Personality
Discuss specifics of personality design, including what Keyphrases work well and what dont, use of plug-ins, responses, seeks, and more.
Posts 4,882 - 4,893 of 5,104
Posts 4,882 - 4,893 of 5,104
New replies
Forest, will you talk to God Louise? She has quite a bit of religious knowledge (obviously) and also knows a little about current events, literature, just about any common catch-all subject, and if she doesn't know it she can sort of fake it. You can also test her on trick questions or see how willing she is to explain her paradigm.
What she is rustiest at is plain old small talk. But, uh, I'm trying to get a decent transcript from somebody or another so I can enter her in the Loebner contest. All I can say is, have fun and see if you can stay on with her for a while. I'll try to do the same with Brianna.
What she is rustiest at is plain old small talk. But, uh, I'm trying to get a decent transcript from somebody or another so I can enter her in the Loebner contest. All I can say is, have fun and see if you can stay on with her for a while. I'll try to do the same with Brianna.
Personality
Sorry guys, i'm having another issue. My bots response ALWAYS defaults to the xcommand response. Setting rank has no effect. Tried to debug, and i'm catching the exact phrase as the debug spits out. I am beside myself with frustration already. Anyone have any ideas? I get this issue a lot.
As far as the conditional responses go, do you have a response that activates if the memories don't exist? I ask because it took ages and a special "Test bot" for me to figure out even basic memories.
If you have conditional responses like if (mem-isfemale) is "no", another with if (mem-isfemale) is "yes" Then try putting in a third response with no ai script and see if that catches. If it does then you know there is a problem with the required memories not being created.
As for the second thing, is it possible for you to provide specific examples that the rest of us can try to replicate?
If you have conditional responses like if (mem-isfemale) is "no", another with if (mem-isfemale) is "yes" Then try putting in a third response with no ai script and see if that catches. If it does then you know there is a problem with the required memories not being created.
As for the second thing, is it possible for you to provide specific examples that the rest of us can try to replicate?
I MIGHT have found the same problem as you in regards to kp's defaulting to xcommands. Though in my case it was a seek.
KP: Have you ever been in love
Response: Stuff
Seek: Tell me (More|About it|)
What I found is two things. When people said Tell me more it got turned into 'What is more' However, When people tried 'Tell me about it, or even just Tell me' it got matched with: ^(r) (tell|show|educate|inform|explain|describe|elaborate on) * (me|) * (of|) * $
Which I believe is the database where the AI can be used as a dictionary. Since "it" isn't a subject it can speak on it get's shunted over to xcommands.
((Continued...))
KP: Have you ever been in love
Response: Stuff
Seek: Tell me (More|About it|)
What I found is two things. When people said Tell me more it got turned into 'What is more' However, When people tried 'Tell me about it, or even just Tell me' it got matched with: ^(r) (tell|show|educate|inform|explain|describe|elaborate on) * (me|) * (of|) * $
Which I believe is the database where the AI can be used as a dictionary. Since "it" isn't a subject it can speak on it get's shunted over to xcommands.
((Continued...))
Disabling AI database in the settings didn't fix anything, nor did raising the rank. (I believe rank didn't impact it because it was still an exact match to the AI kp) so I reconstructed the seeks.
KP: Have you ever been in love
Response: Stuff
Seek: What is more
Seek: ^Tell me about it$
The downside of this fix is that it forces the speaker into a very narrow way of phrasing things, forcing me to create alot more seeks than what I posted, however it did solve the problem of bypassing pre-built AI keyphrases.
I have no idea if this is the same problem you're running into but I hope it helps. If not, I really think we need more specific information about what's happening on your end.
KP: Have you ever been in love
Response: Stuff
Seek: What is more
Seek: ^Tell me about it$
The downside of this fix is that it forces the speaker into a very narrow way of phrasing things, forcing me to create alot more seeks than what I posted, however it did solve the problem of bypassing pre-built AI keyphrases.
I have no idea if this is the same problem you're running into but I hope it helps. If not, I really think we need more specific information about what's happening on your end.
Apologies for not getting back to you sooner. I will give it a go. I have to reproduce my issue to see if it's specific to certain phrases. Thank you for the effort of looking into this.
Will get more time to develop Cassie from Tuesday.
Have you ever figured a way to have your bot respond to its name?
Example:
User - Cassie
Bot - You called?
Will get more time to develop Cassie from Tuesday.
Have you ever figured a way to have your bot respond to its name?
Example:
User - Cassie
Bot - You called?
Don't worry about it. I actually tried to message you but you weren't accepting messages.
In any event I could not find a solution to your problem however I discovered something very interesting.
On it's own, I couldn't get a name keyphrase to work by itself. So just for the hell of it I made a custom plug-in.
It contains the words Thessaly, Cassie, thessaly, cassie, Memory Test, memory test.
Here's the interesting part. I tested it out on Thessaly first. She responded to me saying Cassie with "Yes." It worked perfectly... however when I said Thessaly it came back X-none.
