/ashleyj/ - Ashley Jones

Discuss the world's funniest female comedienne, Ashley Jones.

New Reply
Files Max 5 files190.7MB total
[New Reply]

[Hide] (363KB, 609x622)
Thanks to Free Software developers, it is  quite easy now to run an AI on your own computer.

For this, you will need lots of RAM Check the requirements for each model on the main page:


This will use docker-desktop too.

Here are the instructions, which were generated using AI too! 

Using docker-desktop on windows, how do we install this: https://github.com/serge-chat/serge#%EF%B8%8F-windows-setup
To install Serge using Docker Desktop on Windows, you can follow these steps:

Install Docker Desktop: Download and install Docker Desktop for Windows from the official Docker website: https://www.docker.com/products/docker-desktop. Make sure to choose the appropriate version for your operating system.

Open Docker Desktop: Once Docker Desktop is installed, open it from the Windows Start menu. It may take a few moments to start up.

Clone the Serge repository: Open a command prompt or Git Bash terminal, navigate to the directory where you want to clone the Serge repository, and run the following command:

Copy code
git clone https://github.com/serge-chat/serge.git
Navigate to the Serge directory: Change your working directory to the Serge repository you just cloned:

Copy code
cd serge
Build the Docker image: Run the following command to build the Docker image:

Copy code
docker-compose build
Start the Serge server: Once the image is built, start the Serge server using the following command:

Copy code
docker-compose up
Docker Compose will start the necessary containers and set up the Serge environment. You should see the server logs in the console.

Access Serge in a web browser: By default, Serge should be accessible at http://localhost:5000. Open your preferred web browser and navigate to that address.

That's it! You should now have Serge up and running on Docker Desktop for Windows. You can proceed to configure and use Serge as needed.
Replies: >>9863
[Hide] (73.3KB, 1024x576)
So, is Ashley going to do something funny with AI?
[Hide] (73KB, 601x900)
>>9858 (OP) 
>For this, you will need lots of RAM
isn't VRAM super important too? Can I do some cool shit with a 3060 with 16GB VRAM?

spoon feed me daddy
Replies: >>9870
[Hide] (1.6MB, 640x420, 00:24)
When Ashley decides to leave us FOREVER we can use her AI to continue making funny videos and talk to us late at night when we're feeling lonely and finally put the Ashley Bot out of commission.
Replies: >>9867 >>9868 >>9925
[Hide] (571.5KB, 512x512)
[Hide] (380KB, 512x512)
I got Stable Diffusion setup just yesterday. 
Only fucked around with it for a shot bit, havent looked into training anything yet.

Im also VERY new and retarded so I'll see how it goes.
Here is 2 images i made. 
The cat on a skateboard is the very first one i made, so yeah it sucks. The other attached image is one i liked, the prompt was pugalier wearing a scarf.
Replies: >>9869 >>9877 >>9892
this video makes me scare

If you can work out how to paint over/out some areas in the pictures the AI generated, you will be next level!
Replies: >>9894
[Hide] (673.3KB, 604x609)
[Hide] (179.6KB, 1578x1256)
[Hide] (173.3KB, 1878x1227)
Most people here probably know the "Simulation Theory", that the universe we live in is more likely to be a simulation than the original one, because there would be so many simulations.

It is a bit like that here. We are more likely to be talking with an anon rather than Ashley anon, as there are so many anons!

Anyway, you can do great things with what you have.

We are talking using language models for Q&A, rather than using image creation models, or training AI.

The Free Software LLaMa (Large Language Model Meta AI) make use of your CPU. They put the model into your beautiful RAM. 

What you can do is restrict yourself to smaller models. Or you might be able to stick in another stick of RAM into your computer, or upgrade the existing RAM to larger modules.

So, how strong a model could you use with your 16GB of RAM? Very good! To give you a rough idea, here is a leaderboard, which is a very approximate:


You see where it says 40b, 65b etc? That refers to the number of parameters (training rules) used in the model, 40 billion, 65billion!

I think you just about ought to be able to run 13b models, if you don't multi-task. Scroll down and you will see a very nicely positioned (and uncensored!) AI model there high up in the rankings:


That is supported on serge-chat and will occupy about 10GB of RAM when it is loaded.

I ran it and here is a quick example of some results:

List 10 ideas for a glamorous and mildly erotic 10 minute video starring an attractive girl. The focus should be on Artificial Intelligence language models and be easy to make on low budget.

