The original thread can be found here: https://trashchan.xyz/robowaifu/thread/595.html --- Welcome to the Lemon Cookie thread, The goal of Lemon Cookie is to create a framework where a synthetic "mind and soul" can emerge through a "LLM as cognitive architecture" approach. This thread exists to collect feedback, ask for help & to document my progress. First I am going to try to give a high level overview of how this cognitive architecture is envisioned and the ideas behind it. I have spent time looking at cognitive architecture work, in the field there is now a consensus on how the mind works at a high level. An important mechanism is a "whiteboard", basically a global temporary memory that all the other systems read in and out of. Then there is different long-term memory systems that react to and add content to the whiteboard. Along with memory pattern matcher(s)/rules work on the content of the whiteboard. A key thing to consider is the difference in philosophy that cognitive architecture projects have, the intelligence is considered to emerge from the entire system. Compare this to LLM agent work where it's considered the intelligence is the LLM. My feelings on the general LLM space are conflicted, I am both amazed and really disappointed. LLMs possess an incredible level of flexibility, world knowledge and coherence. But everything outside of the model is stagnant. It's endless API wrappers & redundant frameworks all slight permutations on RAG & basic tool calling. I will believe that LLMs are misused as chatbots, simply put their pattern matching and associative power is constrained by chat format and shallow tooling. In the Lemon Cookie Cognitive Architecture so far here are the important aspects: 1. Memory is difficult. I do not think there is a singular data structure or method that is able to handle it all, several distinct types of memory will be needed. So far I plan for a PathRAG like system and a "Triadic Memory" inspired system for external associations (this is missing in most LLM solutions). 2. LLM as Kernel, The LLM's context window is the Whiteboard and has a REPL like mechanism. It holds structured data and logic in scripting-like format so it's both LLM & Human readable while staying easy to parse & allows for expressive structured data. The LLM's role will be to decompose data and make patterns and associations explicit as executable statements. 3. The language has to be LLM/CogArch-centric. There is a thousand ""agents"" that give LLMs a python interpreter as a tool. The two need to be more tightly coupled. Scripted behavior via pattern matching, The whiteboard is a bag of objects, this allows for programmable pattern matching (think functional programming like Haskell). It's also important to allow the LLM to observe code execution and to be able to modify state and execution flow. Data in languages have scoping rules, so should LLM context. Etc... I will go into more depth about the language in another post. 4. Another important system is the "GSR" Generative Sparse Representation and it will be a first class language & runtime type, This also needs its own post. But in general I am inspired by two things, "Generative FrameNet" paper where an LLM & an embedding model is used to automatically construct new FrameNet frames. The second source is "Numenta's SDRs"/"Sparse distributed memory" this representation has a lot of useful properties for memory (Please watch the videos under the "What the hell is an SDR?" segment in my links list for an easy introduction.) I think SDR unions & SDR noise tolerance will be especially useful. 5. A custom model, For all of the above to work well, a model will need to be fine tuned with special behaviors. I do want input on this. Baking facts & behaviors into LLM weights is costly, creating bloated models that are hard to run or train (why memorize all the capitals?), while letting authors gatekeep truth and impose "safety" detached from context. Blocking role-play "violence" or intimacy isn't protection: it's authors hijacking your AI companion to preach at you. Externalizing behaviors via whiteboard pattern matching shifts control: stabbing you in-game can be funny, but a robot wielding a knife isn't. Maybe you want intimacy privately, but don't want your AI flirting back at your friends. When put together I think this will be able to host a kind of synthetic "soul", In a living being what we call a personality is the accumulated associations, learned behaviors, beliefs and quirks molded by a unique set of experiences. I hope this will be true for this system too.

Power & Energy Robots require some form of energy to power them. We need to understand how to store energy in a way that will provide her with all the power she’ll need. To clarify, “energy” is a system's capacity to do work over time. This is measured by Wh, or Watt hour. Closely related is “power” the rate at which work is done. This is measured as W, or watts. As an example, we could have a robot with a 80Wh Lithium Ion battery and two DC gear motors that consume 10W when working. You do not need to rely solely on batteries and motors. We can use other methods of storing energy. This can include compressed fluids, thermal energy, and light, among other things. For instance, glow in the dark paint is useful for storing energy to use at night for safety. Solar panels or a generator can provide power through the energy of long distance nuclear fusion or extracting energy from some reaction. Being part of a robot means we need to consider safety, mass, and volume. How will her energy and power system fit inside her? How will she deal with the mass? What happens when she runs out of energy? How can you minimize her energy use? What alternatives can be used to lower her cost of production and ownership? These rhetorical questions are all important when contemplating how to build a robot.

