INSTALL UNCENSORED TextGen Ai WebUI LOCALLY in 1 CLICK!

are you sick and tired of chat GPT censorship do you wish you can do some uncensored roleplay on your computer without some big corporations spying on your conversations well boy do I get a solution for you hello humans my name is k a overlo and recently open AI unveiled a bunch of new tools and updates to CH gbt which is pretty cool however there is one thing that will never change and that is the fact that Chad gbt is still a sensored AI model that is controlled by a big Corporation you don't have any control and you cannot do whatever you want with it so how would you like to run any uncensored large language models with multimodel capabilities where you can use your own microphone to talk and receive audio answers all of that running on your local computer oh geez well this is exactly what I'm going to show you how to install today because to be able to run those un sensor large language models we use a really cool piece of software called uaba Tex generation web UI which is basically just an interface that will allow us to run any AI text models we want and to install it it's very simple but if you want an even simpler method I offer a simple oneclick installer for my patreon supporters which will automatically download and install everything you need just download the file on your computer double click it and then choose started Windows installation and then it to automatically download and install everything you need simple as that and the second way is of course the manual installation and I'm going to show you how in the thir feet that you need to install is G for Windows which is the tool that will allow you to clone the repository just click the link in the description you're going to arrive on this page then click on download for Windows then click on 64-bit G for Windows setup then once you have the exit file you're going to double click on it click on yes and then click install to proceed with the installation so then once this is done you're going to uncheck this box then click finish then you're going to click the second link in the description down below you're going to arrive on this page you're going to click on code and then you're going to click on this little icon to copy this this entire line then you're going to create a brand new folder on your computer in my case I called it ubaba but you can name it anything you want but try to avoid any spaces in the name because it might cause some issues then you're going to click on the folder path type CMD press enter which will bring the command PR window and then here you're going to type get clone and then you're going to press contrl V to paste the line that we copied previously and then you're going to press enter which will then CL the repository onto your computer and by doing this it will create a brand new folder called text generation webui and now if you go inside you will see a bunch of different files but don't worry we only need to use a few of them now if you are on Windows the only two files that you need to know is startor windows. bat and update windows. bat the first file is to run the installation and launch the web UI and the second file is basically to update the software this is really all you really need to know so then the next thing that you need to do is just double kick on the start windows. bath file which will then download and install all the files that it needs to run and then after a few seconds it will ask you what is your GPU do you have an Nvidia GPU an AMD are you on an Apple M series or the Intel Arc or if you don't have a GPU at all because yes even if you don't have GPU you can pretty much run any models you want using only your CPU which is really super practical but in my case say have an Nvidia GPU I will press a and then press enter it will then ask would you like to use Cuda 11.8 instead of 12.1 but this is only necessary if you have a very old GPU like Kepler and in my case say have a 3090 it is a pretty new card so for me and probably for the rest of you you're going to input n and then press enter and then it will continue with the installation and after a few minutes the installation will be finished and it will give you a local URL that if you press control and then left click on it will open the web UI and there you go we have now finished the installation and we can start having some fun right well not exactly because all we we did is just install an interface an interface that is used to run AI text models so the next thing that we need is an AI Tex model but how do we do that how do we download an AI model and where do we find them well first you're going to click on the model tab which will then look something like this and I know that if you are a complete beginner this sounds really really scary but don't worry I will explain everything you need to know this is actually really simple to use now in this section right here this is what we're going to use to download a model and to find an AI model you're going to use a website called huggingface doco which is basically an amazing website where you can find plenty of AI models data sets and test a bunch of AI softwares for free and the way we're going to download our models is simply to download them from a user code of the block which is an absolutely amazing guy that has a huge list of models available for you to download and as you can see there is really a lot of models to choose from so in that case the next question you might have is well first which one do I choose and then which format do I use because yes as you can see right here for some reason each model has three different version like for example you have here a model called vion 2 70b chat GG UF then you have awq and