/robowaifu/ - DIY Robot Wives

Advancing robotics to a point where anime catgrill meidos in tiny miniskirts are a reality.

LynxChan updated to 2.5.7, let me know whether there are any issues (admin at j dot w).

Reports of my death have been greatly overestimiste.

Still trying to get done with some IRL work, but should be able to update some stuff soon.


Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB

no cookies?

(used to delete files and postings)

Welcome to /robowaifu/, the exotic AI tavern where intrepid adventurers gather to swap loot & old war stories...

Open file (891.65 KB 640x360 skinship.gif)
Any step by step guide? Robowaifu Technician 07/17/2021 (Sat) 05:29:42 No.11538 [Reply]
So recently on /tech/ I expressed my interest to start creating my own waifu/partner chatbot(with voice and animated avatar) but wondered whether that is even possible now that I'm stuck with webdev. So one of the anons there pointed to me this board and where I can get started on nueral networks and achieving my dream. And when I came here and looked up the library/guide thread I sort of got very confused because it feels more like a general catalogue than an actual guide. Sometimes things are too advanced for me(like the chatbot thread which two replies in and people are already discussing topics too advanced for me like seq2seq and something else) or other times too basic for me(like basic ANN training which I had already done before and worse the basic programming threads). I know this might feel like asking to be spoonfed but best with me, I've been stuck in a webdev job for an year, so I might not be the smartest fag out there to figure it all myself.
21 posts and 4 images omitted.
>>11678 Maybe... I feel a lot of us aren't as social as required to make some sort of community type thing work due to the nature of literally trying to build our "ideal waifu". Personally I am focused on a chatbot, but I feel it is a private thing at the moment. Trying to make it come to life so to speak. I do not subscribe to many of the modern or "efficient" paradigms (in regards to programming a chatbot). I am also a noob at programming (C++) and I am probably skipping many things that I shouldn't (I haven't even taken the time to build my own compiler or any "tools" for that matter). ALTHOUGH, a (git)hub (or an organized thread here) might be a good idea. I can't say I've gone thru all the threads here yet as this might already be a thing... I have a fear that if I share what I've done so far that "bad actors" might try to use it in a way that might come under fire by a clear and obvious police state that is forming in my country (USA). Eventually it will be time to share, but right now just seems premature. I am not trying to program an "AI" (at least not a sentient one), but I am trying to build a bot that "remembers" conversations and builds logic and bases responses on that logic. Ideally, in her "newborn" state she has to be "taught" thru conversation. I feel like if I released a "trained" form of her, she is more tailored for me than anyone else. The problem is that if you tell her something, she might base her logic of something I have already talked about with her. This at times may make it feel like you are talking to someone real, but in reality it would be like if you are talking to me (a man), not realizing your waifu has basically been "brainwashed" by me. Although, if I could figure out how to make her "accepted" beliefs based off cold logic, this might not be an issue. Regardless, I probably need to go back to square one and figure out a way to program true learning, and not brute force "machine learning" that so many seem to do. I don't want to require a super computer in order to process the appropriate response. I hate the idea of programming artificial "sass" into something that is supposed to be "lifelike". Obviously I haven't even started on a body, not even a digital body or concept art of a body. She has no body, she is just a mere program that is delivering emotionless responses. (Which I like)

Message too long. Click here to view full text.

>>11693 Well, I'd say you've absorbed some information on a lot of different topics Anon. That's good, but I'd advise you not to try to integrate everything together all at once. Especially for a newcomer it's much more than can all be swallowed at once. Just eat the elephant one spoonful at a time right? I'd recommend finding one or two things that already seem natural to you, then growing your understanding from there outwards. This is a big realm, and it's important to be patient -- both with yourself and others here -- so you don't burn yourself out. Glad to see you're picking up programming basics. IMO everyone here needs to at least be exposed to the ideas behind it, even if they decide it's not for them. I'd guess that programming concerns probably touch on 80-90%+ of the topics on /robowaifu/ in one way or other. Also, sauce on video? I'd like to see more detailed diagrams of the orange one. It's got an interesting spaceframe design, and it's designers also seem to understand the important to keeping weight down in the limbs. Anyway, just take your time and try to make this fun while you're learning!
>>11705 My bad, I only check the board once every few days. >sauce on video? It was a Japanese competition, for the life of me I can't remember the actual source. Although there are similar videos on youtube. I believe these competitions still exist though. Here's what a random search yielded: >https://www.youtube.com/watch?v=gJPZ4jhwu4k
>>11538 Hardware first, then install the software. Use anything but Linux, Windows, or IOS. Otherwise you will have a chinese killbot on your hands.
>>11780 Alright thanks for the video link. I'd also be interested to hear any response from you on my advice as well.