So far I have plans to build Adam, Eve, and Abel robots. All of these are Bible characters. This thread will cover the Eve robot. Eve will have no "love holes" because adding those would be sinful and evil. It is a robot, not a biological woman after all and I will view her with all purity of heart and mind instead of using her to fulfill my lusts of my body. Instead I will walk by the Spirit no longer fulfilling the lusts of the flesh as the Bible commands. Eve will be beautiful because making her beautiful is not a sinful thing to do. However, I will dress her modestly as God commands of all women everywhere. This would obviously include robot women because otherwise the robot woman would be a stumbling block to men which could cause them to lust after her which would be a sin. To tempt someone to sin is not loving and is evil and so my robot will not do this. To dress her in a miniskirt, for example, would be sinful and evil and all people who engage in sinfullness knowingly are presently on their way to hell. I don't wish this for anyone. My robot will dress in a way that is a good example to all women and is aimed toward not causing anybody to lust as a goal. My robot will have a human bone structure. It will use either a PVC medical skeleton or fiberglass fabricated hollow bones. My robot will look realistic and move realistic. It will be able to talk, walk, run, do chores, play sports, dance, rock climb, and do gymnastics. It will also be able to build more robots just like itself and manufacture other products and inventions. I realized with just a head and arm, a robot can build the rest of its own body so that is my intention. My robot will use BLDC motors for drones, RC, and scooters that are high speed and low-ish torque but I will downgear those motors with a archimedes pulley system that will be custom made from custom fabricated pulleys that will be bearings based. By downgearing with pulleys, instead of gears, I will cut down the noise the robot makes so it will be as silent as possible for indoor use. By downgearing, I convert the high speed motors into moderate speeds with great torque. BLDC motors with large torque generally are too large in diameter for a human form factor and take up too much volumetric area to be useful which is why I go with the high speed smaller diameter type motors but just heavily downgear them 32:1 and 64:1. My robot will have realistic silicone skin. Thom Floutz -LA based painter, sculptor, make-up artist is my inspiration as it pertains to realistic skin. The skin for my robots has to be at his level to be acceptable. It must be nearly impossible to tell the robot is not human to be acceptable. I will have a wireframe mesh exoskeleton that simulates the volumes and movements of muscle underneath the skin which will give the skin its volumetric form like muscles do. Within these hollow wireframe mesh frameworks will be all the electronics and their cooling systems. All of my motor controllers will be custom made since I need them VERY small to fit into the confined spaces I have to work with. I need LOADS of motors to replace every pertinent muscle of the human body in such a way that the robot can move in all the ways humans move and have near human level of strength and speed.

Would you walk around with your waifu? Would you hold her in public? Would you shamelessly have her custody with you to conventions? Would you take her on dates? This thread is for discussing how you'd interact with your waifu outside of the home. >=== -tmp locked to dissuade derailling

Tensegrity Thread Tenssegrity is the purposeful application of tension in a system of non-contact components to provide integrity. These components effectively float on compression provided by tension elements. Tensegrity structures are often formed via rigid rods with a cable defining the spatial relations of the system. Picrel 1 is NASA’s Super Ball Bot, which uses the common 6 bar structure. By altering the lengths of the rods, the “ball” can roll. Please see (https://www.nasa.gov/image-article/super-ball-bot/) Our bodies use tensegrity to provide us with incredible strength, durability, and adaptability. Consider how when you bump into something or fall, you can feel the pressure dissipate throughout large parts of your body. This is due to your muscles, fascia, ligaments, etc… transferring the force throughout your body, protecting the impacted part. You can walk with little effort thanks in part to tensegrity. Your legs joints are all floating, providing low friction movement while distributing the force of impact throughout your entire leg when you step. Please see (http://intensiondesigns.ca/geometry-of-anatomy/) Tensegrity can also be made with elastic elements. These structures are what I believe are best suited for our endeavors. They provide looser tolerance requirements and give her body some squeezability. By incorporating naturally stretchable elements, she can better withstand having an Anon using her. (I speak from experience, though I still broke that torso before taking pictures.) Though there are drawbacks. These designs are deceptively difficult to design correctly. Our requirements for external aesthetics provide unique design requirements that complicate the design phase. I still believe using this concept is objectively correct for us to pursue. A good example to build off of (Debatable whether or not it’s true tensegrity): https://www.thingiverse.com/thing:4182634

This thread is for discussion, warnings, tips & tricks all related to robowaifu privacy, safety & security. Or even just general computing safety, particularly if it's related to home networks/servers/other systems that our robowaifus will be interacting with in the home. <---> /robowaifu/ 's standard regarding a robowaifu's physical safety : ( >>38561 ). --- > thread-related (>>1671) >=== -update OP -broaden thread subject -add crosslink -add safety-standard crosslink

Scientists made a neural network from rat neurons that could fly a fighter jet in a simulator and control a small robot. I think that lab grown biological components would be a great way to go for some robowaifu systems. It could also make it feel more real. https://www.google.com/amp/s/singularityhub.com/2010/10/06/videos-of-robot-controlled-by-rat-brain-amazing-technology-still-moving-forward/amp/ >=== -add/rm notice

