[LOCKSUIT]

This is most of all my notes, images, algorithms etc summarized/unified in their recent forms. Here. I am 24 and started at 18 all this work. I make discoveries in my brain using mostly vision (visual language; shapes, cats, etc which has context, each explain each other like in a dictionary, a small world network of friendly connections).



I have 2 more videos to share but not yet uploaded. And 2 more notes. The last note will have some more very recent good data but these not yet given are of less immediate importance. The long file though does have a lot of recent knowledge in it though still. It's better you know all it, at least the movie.

notes
https://paste.ee/p/mcnEk
https://paste.ee/p/kQLCx
https://paste.ee/p/CvSsB
https://paste.ee/p/lJYMP
https://paste.ee/p/EmCZt

Code-only of my advanced-ngram 'gpt2':
https://paste.ee/p/7DG3M
https://paste.ee/p/XvVp5
result:
The software was made on a
The software was made on a wide variety of devices, and operating apps and applications that users can easily read as an app for android. It is a bit of a difference, but i was able to get it. The developers are not going to make it through a web applications, and devices i have seen in the running for the mobile apps. Applications allows users to access applications development tools, and allow applications of the app store. A multimedia entertainment entertainment device, and allows platforms enabled access to hardware interfaces. Using a bit of html application app developers can enable users to access applications to investors, and provide a more thorough and use of development. The other a little entertainment media, and user development systems integration technology. Applications allows users to automatically provide access to modify, optimize capability allows users to easily enable. Both users and software systems, solutions allowing owners software solutions solutions to integrate widgets customers a day. And if you are accessing services product, and mobile applications remotely access to the software companies can easily automate application access to hardware devices hardware systems creators and technologies. Builders and developers are able to access the desktop applications, allowing users access allows users to
((I checked the 400MB, not too long copies pastes, like only 3-5 words at most))


I almost got GPT-2 understood as shown at end of movie but need help, anyone understand it's inner workings? Looking to collaborate.

I recommend you do this as well and mentor each other.

more data by my top-of-mind pick:
AGI is an intelligent Turing Tape. It has an internal memory tape and an external memory tape - the notepad, the desktop, the internet. Like a Turing Tape it decides where to look/pay attention to, what state to be in now, and what to write, based on what it reads and what state it is in. The what/where brain paths. It will internally translate and change state by staying still or move forwards/backwards in spacetime. It'll decide if to look to external desktop, and where to look - notepad? where on notepad? internet? where on internet?

It's given Big Diverse Data and is trying to remove Big Diverse Data (Dropout/Death) so it can compress the network to lower Cost/Error and hence learn the general facets/patterns of the universe exponentially better while still can re-generate missing data despite having a small world network (all quantinized dictionary words explain each other). It uses Backpropagation to adjust the weights so the input will activate the correct node at the end. That node, can be activated by a few different sorts of images - side view of cat, front view of paw, cat ear, mountain, it's a multi dimensional representation space, Since the network Learns patterns (look up word2vec/Glove, it's same as seq2seq) by lowering error cost by Self-Attention evolution/self-recursion of data augmentation (self-imitation in your brain using quantinized visual features/nodes), it therefore doesn't modify its structure by adjusting existing connections (using weights/strengths) to remove nodes, it rather adjusts its structure by adjusting connections weights to remove error and ignores node count Cost.

Intelligence is defined as being flexible/general using little data, but walker bots are only able to solve what is in front of themselves, we need a thinker like GPT-2, and you can see the mind has evolved to simulate/forecast/predict the future using evolutionary mental RL self-imitation self-recursion of data. And intelligence is for survival, immortality, trying to find food and breed to sustain life systems, it's just an infection/evolution of matter/energy / data evolution.

[HS]

Oh my gosh, you're so logical, I did not expect that. Use your power for good. 

[HS]

You showed the optical illusions I posted! And talked about redundunduncy
Crazy text you is real life me, and real life you is chilled out Richard Feynman... This was too much I need to sleep...

[LOCKSUIT]

I think you can skip the heterarchy maybe....simply the hierarchy nodes get activated ex. nodes cat, etc, which parallelly leaks energy to their nearby context ex. 'the cat ate' 'the cat ran' 'our cat went' and these handles leak energy to nearby context 'the dog ate' 'the dog ran' 'some dog went' and so on for proving 'some'='our' as well, including if cat=zebra/horse and dog=zebra/horse then cat=dog! Hence no w2V, just on the fly activated by leaking connections. Solves typos, rearranged phrases, unknown words ex. superphobiascience, alternative words, related words, names, references ex. it/he, and blanks. Then for the candidate words, the winner is the one that is most frequent in knowledge (has energy) and in Working Memory Activation Context (which fade energy/leak), most related to story word (activation leak), most favorite (has energy). This is how to recognize/understand a window where you look/how wide, then which candidate Next Word to choose, then you may also adapt it by translating it too.