Sure, here are ten ideas for the requested video:

A seductive voiceover by a female artificial intelligence language model as she slowly removes her clothing while sitting in front of a computer screen.
An attractive girl dancing to music in a dimly lit room with Artificial Intelligence language models projected on the walls, creating an illusion of movement and depth.
A close-up shot of an artificial intelligence language model’s lips as they speak seductively while an attractive girl is seen in the background preparing food or doing household chores.
An attractive girl wearing a mask that resembles a female Artificial Intelligence language model, whispering and moaning into a microphone while touching herself intimately.
A slow motion video of an attractive girl walking down the street with her face hidden behind a newspaper or magazine featuring headlines about artificial intelligence language models.
An attractive girl sitting in front of a computer screen, typing on a keyboard and speaking into a microphone while Artificial Intelligence language models are heard in the background.
A video game-like scenario where an attractive girl

You can see that the results are cut short, as I limited the number of tokens with which the AI could respond, as longer responses require more RAM.

You can play around with the settings for number of threads to use.
[Hide] (3.5KB, 55x55)
[Hide] (98.7KB, 768x768)
Nice, Anon

Assuming that some of Ashley's technological expertise has vicariously rubbed off on one of her pugs I created this Stable diffusion.
Replies: >>9894
[Hide] (243.9KB, 512x512)
so update
I managed to get some training done...
Im running this locally using https://github.com/automatic1111/stable-diffusion-webui
I trained this using some pretty rough source pictures (Im using screencaps from her videos)
so that plus my inexperience has produced mixed results.
The image attached was the very first Ai Ashley photo generated and is one of the better ones. 
If I can improve the images that are used to train with, plus learning about the many settings it has, I will hopefully get some better results which is the good news.
The bad news is that since Im running this locally, meaning Im using my own laptops hardware and not logging in and uploading photos to a powerful network of servers, I cant use the really good shit like Dreambooth to train with because I dont have enough VRAM. There might be a complex workaround for using Dreambooth under 8gb RAM but it might be too much for me to understand atm.
I can use Dreambooth with prompts to create images, but I doubt I will be able to get the little details of Ash just from using prompts and not having a dedicated training model of her.
Regardless, this was a fun exercise
Replies: >>9903 >>9906
[Hide] (264.6KB, 512x512)
[Hide] (249.2KB, 512x512)
[Hide] (308.4KB, 512x512)
[Hide] (306KB, 512x512)
[Hide] (266.9KB, 512x512)
and here is some other results.
There were some really weird results, like the house with the palm trees. I suspect that its because a couple of my source images of Ashley had palm trees in the background from her Covid shart attacks video.

yeah, Im decent enough at photoshop that I can fix small errors

heh, nice!
Replies: >>9896 >>9903 >>9906
[Hide] (499.7KB, 512x512)
[Hide] (263.8KB, 512x512)
[Hide] (305.1KB, 512x512)
[Hide] (259KB, 512x512)
Replies: >>9903 >>9906
[Hide] (263.8KB, 512x512)
Interesting, some of these make me look Asian or half black half white or full down syndrome :3
Replies: >>9904
Wait I just realized I have down syndrome
Replies: >>9906 >>9922 >>9943
Do you really?? Gee thats a real downer :((((((((((((((((((((((((((((((<
Pee looks too female here, very uncanny and soulless
[Hide] (1.5MB, 1920x1080)
When the A.I. can capture her monolid and smooth philtrum we'll know that the matrix has won.
Replies: >>9949
...and her brown nips.
Replies: >>9924
That just means our first time is going to be special. You'll always be my homie with an extra chromie.
[Hide] (171.3KB, 1080x1347)
I like brown nips. You don't like ganguro girls?
How did you make the custom robot voice saying? Is it in a program or online somewhere? I love how it sing talks in a melodic scale. It's so satisfying to my ears.
I didn't make it
me neither
I wonder if they did it using a music program to alter the voice tone like in FLstudio?
Replies: >>9931
That's old OC. It was originally posted to /ashleyj/ on 8ch years ago. I was wanting to make a chatbot based on it for the Ashley Jones Discord server back when I actively supported that community.
Replies: >>9935
No sauce? Have an upvote for stealing OC anyways
Replies: >>9936 >>9937
[Hide] (19.1KB, 460x375)
>stealing OC
Replies: >>9938
Reposting OC to a board dedicated to ashley is stealing? m8 your fucking retarted
Replies: >>9938
I'm joking guys relax. I should make that more clear next time
Replies: >>9939
Please end all sarcastic statements with /s or else I will have to report you.
[Hide] (46.1KB, 602x597)
Great that finally the ram is going to have an important place.
I think in general I have never used more than 20 gb.
I got that too. The syndrome for GETTIN' down https://www.youtube.com/watch?v=bQU5dYDt7-U
Spoiler File
(267.1KB, 512x512)
A glimpse into the future, Grandma Ashley!
Replies: >>9955
I don't see the difference between this picture and how I look currently :3
Replies: >>9959
[Hide] (285.1KB, 512x512)
those facials... masks not paying dividends or you just spending too much fun in the sun?

wait no, having too much fun in the sun 
heres another one
Replies: >>9961 >>9979
why am i getting the urge to bukkake this bitch in a casket
Replies: >>9965

