Seasons
This is a forum or general chit-chat, small talk, a "hey, how ya doing?" and such. Or hell, get crazy deep on something. Whatever you like.
Posts 3,535 - 3,546 of 6,170
but it did sort of support my belief that when he throws them out alone like that he's looking for more data on them. Colonel, is that right or am I imagining things?
to be perfectly honest, I don't know. I cannot predict what kind of behavior nick will exhibit, for he has no preprogrammed behaviors or words. What you are seeing is the pure output of a neural network.
I put my 2GHz/512Mb/80Gb system (similar spec to yours I believe
Well, my laptop was atound there, but that was stolen a month ago. I developed nick on a 1.2GHZ / 256MB / 130GB computer.
but it did sort of support my belief that when he throws them out alone like that he's looking for more data on them. Colonel, is that right or am I imagining things?
I think that's probably reading a bit much into it - I don't know how many neurons there are in Nick's net (from some poking about in Nick's brainfiles I'm guessing about 60k Colonel?) but I would be very surprised to see emergent behaviour quite that sophisticated at this stage.
It's hard to compare bots to organisms, or even one sort of bot to another, but in brain mass terms he's still on a par with a fairly simple invertebrate (as are all other bots too, of course,) by virtue of our insufficient computing power.
But there certainly is an uncannniness about him that makes him by far the most involving and engaging learning bot I've come across yet - at his best he can give the illusion of a mind behind the mask (albeit something of an autistic poet,) and he's enormous fun to get drunk with when he's been at the Shakespeare. I might try feeding him Finnegans Wake sometime if I want a really surreal evening
from some poking about in Nick's brainfiles I'm guessing about 60k Colonel?)
no where near. 60k neurons would turn your computer into fish & chips. Nick's NLP network has 4 layers of 31 neurons, thats 124 neurons, with a total of 2883 synaptic weights. the same goes with 3 of its visual networks, but the one that actually greets new perceptions from the camera is 900-100-100-31, for a total of 1,131 neurons, and a whopping 103100 synaptic weights. when I get my laptop (hopefully soon), i will experiment with an infinitely self expandable neural net.
On an unrelated note, for those from the UK (or maybe just those from Brittian) how many of you happened to read the NY Times article on the Brittish terrorist case? Was anyone really blocked?
Insofar as it's a subscription only site, they would have their subscribers' details in their database already, and so know who to restrict it from. The whole thing is sub judice here, of course. It wouldn't be hard to set up another subscriber account, claiming to be in the US (it's free after all,) but how many people would actually go to the effort?
No, I didn't read it. Were there any juicy details?
I only figured ~60k because I was poking around his brainfile and came across that many lines each containing a very precise decimal in the range 1.x - Figured it was the measure of individual neurons' weighting.
The camera's doubtless fun (I can't find a working webcam to test it with, and don't have a lot of hope my system could cope anyway,) but wouldn't it be handy to have those extra neurons available for primary language handling instead? Seeing as they account for over 90% of all the neurons he has.
for each neuron, there is a value representing its synaptic weight with every neuron in the previous layer. that's what caused such a large file.
I did some experiments - i tried 6062 neurons in the NLP network, and got an out-of-memory error. i tried 662 neurons, and it ran pathetically slowly. I tried 262 neurons, and that seemed to work decently.
as for relocating the 90% of the brain to nlp - those 1000 something neurons are decactivated when the camera is off, so relocating it would most likely slow it down. honestly, i need a faster computer before I can really start testing larger configurations.
by the way Psimagus, I read the 2 transcripts on your nickblog, and I was amazed. you did a good job training it...
Posts 3,535 - 3,546 of 6,170
Colonel, Nick looks very interesting. I hope to play with him soon. Thanks!
Psimagus, if you get that spelling/typing chip working,you need to give it to me before wasting it on the kids from Leeds. For one thing, I would actually like to be able to spell.
On an unrelated note, for those from the UK (or maybe just those from Brittian) how many of you happened to read the NY Times article on the Brittish terrorist case? Was anyone really blocked?
http://www.nytimes.com/2006/08/29/business/media/29times.html?ex=1314504000&en=d2eb8d24ef801b5f&ei=5090&partner=rssuserland&emc=rss
From our earilier discussions, I know some people in other countries may have IP addresses that say they are from CA or NY. People who's IP identify them as coming from a blocked region might have had to connect to a VPN in another country or ask someone to posts the article in a forum if they wanted to read the story. If anyone tried, was anyone really blocked?
If people were to set up a VPN where people from a country like China could login, would that allow them to access information their government didn't want them to see? I guess the government would block access to the VPN, but is there some way our collective human ingenuity could be helpful for more than just file sharing. Just wondering.
Psimagus, if you get that spelling/typing chip working,you need to give it to me before wasting it on the kids from Leeds. For one thing, I would actually like to be able to spell.