It can be run faster to do it other ways but to understand it can be easier by other ways like this.

[Art]

Finally we get to hear the real Lock!! Good one!

[LOCKSUIT]

The hierarchy can self-organize to lower node count Cost (error) by re-arranging the connections that exist, to learn Byte Pair Encoding (segmentation) on the fly too, not just translation or sequence-building on the fly. You don't have to look at all areas/layers of the hierarchy.

[LOCKSUIT]

Attention types: Decides which sense to pay attention to ex. buzzing noising or pop up goals, where/how wide to window on that sense, and which Next Word to predict using more attention.

[LOCKSUIT]

Think if you had a GPT-2 prompt 1 word long only ex. scratching, it could be seen as scratch ing, so let's say we had only scratch,  the word to predict next is the one seen before in the data, and the frequent one is more likely chosen. This word scratch can also match related words itch scrap etc and then see what frequent word is next. So the closest related word, and the highest frequent word is chosen.

scratch the x64
scratch it x58
scratch back x38
itch my x76
itch shoulder x50

So my is predicted because itch=scratch a lot and my is seen after itch x76 times. But we aren't done yet. The candidate words/tokens are given an extra score, now, which is relational score to story words 'scratch' (no itch exist) in your input prompt. So we end up picking shoulder say. We may also then translate shoulder to adapt it in ex. stomach if your story was about stomach itch.

If we have 2 words in our generated story so far: 'Scratch stomach', we can predict the next word using either word but we want to use both, so we see what nodes/related nodes including positional rearranging exists in the data ex. scratch stomach, itch bladder, bladder itch, etc, and the more distant they are positioned the lesser of course a match it is.

So say we got a 'first I go into the lake and stay there, second I cold freeze in the winter', the 'first' votes on 'second' because it could be 1st or etc translated, frequency, relation, but it is far back and there's other words, so it definitely has less vote being 1/10th the weight and fading in energy by this time, yet it does have a strong vote too.

Say we find 'stomach itch but it has a word in the middle or one missing ex. 'stomach really itch' or 'stomach', this also has less score but still has some.

What if I say 'second, this is that, third, i am this, but i won't say the next because i refuse to repeat myself'. Here I pay attention to ignore the word 'fourth' being said.

[LOCKSUIT]

Let me know if you want more, if no one says so then I assume no one is interested.

[goaty]

If your sure this is the future of a.i. you should just pursue it further yourself,   if you were just to release it all early for complete free, whats new about it theoretically that's exclusive to your system?

[LOCKSUIT]

It is free. I am sharing my knowledge/current stage. There's a lot of new knowledge there. My AGI description talks on its own to itself and asks us/tells us knowledge. It unsupervisedingly collects internet data and researches, and learns patterns/building blocks, then it Answers Questions using RL using 4 steps. It is transforming the data, old to new, then analyzing it again, old to new, this is bootstrapped Self-DataRecursion, data evolution, this is how evolution happens. And bigger brains are smarter by more context data. The tools/skills are another part for that and can be described by text/visual language of the universe. I also tell you what will happen in the end etc.

[LOCKSUIT]

Storing, Forgetting, and Recall. Attention does all. Which sense or where to pay attention to. How much attention. To recognize it. To choose the Next Feature by looking at story words and which to ignore (forget). To Adjust the Next Feature. The nodes that store/recall are attentive / pop up as agenda questions.

[Korrelan]

Korrelan, I'm really sorry you won't be able to fully process the green/red colors being color blind but, it doesn't matter much lol.

Thanks for the individual consideration Lock, it's much appreciated, but it's ok because I can read.

You did leave that comment in though, before you posted it on every single forum known to man. 

 

ED: OH! I see you have removed it from one non Email based, editable forum... cool.

 

[goaty]

Looks like you've got a good system (I didn't think of it myself, sounds good technique),  but is it intelligent yet?  If not your work isn't finished,  just somewhere along.

[LOCKSUIT]

I spammed OpenAi with it too hehe, more 2 come

I call this ignition phase, spread it like fire