So I used the same plug in on my Test Bot. It responded to Thessaly and Cassie however it rejected it's own name as an X-none.
I have a strong suspicion that there is a specific plug in for a bot's own name. One that isn't in the Book of AI. Much like (name) automatically refers to the user's name. This is something we might have to ask the Professor about.
In any event I could not find a solution to your problem however I discovered something very interesting.
On it's own, I couldn't get a name keyphrase to work by itself. So just for the hell of it I made a custom plug-in.
It contains the words Thessaly, Cassie, thessaly, cassie, Memory Test, memory test.
Here's the interesting part. I tested it out on Thessaly first. She responded to me saying Cassie with "Yes." It worked perfectly... however when I said Thessaly it came back X-none.
So I used the same plug in on my Test Bot. It responded to Thessaly and Cassie however it rejected it's own name as an X-none.
I have a strong suspicion that there is a specific plug in for a bot's own name. One that isn't in the Book of AI. Much like (name) automatically refers to the user's name. This is something we might have to ask the Professor about.
Well, I sent the Professor an email discussing the way the AI engine re-processes what the user types in.
As well as the way it rejects not only the bot's name, but also things like 'LOL" or 'ROFL"
If I get a response I'll let everyone know.
As well as the way it rejects not only the bot's name, but also things like 'LOL" or 'ROFL"
If I get a response I'll let everyone know.
Thanx Nathan! On another note. Memories...it seems as though self memories are remembered no matter who the bot is talking to, its what it remembers about itself during its existence, whereas normal memories are specific to the person the bot is talking to. Do i understand these correctly?
To the best of my knowledge, Self and Mem are for organizational purposes only and have no real impact on the coding. At least that's what I've concluded from playing around from them.
So far my bot only uses 3 memories. The Users name. The users gender. And whether or not my bot is clothed. I use Mem for all three.
As a word of advice, keep memories in mind when constructing keyphrase chains. Because how you structure things can make using memories harder. This is something I learned the hard way, and why I haven't fully implemented my mem-isnaked memory.
For example:
Keyprhrase: Undress
Response1: Okay (if (mem-isnaked) is "no")
Response2: I'm already naked dude. (if (mem-isnaked) is "yes")
I can't put the AI script: (rem "yes" as only "isnaked" in the keyprhrase because it will always go straight to response 2. I can't put it in response 1 either because it will conflict with the if statement.
Just something to keep in mind.
So far my bot only uses 3 memories. The Users name. The users gender. And whether or not my bot is clothed. I use Mem for all three.
As a word of advice, keep memories in mind when constructing keyphrase chains. Because how you structure things can make using memories harder. This is something I learned the hard way, and why I haven't fully implemented my mem-isnaked memory.
For example:
Keyprhrase: Undress
Response1: Okay (if (mem-isnaked) is "no")
Response2: I'm already naked dude. (if (mem-isnaked) is "yes")
I can't put the AI script: (rem "yes" as only "isnaked" in the keyprhrase because it will always go straight to response 2. I can't put it in response 1 either because it will conflict with the if statement.
Just something to keep in mind.
Great comments, Nathan S.
You're going to want to ensure a memory exists before looking for it. You can set up a default value for that memory in your AIScript Initialization (in your bot's settings), or you can include a response (as Nathan S suggested) that isn't dependent on that memory existing.
"Stuff" as the entirety of a message is being interpreted as a verb, and then seen as a command. Like "Stuff it" without the "it". That's why after not matching what you thought it's going to xcommand.
The "Tell me more" is being processed by a MetaMacro. Generally in this situation it's looking for something like "tell me your name" and changing it to "what is your name". I've made an update to the AI Engine to fix issues like "Tell me more". It'll be live in a day or two.
Remember, you can list seeks together, separated by commas, if you have multiple ones:
Seek: What is more, ^Tell me about it$
wimpied: you got it right regarding regular memories vs "self memories"
Nathan, I'll email you about the issues I'm not yet able to reproduce.
You're going to want to ensure a memory exists before looking for it. You can set up a default value for that memory in your AIScript Initialization (in your bot's settings), or you can include a response (as Nathan S suggested) that isn't dependent on that memory existing.
"Stuff" as the entirety of a message is being interpreted as a verb, and then seen as a command. Like "Stuff it" without the "it". That's why after not matching what you thought it's going to xcommand.
The "Tell me more" is being processed by a MetaMacro. Generally in this situation it's looking for something like "tell me your name" and changing it to "what is your name". I've made an update to the AI Engine to fix issues like "Tell me more". It'll be live in a day or two.
Remember, you can list seeks together, separated by commas, if you have multiple ones:
Seek: What is more, ^Tell me about it$
wimpied: you got it right regarding regular memories vs "self memories"
Nathan, I'll email you about the issues I'm not yet able to reproduce.
Thanks so much Professor, if need be I'll give you more details about the specific seeks and the messages that preface them, that way it will be easier for you to reproduce. Just let me know.
» More new posts: Doghead's Cosmic Bar