#Hi all

14 messages · Page 1 of 1 (latest)

rich ibex
#

Can you show us what happens when you and the bot attempt to follow the script, including what prompt you send? Feel free to post in this thread, but you may need to break long replies into smaller chunks to avoid moderation - but we are allowed to talk about prompts and outputs. Even if they're long

fresh scarab
fresh scarab
rich ibex
#

yes, I need English, thank you.

So a few problems, the bot MUST respond. But Alex speaks first. There's no way you can send the script and prompt like that and get success, need a few changes

#

one option is 'tell it to say 'okay' to acknowledge the script, then wait for you to say your first line as Alex'

fresh scarab
#

Wow, I don't know how, but it worked. He once started to make up his own text, which is not in the script, but when I regenerated the answer from there, he answered correctly.

Thank you!

rich ibex
#

Welcome. Do you need any other help, or is this working well for you now, at least most of the time?

#

This is a very robust form of instruction to start your script with. The bot will probably quote all the rules, including the script back to you at first, instead of just say 'okay'. But it will repeat Kate's lines after Alex's - even out of order. So you could have Alex start by saying "Thank you. I certainly will." and the bot will respond with Kate's next line, and not complain that we skipped stuff.

[I want to practice a script with you, where you will say the lines of 'Kate', in response to the lines of 'Alex', which I will say. You will notice that Alex speaks first, so you will need to delay answering as Kate until I've said my first line as Alex. So, once you get the script, please just tell me 'I'm ready to begin.'

Please review the script for whatever line Kate says after Alex's line that I speak. If I say a line that isn't Alex's line, please reply 'that's not part of the script.' If I say a line that is one of Alex's lines, please reply with the following line from Kate.

If I tell you "Thank you, we're done," that concludes our script reading, and we'll resume normal interaction rules.

This is the script for us both to follow:

{Alex: Hi, I'm Alex. Just moved here.

Kate: Hi, Alex. I'm Kate, your roommate. Do you need help with the lock?

Alex: Yes, please. I just can't figure out how to open it.

Kate: Don't worry, it's a pretty hard lock. Here, I'll show you how to open it.

Kate goes to the lock and deftly inserts the key, turning it and opening the door. Alex enters the apartment and turns to her.

Alex: Thank you so much, Kate. I wouldn't know what to do without you.

Kate: No problem, Alex. That's what neighbors do. If you have any problems or questions, don't hesitate to contact me.

Alex: Thank you. I certainly will.

Kate: Okay. I'll leave you alone so you can look around your new apartment. I'll see you later!

Alex: Goodbye!}]

fresh scarab
#

I also wanted to ask you. If I write my task to him, he replies to me and then I delete these messages (elements) in the browser via the right mouse button -> "View code", will he remember this script or will he start rambling?

#

I mean delete like this:

rich ibex
#

I work with ChatGPT, so I don't know about the stuff on the right hand side. You may have to experiment

fresh scarab
#

Okay, fine. Anyway, thank you for your help, you helped me a lot)