then you have gptq what do these models mean and which one should you choose well basically all of those are different format for the AI models the GG UF is a special format that is only used if you want to run an AI model using your CPU so if you don't have a GPU at all and you want to use your CPU to run an AI model you need to download GG UF models however if like me you have a GPU you have the choice between awq and gptq now these are basically very very similar in the sense that you're both using your GPU to run those models but one format is more recent than the other basically gptq is an old format model that has been recently replaced by a format called awq so basically if you have a GPU awq is the model format that you want so basically tldr if you don't have GPU you want to download ggf models but if you have a GPU you need to download awq models so now that we know that which one of those 2,362 models do we download especially because every model has a different size like for example this is a 70 billion parameters model and this is a simple 7 billion parameter model what is the difference well first the high the parameters the smarter the model is but also the more resources it's going to use which is why I'm going to tell you right now most of you even with a good GPU will not be able to run anything above a 13 billion parameter model if like me you have 24 GB of vram if you have like a 30390 or a 4090 the maximum size model that you can run on your GPU is a 33 billion parameter model that is the absolute maximum everything else is only for professional grade gpus and if I can give give you some advice on which model you can try the first small model that I recommend is called Mistral 7B which is basically a small 7 billion parameter model that is apparently very very powerful for its size and that only requires around 4 GB of vram to run or if you're using the CPU version around 7 GB of RAM which is pretty good and if you want a more powerful model like a 13 billion parameter one you can try this lat 2 30 billion th fatter model which is again really really powerful for its size and that uses around 7 GB of vram or if you're using the CPU version around 10 GB of RAM so then once we know which model we want to download how exactly do we do that well it's actually really really super easy all you need to do is once you find the model that you want to download you're going to click on this little icon right here to copy this entire name then you're going to go inside your wave UI inside the model Tab and then here are the download model or Laura in the first field you're going to paste the name that we just copied and then simply click on the button download which will then so download loed the model onto your computer simple as that and there you go after a few minutes the model is downloaded and for the model to appear in the list just click on this little refresh button right here to refresh the list and have our model appear however before we select it you might have noticed that if you use the CPU version of the model you will not have one but actually a bunch of different models to choose from because the block is actually really nice and provide a bunch of different version of the model that you can choose to suit your need and for each model you basically provide the size of the model the maximum vram required to run it as well as the recommended use case so what do you do when you have multiple models to choose from how do you download a specific model well first you're going to scroll up once again you're going to click on this little icon to copy this entire name then once again you're going to paste it right here and then you're going to click on get file list which will basically give you all the list of all the models available on that page and let's say that for this example the one that we want is the Q4 km which basically is for 37 GB in size and requires 6.87 GB of RAM which is described as medium and balance quality and also recommended if we now look at this list we can see that the Q4 km. GF is right here so I'm going to copy this entire line and then I'm going to paste that file name right here and then finally I'm going to click on download which once again will download the model onto our computer and there you go and then once again if you want the model to appear in the list just click on this little refresh button right here and now we can select our model however now that we have our models what option do we choose right here because yes to be able to load the model to load the model into the wave UI you actually have multiple options now see in our situation the only two loaders that we really need to pay attention to is Auto awq and Lama CPP now basically the auto awq is for loading models that are in the awq format while the lv. CPP is for loading models that can run on your CPU such as jjml or GG UF and in reality you don't even need to choose them because once you s the model that you want to run the loader will be chosen automatically like for example if I choose the awq the auto awq loader will be selected and if I choose the GF model L CPP will be chosen automatically so now if I want to choose my awq model I'm going to select it and then click on load and after a few seconds the model has been loaded successfully and now we can really have some fun and as you can see in only a few seconds I get a response from the model so now that this is done what exactly can you do with it well oh boy you can do a lot of cool things so for example let's say that I want to chat with a completely different character because yes this is not chat GPT you can actually talk with any virtual characters that you want characters that have their own personality their own way of