On your output, you should select it first, then use the selected item.:
output
[Lpheno = Furlength.selectOne] ... [pheno]
pheno
[Lpheno.pheno]-furred
Doing Perchance.org things…
Links
On your output, you should select it first, then use the selected item.:
output
[Lpheno = Furlength.selectOne] ... [pheno]
pheno
[Lpheno.pheno]-furred
On the ‘gear’ icon on the top left of each editor, you can disable the ‘IntelliSense’/AI prediction for both editors as well as the AI bug checking. You can also enable and disable it through a short cut ctrl+shift+space
to toggle. If you want the AI to predict on command, you can use ctrl+space
to run it.
Btw, it seems to load/hang a little bit each time the button to randomize is clicked, so I’ve made an adjustment with the code, so it loads and selects an item faster, so just change the following lines:
getTxtList(url) =>
if (!url.endsWith('.txt')) return;
let fileName = url.split('/file/')[1].replace('.txt', '');
if (!root[fileName]) {
const request = new XMLHttpRequest();
request.open("GET", url, false);
request.send(null);
let list = request.responseText
list = list.split(/[\n\r]/).filter(a => a)
// changes
root[fileName] = {}
let maxPerSubset = 5000
for(i = 0; i < Math.ceil(list.length / maxPerSubset); i++) {
root[fileName][i] = [... list.slice((0 + i) * maxPerSubset, (1 + i) * maxPerSubset)]
}
}
return root[fileName][Object.keys(root[fileName]).selectOne] // change
Most of the code that runs on the /ai-character-chat is on the page itself. You can click the ‘edit’ button on the top right and it would open up the code panels. On the HTML panel is most of the code while there are also some on the Lists Panel. It is also mainly based on the OpenCharacters but with some tweaks and changes.
If you are talking about how the AI model is set up on the server, I’m not sure, but I’ll ping the Perchance Dev for insights.
Glad to help!
Might be because it is running a JavaScript script. Try to disable it first, reload the page, then run the /print
on the field. It should show something like this after running the /print
:
Then you can click select all, then right click the window -> print:
By creating a backup, I mean something like the version control on perchance in which every save creates a version of the previous generator without the changes. If you are editing with the collab link, and you aren’t the owner/creator of the page, there is no ‘backup’ icon on the top right, meaning the changes aren’t saved in a version although the changes are saved on the generator, thus I think it requires the owner to manually save to create a backup version, unless I’m mistaken.
As for the edit password, I’ve attached an image on my previous comment of the example prompt that shows up when I click the “Collab Link” on another generator that I don’t own. For example, on this one, I clicked the “Collab Link” on /beautiful-people and it shows this prompt:
Also, when I previously created a collab link, it asked me for an edit password, but now I don’t see it, which is why I provided the Password: hello
previously since I’ve set it to that, and checking with you if there should be an edit password if you don’t have the collab link.
You can add this function on the lists editor:
url = https://user-uploads.perchance.org/file/78435dca8c2d1750fe1affd02ef2886f.txt
list
[getTxtList(url)]
getTxtList(url) =>
if (!url.endsWith('.txt')) return;
let fileName = url.split('/file/')[1].replace('.txt', '');
if (!root[fileName]) {
const request = new XMLHttpRequest();
request.open("GET", url, false);
request.send(null);
let list = request.responseText
list = list.split(/[\n\r]/).filter(a => a)
root[fileName] = list
}
return root[fileName]
@[email protected] pinging Dev. I’ll also attach a previous comment of the Dev regarding privacy of the AI tools and advertisements.
The problem with that is the generator 78435dca8c2d1750fe1affd02ef2886f
doesn’t exist. Meaning, the page https://perchance.org/78435dca8c2d1750fe1affd02ef2886f
doesn’t exist. Thus, you need to create that separate page before you can import it. You can’t use the import to open up local files.
Here are some links on creating an Import/Export of another generator:
If you need some help with recreating the generator, feel free to ask 😊
Can you specify which parts of the chat you want to export into a readable format? I’ve recently made a custom code to create a print view for the current thread which you could print as pdf.
Not sure, but pinging dev for thoughts.
You could delete/edit the memories with the /mem
command if the memory generated is not in tune with what happened from chat. Sadly, there isn’t currently an option to prioritize one over the other.
Maybe this is a related thread?
I will reiterate what I said there:
You can check (and edit) all the memories the AI has with the /mem command, and /sum for summaries (if it has them). If you want to check which memories/lore/summaries are being looked up during message generation, I suggest opening the browser Dev Tools (F12 or Ctrl+Shift+J opens the console) before sending a reply, and when the AI starts to reply you should see the AI’s message generation process.
The way the AI generates replies is they query the memory (LTM or Long-Term Memory) and the lore. It might prioritize the memories since it is relevant to the chat whereas the lore might not be since it is for general knowledge/info while the memories are specifically about the current chat.
Currently testing the Collab with a few people on https://perchance.org/collab-testing-01-01
Here are some notes and bugs that I found:
Bugs
Notes
@[email protected] pinging dev. Though, you can try to set a temporary value to a character custom data since that triggers updates like oc.character.customData.test = ''
.
There is no imageUrl
or instruction
property/option to pass in the text-to-image-plugin
so it doesn’t work. There is no AI Image Editor (Image to Image) in Perchance.
Sadly, I don’t know any other way to troubleshoot this, so I’ll ping the dev again @[email protected].
The models used are quite old here in Perchance, so it isn’t as contextually aware as newer models. Here is a prompting guide that might help. I also recommend creating the prompt first in No Style, then delve into styles later.
Hi there, I’ve modified your code and here is the working code. Just specify your API Key and User ID on the top variables, then upon loading the custom code, select the voice you want to use.