioning_latents(speaker_wav)
| AttributeError: 'TTSWrapper' object has no attribute 'Model'
Could you please help me with that.
Thanks for your hard work.
"Megumin talking with Trump's voice, working as intended."
Is there a mobile version of this interfaces? so the model runs on my pc but I can use it on my phone
Yeah you can visit the url on your phone, if that's what you mean
when I talked to the AI I had and Intern API Error. Does anyone know what is this? One of the cmd's showed an error saying that a tokenizer wasn't selected, too
Hello man, real talk. I've only found one person that dealt with this maybe you know some thing about it. I get an error Total sequence length exceeds cache size in model.forward, no clue how to fix it. It happens when I'm generating a message
An AI companion is kinda the end of the road...
I need something less censored than GPT for writing violent giant monster scifi, but I don't have the hardware capability of running an LLM locally. Are there any services like ChatGPT that folks can recommend? If I get one more AI suggestion for peaceful giant monster dance-off instead of the battle it refuses to help me with, I might scream!
Is there a way to get this to not default to installing on the c drive? Mine is solid state and is really just for system files, very little space to work with.
I'd pay good money for this in a game.
Brackeys? Is that you? o.o
i got this just now"Argument alias was not recognized for the current command: '-y' " did i do something wrong?
Should i just manually install since i already have the WebUI?
what is the website mentioned at 11:57 it was hard to understand you, please let me know thanks
chub.ai
I tried this, but the replies are a little slower; it takes a few seconds to get the reply from AI. I am on 3060Ti. is it normal or can i do anything to speed up the process? Like i am running a 7B parameter model. Should I switch it to anything lower (if available) or...?
I am on the same graphic card, And it takes around 10 seconds to get a reply which ruins the experience
Brother larping a conversation with himself.
@@SaltyMaud it’s not me its AI 🤣
@16:53 copy and paste the 4th command I see no such command in the description.
yeah sadly you got to type it by hand.
Sorry to ask this here,
I needed help with fooocus Ai, I'm having a connection error, Is there a fix for that ?, using a 1660ti 😅
Promised commands are not in the description.
on the silly tavern xtts install page if you're still looking
Is there anything more about the journey than chat? Something where the world could change dynamically, new characters could appear and old ones could disappear or their status could change and the MC could also change dynamically in appearance and status. With the possibility of NSFW, of course. Especially NSFW in regards to the MC's position and his/her status and appearance.
What is the best uncensored 7B . I can't run anything higher sadly.
Is there some way to run the xtts api server without using this bat file? Cause it does nothing....
Dear Aitrepreneur!
Can you share the way how to make a talking avatar like the one you have at the corner?
I would be very thankful for that.