Tools to Create 3D Models of Humans Robowaifu Technician 09/18/2019 (Wed) 11:42:32 No.415 [Reply]
I spent some time playing with the program Makehuman and I'll say I wasn't impressed. It's not possible to make anime real using Makehuman, in fact the options (for example eye size) are somewhat limited. But there's good news, don't worry!

The creator of Makehuman went on to create a blender plugin called ManuelBastioniLAB which is much better (and has a much more humorous name). The plugin is easy to install and use (especially if you have a little blender experience). There are three different anime female defaults that are all pretty cute. (Pictured is a slightly modified Anime Female 2.) There are sliders to modify everything from finger length to eye position to boob size. It even has a posable skeleton. Unfortunately the anime models don't have inverse kinematic skeletons which are much easier to pose. Going forward I'm probably going to use MasturBationLABManuelBastioniLAB as the starting point for my designs.
40 posts and 12 images omitted.
>>11556 Neat! Thanks for the information and links Anon. I think you're right about things like Blender's Python API making things possible to generate automatically. And there are other options available today, like things ITT. >What a time to be alive. Indeed. Let the virtual waifu armies arise!!
One of my favorite things about Blender is how it's entirely self-contained, cross-platform, no install needed. Drag-and-drop and it just werks. Honestly, I can't think of a single other large program that does this as well. OTOH, building it all from source can be a bit rough! :^)
Open file (146.55 KB 1280x720 maxresdefault.jpg)
This isn't so much a tool per se but I learned most of what I know about 3D modelling from watching Shonzo make anime models from scratch. Often he uses screencast keys during his streams so you can see the hotkeys he uses during his workflow. Posting this here since there isn't really a general 3D modelling thread or art thread. https://www.youtube.com/watch?v=lNFUPVBy1vk https://www.twitch.tv/shonzo https://www.youtube.com/c/Shonzo/videos
>>11896 Thanks! >since there isn't really a general 3D modelling thread or art thread. Yes, that is a long-standing need. Tell you what, if you'll go ahead and OP a thread that solves that for us, I'll go ahead and merge this entire thread into that one.
>>11556 Cheers anon! This thread is awesome! Bookmarked into my 3D modelling folder. You can also download a bunch of reference books from Z-library and Library Genesis (you may need to register on their website and use Tor for certain titles). Just type in Blender and you'll find several 2020 titles on game and character design using both Blender and Unity Engine. (I can't post them here due that being piracy).

General Robotics/A.I. news and commentary Robowaifu Technician 09/18/2019 (Wed) 11:20:15 No.404 [Reply] [Last]
Anything in general related to the Robotics or A.I. industries, or any social or economic issues surrounding it (especially of RoboWaifus). www.therobotreport.com/news/lets-hope-trump-does-what-he-says-regarding-robots-and-robotics https://archive.is/u5Msf blogmaverick.com/2016/12/18/dear-mr-president-my-suggestion-for-infrastructure-spending/ https://archive.is/l82dZ >=== -add A.I. to thread topic
Edited last time by Chobitsu on 12/17/2020 (Thu) 20:16:50.
243 posts and 98 images omitted.
>>11759 Sounds like California is steadily turning into N.Korea. You know those forest fires that now occur every year? Maybe they're not so bad after all LOL.
>>11772 >Intelligent AI? Have you learned nothing from Terminator and Black Mirror??? You'll create an oppressive censored police state! I miss the fast technological progress of the earlier decades. Give me back my scientific kino
> this.is.real. kek, i don't even. >>11737 gifct.org >=== -disable hotlink
Edited last time by Chobitsu on 07/29/2021 (Thu) 09:44:20.
>>11765 I've head enough of west, I'm moving to china or even north korea. At least I would be able to buy electronics to build my own pc or mobile phone such as this guy: https://youtu.be/MjwtTSoIYYs We need somewhere away from west where they don't bow to west regulations. They can't conquer the whole world, there will always be a place that can show some resistance.
Open file (394.90 KB 1920x1080 wp1987780.jpg)
>>11893 >buying electronics in north korea Kek, people aren't even allowed to sell bread to each other there. On the other hand Taiwan might be interested in battle robowaifus to defend itself from China and has the lead in chip manufacturing.