talking and you can do all of that for free so like for example if you scroll down and you click on character gallery here in my case I have a bunch of characters to choose from the normal AI assistant that is very very plain very boring then you have this example character that if you click on it you can start directly a conversation called chiharu Yamada now I'm not sure if CH from an anime or if this is just a original character but basically if you start chatting with her like say something like he what are your hobbies in fact you can generate she'll basically answer well aside from an obsession with computers I enjoy reading particularly fin fantasy novels it helps me keep up with new technology etc etc I'm not going to read everything but basically each character has their own personality and you can create your own characters yourself like for example I created this Sandra the loving girlfriend that basically um well um plays the role of your actual girlfriend in an actual real scenario like for example right now you are inside a coffee on a date and you can start a normal conversation with her about your day or or start anything else and yes I really mean anything else and you know you can start some real normal conversation and obviously you can really do everything you want so yeah really leave that to your imagination oh and also do not forget to use asteris between the words to describe the action so like for example if I want to you know unzip uh something and run around screaming and now if I click on generate you will see that the text described in the action will be written in Gray whereas the text that you say out loud will be written in white and also the same for the character okay so now that we have this what else can we do and how can we make this a little bit more fun well how about instead of typing text you use your own microphone and then have the character respond to you in an actual voice yeah that's right and you can do that very very easily so for this you're going to just click on the session Tab and then here we're basically going to enable a bunch of different extensions so here for example you're going to click on 11 Labs you're going to click on srow and then on whisper and then you're going to click on apply Flags SL extensions and restart so now if you scroll down you will see a bunch of new stuff appear you're going to have here the 11 Labs text to speech which is basically a way to use a text to speech from the website called inent Labs which is kind of like the best website for text to speech audio where you can use the API to connect to the web UI so like for example if I click on my account then click on profile you will see here an API key and if I click on this little button right here it will make the API key visible so now if I select it then control C to copy it and then paste the API key right here then I'm going to choose the text to speech voice so let's say I want Charlotte then I'm going to click play text to speech automatically and now if I choose my character and I make a new conversation we get this hey love I hope you haven't been waiting long the traffic was insane and yes you heard it right now we have our character actually responded to us using action ual audio but here's the best part I can actually use my microphone to answer that character without having to type anything meaning that we can basically have a real conversation as if we were actual humans talking to each other I mean why need an actual girlfriend when you can talk to an AI come on now because if I scroll down you will see here a new box called whisper stt which is what we're going to use to convert our voice into actual text so now if I click on record from microphone and I see something like hey honey how was your day and now if we wait a little bit it was all right nothing too exciting how about you basically what I said in my microphone was converted into actual text where I then get an answer from the character I mean this is amazing this is really really cool and this is by far my favorite things that you can do inside new W UI just talk to an actual character using your microphone and receive an audio answer I mean this is just fantastic however the problem with this technique is that as of right now we're using an API from a paid website and although you have a free monthly quota of 10,000 words you're going to see that you're going to exhaust that quota very very quickly so is there a way to have a free option well the answer is of course yes and that free option is actually right here it's called srow so now if I click on this it will basically show a very similar box with a bunch of very similar options where here you can basically choose the language and the text to speech voice between a bunch of different versions of male and female voices which are actually pretty decent I mean not as good as the 11 Labs but still pretty decent so if I choose something like I don't know uh like 68 for example and I write some text so like hey honey how was your day with maybe a faster voice speed and click on preview we get something like this hey honey how was your day so yeah maybe not the best voice maybe I'll probably choose something else maybe something like this one and if I click on preview hey honey how was your day so yeah a little bit better so I'm probably going to keep this one just for the test when you really have a bunch of voices to choose from so you can do your own experiment yourself so now if I want to use it I'm going to click on activate text to speech play text to speech automatically and then show message text under audio player then I'm going to scroll up deactivate the 11 LS one and now if we start our new conversation again we get something like this hey love