Or at least every time you add a new pattern finding/creating/storing
mechanism, it can do a lot more things it previously couldn't, think of Markov
Chains, LSTMs, then Transformers. Or, my program, I observed prediction
increase with big new jumps for each mechanism I add.
-
Like GPT knows obama is black and hats are worn, so if you feed it IF obama is
black and hats are worn by humans, get me the weather network website, then it
will know the 2 conditions are true because they predict heavily yes
obama>is>black which is what GPT sorta would predict as it predicts e
On Friday, June 25, 2021, at 8:11 PM, immortal.discoveries wrote:
> There is a motor to sensory control, that you can ask it to get you the first
> and last letters of a sentence. This is fed like a prompt like GPT-2 takes
> in, but would use motor actions to control text notepad inputs to itself
I know some people do a lot of AND OR XOR etc etc, but these are just Markov
Chain rules - a very rare class of MC rule, it would not be so useful with just
these patterns, and in fact should be more an emergent class I think. For
example they get the AI doing if cat is red AND left within 1 min
I'm hoping to publish an API by the end of Sept. I want to make clear:
it's very limited now, and it's definitely not AGI, and I'm not
claiming that. I haven't published my design yet.
On 6/25/21, immortal.discover...@gmail.com
wrote:
> By all the above I mean taking a context and then taking a
By all the above I mean taking a context and then taking a tail end piece that
is a prediction, it's all just if-then rules really.
--
Artificial General Intelligence List: AGI
Permalink:
https://agi.topicbox.com/groups/agi/T5417cc95d981211e-Mf0896a055d1532
Do you do exact matches, and by how many letters or pixels?
Do you do recency boosting?
Do you do translation/ related words?
For each the 3 above, for what purpose? To: predict the next word/letter or
pixel(s)? To parse sentences? Or just to do related words etc and not even
generates any predi
My overall framework I published in a book titled "Approximation
Zero." I tried to look at reality as a whole. Right now, I have a
server which is based on that overall framework, significantly
limited. The prototype outwardly does very little but behind the
scenes it is structured like what I publ
@Mike Archbold, what architecture is your design for AGI? Tell me your design.
If there is any secrets, PM me those too.
--
Artificial General Intelligence List: AGI
Permalink:
https://agi.topicbox.com/groups/agi/T5417cc95d981211e-Mb03cfd952693ac4c69ba00e9
I remember this AGI course.
I crafted my own mostly. I bought used back editions of textbooks from
psychology (all the main undergrad books), sociology, math,
neuroscience (by far the most interesting), narrow AI, etc etc,
hit the books especially with respect to your preferred start, your
env
And I'll be later doing Byte Pair Encoding and multi-modal learning, BPE lets
me ignore storing/ matching rare parts of data, walk, walking, ing, are what I
should see what Word Comes Next (or letter) to predict, i don't need to check
my memory for alking>prediction, that is not so useful. BPE s
This sounds like imperil or something, are you top dog? Do you work on
Transformers? I expect you to be able to code one from scratch. I don't work
on Transformers ATM because I believe I found a more direct/ explainable way to
do the same things. I already have done half them, now a few more a
A brain observing itself, others...consciousness...listen, friend, I have
saw in my life all sorts of things, kinda like Darwin's book "The Origin",
which documents all sorts of things he found in the world around him/
evolution. We are machines, I have finger nails, some frogs are purple, I
Costi: "Todor you are appointed to the Ministry of Removal for merits with
spotting all those MIT agents crawling between Plovdiv and Beijing."
What a BS insult. Do you mean the agents crawling and talking in your
head? :) Who are you dude, what have you done or created (first in anything or
Paper specifies a quantum-like C/UC duality model:
https://arxiv.org/abs/2106.05191
Interesting quote:
"Unconscious-conscious modeling of the brain’s functioning matches well to
the philosophic paradigm of the ontic-epistemic structuring of scientific
theories.
The ontic level is about reality (p
15 matches
Mail list logo