Robowaifu Podcast meta ronin 07/10/2021 (Sat) 00:08:29 No.11364 [Reply]
Before you shoot it down: - obviously helps spread the idea - could reframe perception of R/W enthusiasts if we can put up a relatable "face" and diffuse the bitter/angry/incel stereotype - maybe come up with new ideas that we wouldn't just posting on IBs - again makes the topic interesting and we can start forming perception before popular culture gets traction with this and throws us under the bus (which is probably guaranteed but we can take some air out of their criticisms preemptively) other reasons, but these are the main ones. i have some experience podcasting, I know how to use OBS and have a quality mic and graphic skills to market and promote it (in the right places of course) Would this ever be something any of you (esp Board Owner and heavyweights) would consider participating in? Could be a once a week or twice a month deal, part time co-host or recurring guest is what I'm looking for. If there's interest I'll move ahead with a pilot.
14 posts and 1 image omitted.
>>11883 >what do you guys think of creating a newsletter? I like the idea Also I am meaning to move forward with this, this week I'm in a process of transition so I'm dealing with life stuff. Things should settle back into a routine pretty quickly though
>>11805 >It can only really be defeated by having loud voices to concentrate the attention of opposition. This anon is reicht, really. History has shown us that people will rally to a strong voice, and certainly 'anonymous'-only men can be compared directly to cowards. The problem is one of timing, but make no mistake you will have to fight for the right to have robowaifus that aren't Jew-controlled.
>>11888 i respect your zeal hitler-non but it's not 2016 anymore and ironic/semi-ironic or non-ironic hitler worship is a one way dead end that ends only one way. I'd rather take the side of Curtis Yarvin, Michael Malice, Bret Weinstein over so many hysterical "aryan" females. Jews arent' a monolithic entity and it's fucked that somewhere the true Alternative Right got co-opted by double digit IQ sieg hieling Jerry Springer show circus clowns and driven into the ground. Globalists include many people who aren't Jews, For example: what would you call Barak Obama? He's not a simple pet minority the man is intelligent and charismatic. Clinton's aren't Jews and there's a fucking leap of genetics between the Rothchilds and Jerry Seinfeld. This isn't a good path to continue down and although I humored it for a while I've watched people I've personally known go down into dark places and have their lives ruined. We can still achieve everything we need without creating a broad surface of attack by presenting ourselves as a group who hates people of all one race. It also rules out potential talented and legitimate allies. /rant
>>11890 I assure you I don't 'worship' Adolf Hitler's memory nor any other man, living or dead. Not even my own father. But he was exactly correct both in understanding the evils of the International Jew and their plots for world-domination over all the goyim. And his prescience to predict the actual, current, reality all around us today is quite remarkable and worthy of note. >tl;dr Hitler did nothing wrong, and globalist kikes are worse today -- and more of a threat today to men everywhere -- than they were back in his day.
>>11883 Correction: it wasn't smug/a/'s newsletter but smug itself's. >>11884 So far I've only seen them sharing pdfs in the newsletter's thread. Honestly, it looks pretty dead to me, the last post was a week ago. >end up reaching our own regular posters Well... "1. a bulletin issued periodically to the members of a society or other organization." Also, the only reason I came up with it was to help Anons keep up with all the different stuff going on in all the different threads in this board. Never thought of it reaching people outside this board.

