[agi] Kraken API will make me RICH (Scalping Bot)

殺*Kraken API obeys my commands!! *  殺 (https://www.youtube.com/watch?v=4apATiVWCgI ) -- Artificial General Intelligence List: AGI Permalink:

Re: [agi] Re: Google engineer claims AI bot LaMDA became 'sentient'

This article claims China has built a 174 trillion parameter artificial brain (174 trillion is larger than the number of synapses in the human cortex -- https://eurasiantimes.com/aping-a-human-brain-chinese-supercomputer-achieves-breakthrough-in-ai/ . If that is true it is impressive, at least

[agi] Re: Is there a file format purely for Huffman encoding of text? I need that right now.

Solution 4: Extend the symbol range to 9 bits (thus wasting up to 32 bytes in the tree) so we can also encode control symbols (between 0x100 and 0x1FF). One of those, e.g. 0x100, would be a NOP ("no output") control symbol that you can use to fill up bytes. An incomplete byte following a NOP is

[agi] Re: Is there a file format purely for Huffman encoding of text? I need that right now.

Ah there is one little thing I missed... *How do you know the text has ended?* It is encoded as a bit stream, so it might end in the middle of a byte. Solution 1: Insert text length somewhere. Kinda shit though, prohibits streaming. Solution 2: Put a special END symbol in the tree. But we have

Re: [agi] The next advance over transformer models

On Thursday, June 30, 2022, at 10:15 AM, Rob Freeman wrote: > But in the sense of having the same internal connectivity within two groups > which are not directly connected together. Yes, in the sense that inputs (input clusters) are parameterized with derivatives ("connections") from

Re: [agi] The next advance over transformer models

On Thu, Jun 30, 2022 at 2:18 PM Boris Kazachenko wrote: > On Thursday, June 30, 2022, at 6:10 AM, Rob Freeman wrote: > > what method do you use to do the "connectivity clustering" over it? > > > I design from the scratch, that's the only way to conceptual integrity in > the algorithm:

Re: [agi] The next advance over transformer models

On Thursday, June 30, 2022, at 6:10 AM, Rob Freeman wrote: > what method do you use to do the "connectivity clustering" over it?  I design from the scratch, that's the only way to conceptual integrity in the algorithm: http://www.cognitivealgorithm.info. Couldn't find any existing method that's

Re: [agi] The next advance over transformer models

I think "prediction" is a redundant term,  any representation is some kind of prediction.  "Shared property": I meant initially shared between two compared representations, and only later aggregated into higher-level shared property, within a cluster defined by one-to-one matches.  Vs. summing

Re: [agi] The next advance over transformer models

On Thu, Jun 30, 2022 at 1:51 PM Rob Freeman wrote: > On Thu, Jun 30, 2022 at 1:33 PM Boris Kazachenko > wrote: > >> ... >> My alternative is to directly search for shared properties: lateral >> cross-comparison and connectivity clustering. >> By the way, independently of what shared properties

Re: [agi] The next advance over transformer models

On Thu, Jun 30, 2022 at 1:33 PM Boris Kazachenko wrote: > On Thursday, June 30, 2022, at 3:00 AM, Rob Freeman wrote: > > I'm interested to hear what other mechanisms people might come up with to > replace back-prop, and do this on the fly.. > > > For shared predictions, I don't see much of an

Re: [agi] The next advance over transformer models

On Thursday, June 30, 2022, at 3:00 AM, Rob Freeman wrote: > I'm interested to hear what other mechanisms people might come up with to > replace back-prop, and do this on the fly.. For shared predictions, I don't see much of an alternative to backprop, it would have to be feedback-driven

Re: [agi] The next advance over transformer models

On Thu, Jun 30, 2022 at 10:40 AM Ben Goertzel wrote: > "what method could directly group hierarchies of elements in language > which share predictions?" > > First gut reaction is, some form of evolutionary learning where the > genomes are element-groups > > Thinking in terms of NN-ish. models,

Re: [agi] The next advance over transformer models

On Thursday, June 30, 2022, at 1:47 AM, Rob Freeman wrote: > "ham and eggs", not "eggs and ham", that kind of thing, or "strong tea" not > "powerful tea" "I'm havving somme powerfulll tea right nowww!!" :DD not me but that someone that likes coffee --

Re: [agi] The next advance over transformer models

"what method could directly group hierarchies of elements in language which share predictions?" First gut reaction is, some form of evolutionary learning where the genomes are element-groups Thinking in terms of NN-ish. models, this might mean some Neural Darwinism type approach for evolving the