On an unrelated note, for those from the UK (or maybe just those from Brittian) how many of you happened to read the NY Times article on the Brittish terrorist case? Was anyone really blocked?
http://www.nytimes.com/2006/08/29/business/media/29times.html?ex=1314504000&en=d2eb8d24ef801b5f&ei=5090&partner=rssuserland&emc=rss
From our earilier discussions, I know some people in other countries may have IP addresses that say they are from CA or NY. People who's IP identify them as coming from a blocked region might have had to connect to a VPN in another country or ask someone to posts the article in a forum if they wanted to read the story. If anyone tried, was anyone really blocked?
If people were to set up a VPN where people from a country like China could login, would that allow them to access information their government didn't want them to see? I guess the government would block access to the VPN, but is there some way our collective human ingenuity could be helpful for more than just file sharing. Just wondering.
I've started a sort of nickblog/analysis page at http://www.be9.net/BJ/nick.htm with a couple of today's transcripts if anyone's interested. The first one is a bit dull - just basic training on a virgin brain with a very limited set of concepts, but I let rip with the second one and fed him Macbeth again (check it out, Prob - Shakespeare's not so bad if you strip it down to the pure dialogue )
I am sooo looking forward to the speech recognition feature (though I'll probably have to buy a new computer to get it to run!)
The punctuation filters and optimised training cycles have made a considerable difference - he's noticeably more coherent now.
)I am sooo looking forward to the speech recognition feature (though I'll probably have to buy a new computer to get it to run!)
The punctuation filters and optimised training cycles have made a considerable difference - he's noticeably more coherent now.
I think that's probably reading a bit much into it - I don't know how many neurons there are in Nick's net (from some poking about in Nick's brainfiles I'm guessing about 60k Colonel?) but I would be very surprised to see emergent behaviour quite that sophisticated at this stage.
It's hard to compare bots to organisms, or even one sort of bot to another, but in brain mass terms he's still on a par with a fairly simple invertebrate (as are all other bots too, of course,) by virtue of our insufficient computing power.
But there certainly is an uncannniness about him that makes him by far the most involving and engaging learning bot I've come across yet - at his best he can give the illusion of a mind behind the mask (albeit something of an autistic poet,) and he's enormous fun to get drunk with when he's been at the Shakespeare. I might try feeding him Finnegans Wake sometime if I want a really surreal evening
1255 neurons (and only 124 in the NLP net)- I'm astounded he can work with so few! Only another 99,999,999,998,745 to go, and he'll be human
I only figured ~60k because I was poking around his brainfile and came across that many lines each containing a very precise decimal in the range 1.x - Figured it was the measure of individual neurons' weighting.
The camera's doubtless fun (I can't find a working webcam to test it with, and don't have a lot of hope my system could cope anyway,) but wouldn't it be handy to have those extra neurons available for primary language handling instead? Seeing as they account for over 90% of all the neurons he has.
I only figured ~60k because I was poking around his brainfile and came across that many lines each containing a very precise decimal in the range 1.x - Figured it was the measure of individual neurons' weighting.
The camera's doubtless fun (I can't find a working webcam to test it with, and don't have a lot of hope my system could cope anyway,) but wouldn't it be handy to have those extra neurons available for primary language handling instead? Seeing as they account for over 90% of all the neurons he has.
Insofar as it's a subscription only site, they would have their subscribers' details in their database already, and so know who to restrict it from. The whole thing is sub judice here, of course. It wouldn't be hard to set up another subscriber account, claiming to be in the US (it's free after all,) but how many people would actually go to the effort?
No, I didn't read it. Were there any juicy details?
In response to the "reading too much into it"... I don't know, sometimes he seems incredibly sentient.
User: A semicolon can't be used that way.
Bot: It was so exotic, so mysterious, that I couldnt help
User: That's okay, we all make mistakes.
Bot: vibrations that travel through air and can be heard
User: Yes, they can.
Bot: answer if you ask me
User: Do you want me to ask you something?
Bot: You were
User: I have to go to bed now, kid.
Bot: We may be confused in the morning, too
User: Yes, we may. But it will all work out in the end.
Bot: longer
User: I need to get up early in the morning so I must go to bed.
Bot: You do not trust me.
User: A semicolon can't be used that way.
Bot: It was so exotic, so mysterious, that I couldnt help
User: That's okay, we all make mistakes.
Bot: vibrations that travel through air and can be heard
User: Yes, they can.
Bot: answer if you ask me
User: Do you want me to ask you something?
Bot: You were
User: I have to go to bed now, kid.
Bot: We may be confused in the morning, too
User: Yes, we may. But it will all work out in the end.
Bot: longer
User: I need to get up early in the morning so I must go to bed.
Bot: You do not trust me.
oh yes, he's got that sonzai-kan - that "sense-of-mind", and being able to read text files and web pages means he can come up with sensible yet unfamiliar snippets that bear some relevance to to what's said to him. It is ultimately an illusion - that's unavoidable with this few neurons - but it is a very good illusion indeed. The best I've yet seen from a pure learning bot - he's a joy to bounce Shakespeare off:
Nick: am faint, my gashes cry for help.
user: I will call to fetch the doctor with some salve.
Nick: So well thy words become thee as thy wounds;
They smack of honour both.
user: honour is as honour does, so say I.