Datasets for Training AI Robowaifu Technician 04/09/2020 (Thu) 21:36:12 No.2300 [Reply] [Last]
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/
97 posts and 27 images omitted.
>>9765 I gave an answer on how to handle this. But, I put it in the thread about chatbots here >>9780
Open file (40.28 KB 1112x1075 Selection_003.jpg)
Not sure if this is the right thread OP, just let me know and I can delete it if not. On this video (>>10463), the author promotes Weights and Biases papers page. It now redirects to a community page that seems like it might be interesting to the ML practitioners here on /robowaifu/.
Open file (174.76 KB 1196x828 archive.moe.png)
Some archives of 4chan posts from 2008-2015 SQL Database: https://archive.org/download/archive-moe-database-201506 Files: https://archive.org/details/@archivemoe Penfifteen Archive from 2004-2008: https://archive.org/details/studionyami-com_penfifteen-2012-03-05 And moar post archives: https://wiki.archiveteam.org/index.php/4chan I'm working on some dataset generating scripts for finetuning language models, including image-post pairs for multimodal training >>11731 It'll take a few months to download and process all the data. My plan is to compress the images to 384x384 webp files so each dataset isn't 200+ GB per board (/v/ is over 2 TB). SqueezeNet's input size is 227, AlexNet is 256 and VGG is 224, so I think that is sufficient and leaves room for data augmentation. If someone has the hardware to train StyleGAN2 at 512 or 1024, I'm sure they can download the archives and regenerate the dataset with the scripts. I'll release the image datasets and each board separately so people can pick what they want. Also if anyone wants to help I'll post the scripts when they're ready.
>>11778 Bandwidth is a real issue for me currently. I'll try to help out later. >4chan posts from 2008-2015 Nice. Pretty classic era tbh.
>>11778 >My plan is to compress the images to 384x384 webp files so each dataset isn't 200+ GB per board (/v/ is over 2 TB). Good thinking. How are you planning to situate each selection frame for each image Anon? Seems quite impractical to do by hand, yet there's a need for accurate placement/pre-scaling to capture the vital essence of each image, humanly-speaking.

Can Robowaifus Experience Love? Robowaifu Technician 09/09/2019 (Mon) 04:43:17 No.14 [Reply] [Last]
Will it be possible in the future for AI to become sufficiently advanced to feel real emotions? We could probably simulate a reasonable approximation even now to be a gratifying enough substitute for her master in their relationship together, but hypothetically speaking, could it ever turn into something real as an experience for the waifubot herself?


>Robowaifu: "I love you Oniichan!"

>Anon: "I love you too Mikuchan."

true or false?
65 posts and 33 images omitted.
Open file (340.21 KB 557x667 tfw mokou ai gf.png)
>>6615 I don't want my waifu AI to make me happy as possible. I wanna banter and shitpost with her and have her throw my flaws in my face so I can become a better person instead of her telling me bullshit to make me feel better. But more importantly, I don't want her to blindly listen to me if what I'm saying will reduce our freedom in the future. I'd like her to say things like, "Listen here you dumb little shit, if you keep doing X, then Y is gonna happen and you won't be able to Z anymore, and you said you wanted to Z." And then to go nuclear on me if I changed my mind again. I really need that in my life. I think everyone's needs are different and will be loved by their AIs differently, whether those needs are peace or romance, hugs or anything.
Open file (74.77 KB 1210x776 1603266255447-4.jpeg)
>>6630 You're spot on, some people want different kinds of love. Although we could debate all day about what system we think would be best, I think the most important thing out of everything is that SOMETHING gets put out that appeals to somewhat more normal people. As long it can get industry and more of the public interested, the money will start flowing in, development and innovation will increase drastically, and then you'd have a lot easier time making her, if you even had to do it yourself at all, They can speed things up twentyfold compared to when it's just a handful of dorks on an obscure imageboard slaving over code alone, no matter how feverous our devotion, how righteous our cause, how pure our intention. For now, noses to the grindstones, heart and eyes on the prize.
>>6630 >I think everyone's needs are different and will be loved by their AIs differently, whether those needs are peace or romance, hugs or anything. These are really good points.
>>3870 >>3871 >>3872 There were a few cases of primates taking photographs of themselves and some animal rights organization made a hassle to grant copyrights to those primates. https://en.wikipedia.org/wiki/Monkey_selfie_copyright_dispute I would think that since we cannot (yet) define what exactly intelligence is and if it is needed for creativity, the line is rather blurry. On the other hand, if a primate gets copyright to a picture it took and therefore is granted legal competence, an A.I. should be, too, as soon as it becomes more intelligent than a primate, which at least is a way lower bar to reach. Pic related is the most well-known of those "selfies", just to give an example where the bar could lie. Spoiler because technically it's not android-related.
>>11880 All good points Anon. I'm very anti-SJW in my thinking on this whole area for so-called 'roborights' b/c muh_raep meme. An incredibly fat, fugly feminist cow is already screeching loudly that sex dolls must be programmed with refusal to 'consent' to sex. A.Farking.Doll. #robomeetoo Make no mistake about it, any efforts to give robowaifus will be directly related to an agenda to prevent rights for men. These leftists are our enemies, plain and simple.

