Recently the boss iβve been using responds with stuff like this? Even if i regenerate a response it ends up talking gibberish. Is this a problem on my. End or the bot itself.
Should i just remake a new convo??
#bot response turn odd?
remote vortex
weary vortex
Recently the boss iβve been using responds with stuff like this? Even if i rege...
You likely have your temperature too high or low. Can you share a screenshot of your generation settings please?
weary vortex
It looks okay but maybe lower it to approx 0.7 to see if that changes anything
remote vortex
It looks okay but maybe lower it to approx 0.7 to see if that changes anything
I changed it as you said, still a few run on sentences but at least itβs not giberish, i guess i can edit the response from there. Thank you
weary vortex
I changed it as you said, still a few run on sentences but at least itβs not gib...
No problem. Which model are you using btw?
frigid cape
I don't have the pro version, but this used to happen to me when i was playing arround with the settings.
Please put everything back to default and full fill everything requested for the model you're using and so.
If you're using Cosmos RP Pro, i suggest you to put the Contect Token to 16.000. Try to play with the Temperature from 0.75 to 0.95. And the Max new token from 150 to 250, 250 to 500, 500 to 750. Until you get back to the settings you have putted it before. and see whats the limit the model ai can sustain.
remote vortex
No problem. Which model are you using btw?
Im a little technologically inept so i think these my settings/model
frigid cape
Im a little technologically inept so i think these my settings/model
You are using the 32.000 Contest Token. I suggest you to try what i said before and see if you find "the breaking point."
remote vortex
You are using the 32.000 Contest Token. I suggest you to try what i said before ...
Sorry to have to ask but where/how can i change contact tocken?
rigid valve
Context size, he means context size
It'll be in your generation settings, which will have Temperature, amount of tokens, and Context size
frigid cape
frigid cape
Sorry to have to ask but where/how can i change contact tocken?
I apologies, i made a grammatic mistake. Don't be sorry for asking questions, we are here to help.
frigid cape
Sorry to have to ask but where/how can i change contact tocken?
It could also be an issue of the bot you created... Did you make a bot over 32.000 Tokens size? If so... you need to leave atleast 6000 to 4000 left.
zenith raven
Im a little technologically inept so i think these my settings/model
You're also tryna use pai-001-beta, Pai is no longer in beta.
zenith raven
You are using the 32.000 Contest Token. I suggest you to try what i said before ...
Note that the more data is processed in the input, the greater the chance that the model will lose coherence.
frigid cape
Note that the more data is processed in the input, the greater the chance that t...
I'm aware of the model ai limitations. I suggested the person to play with the settings, but if she is using the Pai model witg 32.000 Token size, then i suggested the bot should be putted to 32.000 in the context token settings.
The person can still make a bot that has 26.000 tokens of informations.
zenith raven
Mhum.
zenith raven
You're also tryna use `pai-001-beta`, Pai is no longer in beta.
Also change pai-001-beta for pai-001, maybe that's the problem.
remote vortex
You're also tryna use `pai-001-beta`, Pai is no longer in beta.
Oh? really where can i find the new model?
zenith raven
Oh? really where can i find the new model?
Since pai-001-beta, a lot has changed, there is also a new, much better model, cosmorp.
frigid cape
Oh? really where can i find the new model?
I suggest you cosmos RP if you use roleplay bots with under 4000 of Context Tokens usage, becouse the model is only 8000 Context token. It is free and it the best one out there for Free roleplay unlimited. It is a little bit slow right now, becouse of the model being used by many people.
If you do have bigger roleplay bots that has over 16000 Context token, i suggest you to use Pai-001 RP 32.000 Context Token.
remote vortex
I suggest you cosmos RP if you use roleplay bots with under 4000 of Context Toke...
Okay thank you I'm changing it now! do I put it in custom model or open ai?:
frigid cape
Okay thank you I'm changing it now! do I put it in custom model or open ai?:
Put the OpenAi Present. And Select GPT 3.5 or GPT 4.0o
remote vortex
Put the OpenAi Present. And Select GPT 3.5 or GPT 4.0o
Thank you so much! this is really helping
frigid cape
Which ever do you like, i'm not sure if it does anything but just select which one you feel that "it does something".
Your welcome, glad i could help. I'm positive that Since pointed out the issue and she did a great job pointing out the issue.
frigid cape
Thank you so much! this is really helping
Have fun and always reach out for help! We're here to help! π
remote vortex
Have fun and always reach out for help! We're here to help! π
Thank you thank you!
I jsut did all your suggestions and it's perfect now! Thank you so much! changing the modle the tokens and the temperature ugh chef's kiss