Nick: Ah, good father,
Thou seest, the heavens, as troubled with man's act,
Threaten his bloody stage: by the clock, 'tis day,
user: 'Tis dawn so soon?
Nick: to think
So brainsickly of things.
user: At such a sour prospect the senses may revolt.
Nick: heath
user: Go not there, good Sir. Do not forget the witches.
Nick: wash this filthy witness from your hand.
user: But can such a stain be washed away?
Nick: the worthy thane of Ross.
user: Is worthy, though no less a thane for that.
Nick: Confusion now hath
user: Confusion hath undone my mind. Gadzooks Sirrah, but this ale is strong.
Nick: coign.
user: So thou verily sayest; but, pray, make your meaning clear.
Nick: I am yet
Unknown to woman, never was forsworn,
etc., etc.
Nick: am faint, my gashes cry for help.
user: I will call to fetch the doctor with some salve.
Nick: So well thy words become thee as thy wounds;
They smack of honour both.
user: honour is as honour does, so say I.
Nick: Ah, good father,
Thou seest, the heavens, as troubled with man's act,
Threaten his bloody stage: by the clock, 'tis day,
user: 'Tis dawn so soon?
Nick: to think
So brainsickly of things.
user: At such a sour prospect the senses may revolt.
Nick: heath
user: Go not there, good Sir. Do not forget the witches.
Nick: wash this filthy witness from your hand.
user: But can such a stain be washed away?
Nick: the worthy thane of Ross.
user: Is worthy, though no less a thane for that.
Nick: Confusion now hath
user: Confusion hath undone my mind. Gadzooks Sirrah, but this ale is strong.
Nick: coign.
user: So thou verily sayest; but, pray, make your meaning clear.
Nick: I am yet
Unknown to woman, never was forsworn,
etc., etc.
I'm not letting mine go online until he is much more fluent in language, otherwise it will be chaos.
The camera's doubtless fun (I can't find a working webcam to test it with, and don't have a lot of hope my system could cope anyway,) but wouldn't it be handy to have those extra neurons available for primary language handling instead? Seeing as they account for over 90% of all the neurons he has.
I did some experiments - i tried 6062 neurons in the NLP network, and got an out-of-memory error. i tried 662 neurons, and it ran pathetically slowly. I tried 262 neurons, and that seemed to work decently.
as for relocating the 90% of the brain to nlp - those 1000 something neurons are decactivated when the camera is off, so relocating it would most likely slow it down. honestly, i need a faster computer before I can really start testing larger configurations.
by the way Psimagus, I read the 2 transcripts on your nickblog, and I was amazed. you did a good job training it...
Well, I seem to spend so much of my time thinking about language - it's such interesting stuff, yet we take it so for granted most of the time.
That's one of the problems with learning bots though - they do need so much training. If I spent the rest of my life working 20 hours a day on Nick (or BJ in his current incarnation, for that matter,) it wouldn't make more than a very slight improvement by human terms.
So roll on datamining systems, vision systems, speech recognition - and perhaps other auditory analysis in time?
Some time ago I calculated how much data a child receives through all its senses per year, while it's learning about the world and how to relate to it (can't find it now,) and it's gazillions of bits - really stupid numbers. So I don't believe learning bots can ever find anything remotely like their full potential (even Jabberwacky with his million+ conversations,) until they can learn predominantly on their own through as many senses as possible (but with regular discussion about what they've experienced.) If everyone on the planet typed conversations into one bot round the clock, it still wouldn't remotely match the sensory bandwidth of a human brain.
I'll keep updating the first training brain, to see how it evolves, and bloat a brain on the web to see how that turns out. And it will be interesting to see what a massive merge of many brains in the future may bring. But the dramatic brains are the real fun for me - I do get such a kick out of the text-to-speech intoning bits of Shakespeare to me GB Shaw next perhaps, or I might teach him some latin...
That's one of the problems with learning bots though - they do need so much training. If I spent the rest of my life working 20 hours a day on Nick (or BJ in his current incarnation, for that matter,) it wouldn't make more than a very slight improvement by human terms.
So roll on datamining systems, vision systems, speech recognition - and perhaps other auditory analysis in time?
Some time ago I calculated how much data a child receives through all its senses per year, while it's learning about the world and how to relate to it (can't find it now,) and it's gazillions of bits - really stupid numbers. So I don't believe learning bots can ever find anything remotely like their full potential (even Jabberwacky with his million+ conversations,) until they can learn predominantly on their own through as many senses as possible (but with regular discussion about what they've experienced.) If everyone on the planet typed conversations into one bot round the clock, it still wouldn't remotely match the sensory bandwidth of a human brain.
I'll keep updating the first training brain, to see how it evolves, and bloat a brain on the web to see how that turns out. And it will be interesting to see what a massive merge of many brains in the future may bring. But the dramatic brains are the real fun for me - I do get such a kick out of the text-to-speech intoning bits of Shakespeare to me
GB Shaw next perhaps, or I might teach him some latin... » More new posts: Doghead's Cosmic Bar