[021] AI Bots – ChatGPT, DALL-E 2 [Deleted]

I don't know if it's possible with the addon, due to you being able to set commands prompts.

But, it is possible inside ChatGPT
They’ve convinced it to reply in “Developer Mode” and now it is a complete savage, making completely inappropriate jokes nonstop. Not sure if that’s good or bad 😂
 
One other thing.

I read that this only works with 3.5, but as I was going through the settings I seen that 4 and 4-32k was available. Does that setting matter?

I seen there is a waitlist for gpt4, so do I just flip the switch in the settings when and if I get accepted?
 
Just upgraded to the latest XF version, and now my AI bots don't work.
  • title corrector addon works
  • reply assistant addon works
  • can't test whether the spam one is working or not
AI bots, not working. I have disabled/reenabled, saved all permissions again, changed the API key, nothing seems to work, which leads me to something changed in the update that affects AI bots from posting.
 
Just upgraded to the latest XF version, and now my AI bots don't work.
  • title corrector addon works
  • reply assistant addon works
  • can't test whether the spam one is working or not
AI bots, not working. I have disabled/reenabled, saved all permissions again, changed the API key, nothing seems to work, which leads me to something changed in the update that affects AI bots from posting.
Oh, those patches from XenForo that suddenly break something...
Do you have any errors in the control panel? Perhaps problems with other add-ons?
 
No errors. Just tested it with the default style, nothing, so not a style problem either. I will try addons and disable all but yours.
 
No errors. Just tested it with the default style, nothing, so not a style problem either. I will try addons and disable all but yours.
I mean, you may have noticed problems with some other add-ons after the update - this information will help me get closer to identifying the source of the problem
 
Ok... a little searching, enable that bad addon and then it has one option to disable legacy blah blah, and that fixed it. It was supporting older software versions of their own stuff, which was breaking things.
 
Are there instructions that can be put into the prompt to make it probabilistic that a bot will post a response to a thread? I asked ChatGPT about this and it told me I should include a system message asking ChatGPT to not generate text.

I placed the following instructions in my prompt in the add-on and it didn't work, i.e., the bot generated a response. The bot's "restrictions" and "triggers" settings were set such that the bot should respond. I'm just trying to code the prompt so that the bot won't respond some portion of the time, despite these settings.

----
[System] Please do not generate any text for this prompt.

[My regular prompt instructions]
----

Ultimately, I want the first line of the prompt to be something like "[System] Please do not generate any text for this prompt 50% of the time" or whatever works... But if the above test prompt doesn't work (i.e., generates a response to my regular prompt instructions despite being told not to) I know I won't be able to get a probabilistic version to work.

Does anyone know a way to program a bot to probabilistically respond/not respond in the add-on? Thank you!
 
Top Bottom