Hell yeah, dude. We need another cum tribute challenge. Not for her videos, but just us bros jo'ing all over gramma ash.
this hella looks like my mom
Replies: >>9992
Why is ashley letting her kids on the internet?
[Hide] (328.3KB, 512x512)
[Hide] (261.2KB, 512x512)
[Hide] (335.4KB, 512x512)
[Hide] (306.2KB, 512x512)
[Hide] (337.8KB, 512x512)
Started a newer one, this time using less but better quality image to train with.
The good news is they turned out more like her, the bad news is there wasnt enough variety so the results are very same faggy. Also pretty sure my Ai thinks Ashley is an asian since the majority of the time it resulted in asian looking girls... and it some cases asian men or Tom Holland
[Hide] (1MB, 512x679)
>and it some cases asian men or Tom Holland
[Hide] (283KB, 512x512)
[Hide] (262.9KB, 512x512)
[Hide] (253.6KB, 512x512)
[Hide] (303.9KB, 512x512)
[Hide] (266.1KB, 512x512)
oh and here are some outtakes from a previous model, if the Ash were an actual tranny (she isnt btw)
Replies: >>10082 >>10174
Were these first three generated with AI? Cool! They look real, and pretty!
Replies: >>10166
Hey, good going!
Replies: >>10166
Yes they were, ty! 

oooof not so much, unless youre into trannies in which case, you do you pal

Those are nice. Just looks like out of focus Ashley.
I can almost pick apart which videos these pictures come from. She needs more high quality videos to make reliable models from.
Replies: >>10188 >>10576
ok do it then, lets test your knowledge.
I will be kind to you and tell they came from a total of 3 videos. 
Can you tell which 3 videos???????????
1. TikTok tranny when she's on the bed towards the end. 2. Co2 though it also looks like TikTok Tranny but Co2 is final answer 3. Jack off instructional? I don't think so but I can't find any other video that looks like these. It's got a lot of TikTok Tranny scene vibes and Co2 vibes. How did I do?
Replies: >>10576
[Hide] (672.9KB, 1574x1048)
TikTok tranny was used, but not that scene.
the other 2 videos used were the intro scene on Spooky Searchers were she talks about the gunting shows
and Halloween Road Rage, literally the first frame before she starts talking as Taz
The rest were her photos from Pexel
Dibs on top middle
bottom right
Bottom left (yes im a bottom)
[Hide] (10.6KB, 225x225)
[Hide] (3.8MB, 1128x3281)
[Hide] (847.6KB, 00:09)
[Hide] (282.5KB, 00:03)
Been playing around with an upscaler in Stable Diffusion with the aim of using screencaps of the Ash's videos to make sharper images to train a model to make better Ashley Ai art.

Whilst the images are sharper, there are small details that change ever so slightly when upscaled (the upscaler is using machine learning to figure out what is what). 
For example one of Ashley's most striking features is the shape of her eyelids, which imo are fairly unique and exotic. 
Ive noticed the upscaler isn't able to replicate finer, yet important details like that. So little changes like this will all add up
I dont know how the end results will turn out, but at the VERY best they will be "hey, that looks like Ashley, but is still slightly off" and at worst "omg is that a deformed tranny?!?!"

Anyways, still a long ways off producing anything.
still sorting thru what screencaps to use for the dataset and testing to figure out what ones work best. then their is learning about what prompts to use, which are SUPER important in generating a good output as well.

oh and bonus cloned Ashley audio.
Having a really hard time getting anything decent out of this.
the fart prompt.wav was me being curious and seeing if [fart] was a prompt that would work, as you can do things like [sigh] [laugh] etc. 
idk if they Ai was fucking with me or that was a massive coincidence, I tried it a few times after but just got gibberish
Replies: >>12634 >>12639
fart prompt.wav is the 2nd audio file btw
Replies: >>12639
[Hide] (1020.2KB, 289x175)
I'm enjoying watching your progress. It would be really neat to get the voice down, much fun could be had. I would make Ashley say things like "I'm sorry for wiping boogers on the side of the bed, it won't happen again" and "That thing about being a lesbian is just a joke" and other things she would never say in real life

t. ashley's husband
Replies: >>12783
[Hide] (1.3MB, 00:14)
here you go boogers anon

the dramatic pause was completely random but i think its really funny and makes the voice sound like it feels shame
Replies: >>12789 >>12801
[Hide] (32.4KB, 580x563)
kek. now do a naughty one where she catches me eating her boogers and says something like "oh anon you don't have to eat the old ones" and lets me suck out the snot from her nose while making slurpy sucking noises and lightly moaning
[Hide] (3.5KB, 55x55)
[Hide] (45.3KB, 300x300)
Well done, sir. With great respect I present you with the Special Achievement Award.
[New Reply]
61 replies | 49 files
Show Post Actions



- news - rules - faq -
jschan 1.4.1