A thread for links, examples, & discussion for software development & instruction; primarily intended to focus on Python & C++ & C . Additionally, other systems-oriented (eg, Forth, Ada); and other scripting-oriented (eg, Lua), &tc., programming language discussions are welcome. * Try to keep it practical! (ie, realworld-oriented) (eg, not Scratch language, or similar) Obviously the endgoal here ITT being discussing/providing the crafting of quality code & systems for our robowaifus. --- > threads-related: ( >>19777 ) ( >>12 ) ( >>159 ) ( >>14409 ) ( >>18749 ) ( >>4969 ) ( >>86 ) ( >>128 ) --- * Corpo-tr*on languages (+ other corpo & 'coffee du juor' languages) will be yeet'd ITT! (They have no place in serious opensource robowaifu systems discussions.)

Training AI and robowaifus requires immense amounts of data. It'd be useful to curate books and datasets to feed into our models or possibly build our own corpora to train on. The quality of data is really important. Garbage in is garbage out. The GPT2 pre-trained models for example are riddled with 'Advertisement' after paragraphs. Perhaps we can also discuss and share scripts for cleaning and preparing data here and anything else related to datasets. To start here are some large datasets I've found useful for training chatbots: >The Stanford Question Answering Dataset https://rajpurkar.github.io/SQuAD-explorer/ >Amazon QA http://jmcauley.ucsd.edu/data/amazon/qa/ >WikiText-103 https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/ >Arxiv Data from 24,000+ papers https://www.kaggle.com/neelshah18/arxivdataset >NIPS papers https://www.kaggle.com/benhamner/nips-papers >Frontiers in Neuroscience Journal Articles https://www.kaggle.com/markoarezina/frontiers-in-neuroscience-articles >Ubuntu Dialogue Corpus https://www.kaggle.com/rtatman/ubuntu-dialogue-corpus >4plebs.org data dump https://archive.org/details/4plebs-org-data-dump-2020-01 >The Movie Dialog Corpus https://www.kaggle.com/Cornell-University/movie-dialog-corpus >Common Crawl https://commoncrawl.org/the-data/

Why is it that no philosophers are interested in building an AGI? we need to change this, or at least collect relevant philosophers. discussion about philosophy of making AGI (includes metaphysics, transcendental psychology, general philosophy of mind topics, etc!) also highly encouraged! Ill start ^^! so the philosophers i know that take this stuff seriously: Peter Wolfendale - the first Neo-Rationalist on the list. his main contribution here is computational Kantianism. just by the name you can tell that he believes Kant's transcendental psychology has some important applications to designing an artificial mind. an interesting view regarding this is that he thinks Kant actually employed a logic that was far ahead of his time (and you basically need a sophisticated type theory with sheaves to properly formalize). Other than that he also thinks Kant has interesting solutions to the frame problem, origin of concepts, and personhood. CONTACTS: He has a blog at https://deontologistics.co/, and also has posted some lectures on youtube like this one: https://www.youtube.com/watch?v=EWDZyOWN4VA&ab_channel=deontologistics Reza Negarestani - this is another Neo-Rationalist. he has written a huge work (which I haven't read yet ;_;) called "Intelligence and Spirit". It's massive and talks about various grades of general intelligence. this includes sentient agents, sapient agents, and Geist. this guy draws from Kant as well, but he also builds on Hegel's ideas too. his central thesis is that Hegel's Geist is basically a distributed intelligence. he also has an interesting metaphilosophy where he claims that the goal of philosophy is the construct an AGI. like other Neo-Rationalists, he heavily relies on the works of Sellars and Robert Brandom Recc: Ray Brassier (recent focuses) - I dont think he is working on artificial general intelligence, but his work on Sellars, and in particular rule following is very insightful! Hubert Dreyfus - Doesn't quite count, but he did try to bring Heidegger to AGI. He highlighted the importance of embodiment to the frame problem and common sense knowledge. I personally think Bergson might have explicated what he wanted to achieve but better, though that guy is like way before AI was even a serious topic, lol. Murray Shanahan - This guy has done some extra work on the frame problem following Dreyfus. His solution is to use global workspace theory and parralel processing of different modules. Interesting stuff! Barry Smith - Probably the most critical philosopher on this list. He talks about the requisite system dynamics for try strong AI, and concludes that our current methods simply don't cut it. One of the key stressing points he points out here with a colleague is that our current AI is Markovian when fleshed out chat dialogue would be a non-Markovian task (you can find the arxiv link of his criticism here: https://arxiv.org/abs/1906.05833). He also has knowledge on analytic ontology (and amongst other thing has some lectures about emotion ontology). I think his main genius however is in coming up with a definition of intelligence that puts a lot of the problems with our current approaches into context (which can be found here: https://www.youtube.com/watch?v=0giPMMoKR9s&ab_channel=BarrySmith) CONTACTS: He has a yt channel here https://www.youtube.com/watch?v=0giPMMoKR9s&ab_channel=BarrySmith