I hope you haven't been waiting long the traffic was insane so yeah there you go pretty much the exact same thing but this time it is completely free meaning that you can chat for hours without having to worry about about the money that you spend I mean this is definitely cheaper than a real date oh and also by the way if you want to have access to my Sandra the Ling girlfriend character which I got to say is by far the best character that I ever created I will actually make it available for my P supporters and to use it and actually upload the new character inside the webui you're basically going to have two files a Json file and a PNG file because to upload the character you're going to click on the parameters tab then click upload character where here you will have an option to drop a Json file as well as a profile picture so now if I select the Json file and I drop it right here and then I take the P file and drop it right here and click on submit and then if you scroll down and then click refresh you can select this new character right here that you can use and talk to for hours and believe me I personally had a lot of fun with that character yes a lot of um fun so yeah once again if you want to have that character that is only exclusive to my P supporters the link for it will be in the description down below however this is not the end this is only part of all the cool things that you can do inside the web UI and one of the cool things that you can do inside is to use the web UI to analyze an image and have a conversation about that image exactly like CH GPT Vision now obviously it's not as powerful as CH GPT but it's still pretty cool now first to be able to use this we actually need to download a very special model we need to download a model that is able to see the image and analyze it you can actually choose a bunch of models to use it with but the most most popular are called lava and you can either choose the 7 billion parameter model or the 13 billion parameter model so in my case I'll will be choosing the 13 million parameter model the link for it will be in description down below you're going to copy it paste it right here and then click on download so then you're going to click on model then you're going to select it with this one you actually need to use the auto gptq loader for the W bits you're going to choose four and group size 128 and then you're going to click on load and now you're going to click on session then click on multimodel and and then click on this Z button right here to restart to AI now sometimes it works but sometimes it doesn't sometimes it gives you an error and if it does don't worry it is actually pretty easy to solve because all you need is just edit this little text file right here that says CMD flags and under this line you're going to copy and paste this command that you will find in the description down below D- multimodel pipeline lavva 13B and then you're going to save the file and now if we Rel the wui again and then we click on the session tab you will see that the multi model extension will be selected by default now to be able to use this you actually need to select the instruct mode which is basically a special chat mode that actually uses a special instruction template now you don't really need to know about this it actually does it automatically just know that if you want to use this trick you need to select this option but then it will also give you a new field called send the picture when you can basically drag and drop a new image inside the chat so like for example if I upload this funny image of a monkey right here and now if I ask what is funny about this image and I click on generate this image depicts a monkey wearing sunglasses and a pink jacket which adds an amusing touch to the scene this humorous situation demonstrate how animals can sometimes appear more humanlike than other creatures leading to interesting comparisons and situations so yeah there you go just like Chad GPT you can actually use the lava 13 billion parameter model to analyze and describe an image and then have a conversation about it with which again sometimes is really just a gimmick but it's still a pretty cool technology and yeah there you go now you know pretty much everything there is to know on how to install and use the Uber bugat generation web UI and once again if you have any issues installing the software do not forget that I offer a one click installer on my patreon page as well as technical support if you have any issues so once again try on yourself and have some fun please say goodbye to the viewers and thank them for watching thanks for tuning in to today's video If you enjoyed it please give it a thumbs up and subscribe to my channel if you haven't already done so until next time see you and yeah there we are it folks thank you guys so much for watching don't forget to subscribe and smash the like button for the YouTube algorithm thank you also so much to my pure supporters for supporting my videos you guys are absolutely awesome you people are literally the reason why I'm able to make these videos so thank you so much and I'll see you guys next time bye-bye

Channel: Aitrepreneur

Views: 179,774

Keywords: ai, Artificial Intelligence, Machine Learning, Deep Learning, ChatGPT, GPT4, GPT-4, GPT3, GPT-3, OpenAI, Microsoft, open ai, ai chatbot, ai robot, open ai chat gpt, chatgpt examples, chat gpt explained, openai chat, ai tools, what is chatgpt, chatgpt explained, chat gpt, openai chatbot, ai chat, gpt 3 ai, gpt-4 demo, gpt 4, ooga booga text-generation-webui, oobabooga texgen, textgen ai, oobabooga github, llm tutorial, llm locally, TextGen Ai WebUI, Run LLM Models, local llm

Id: C-7jGYOGvy4

Length: 20min 52sec (1252 seconds)

Published: Wed Nov 08 2023