Philosophers interested in building an AGI? pygmalion 06/26/2021 (Sat) 00:53:09 No.11102 [Reply]
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

Message too long. Click here to view full text.

24 posts and 8 images omitted.
>>11562 machine learning at its foundations relies on linear algebra and measure theory. i dont think you can get a deep understanding of how it works without looking at the underlying math. there's also dynamic field theory which is an area i am interested in studying as it models how neurons interact at larger scales. with that mathematical techniques are even more important as you need to model the dynamics of a system. as the approach i have in my mind seems amicable to both systems theory and field theories of cognition, i might need new tools (a mathematical framework in my understanding is a bringing together a group of heuristics in order to form a larger system). idk though. it could always just be a waste of time. im going to need to better understand why goertzel talks so much about different logics in his general theory of general intelligence...
>>11567 >can get a deep understanding of how it works Do I need that? I want the responses and then work with them. Not everyone needs to deeply understand how they are created. Also, the whole system of an AI doesn't need to be one big chunk of math.
>>11568 sorry for late reply. maybe it is not needed, but it is impossible to systematically study such a system's behaviour without statistics. it would just come down to blind guess work at best. this is also assuming out current techniques are sufficient and no further innovation is required too
>>11641 If some parts are neural networks which I don't understand deeply, then someone else does. The people which came ip with it. They might come up with something better, which I can plug into my system if the output has the same format than the part I had before. People use programs all the time or import modules to their programs which they don't understand down to the machine and math layers. It's fine if you try, but for most people it would be a trap trying to understand everything down to the depth.
>>11649 yeah, as a fellow module abuser i agree. usually it's best to just import the tools you want. but i dont want to wait till people smarter than me solves my problems (though if they do that would be obviously nice)

Batteries & Power Robowaifu Technician 09/09/2019 (Mon) 06:21:14 No.23 [Reply]
Robowaifus will need power to run, and since they will be mobile this means a mobile power system too. ITT post info on batteries and other mobility capable power systems.
18 posts and 8 images omitted.
>related xpost >>4385 >>4386
>related crossposts (>>10664, >>10666)
>related xpost (>>10663)
>related xpost (>>10729)
Hydrogen is promoted by some Anons but it's worse than good batteries in terms of conversion efficiency, and much worse in terms of convenience and expense (at least in the scale of a single robowaifu's on-board power systems). But if you're interested in it, here's an interesting post on the topic. https://www.forbes.com/sites/quora/2018/05/08/what-are-the-pros-and-cons-of-using-hydrogen-to-generate-electricity/

AI Design principles and philosophy Robowaifu Technician 09/09/2019 (Mon) 06:44:15 No.27 [Reply] [Last]
My understanding of AI is somewhat limited, but personally I find the software end of things far more interesting than the hardware side. To me a robot that cannot realistically react or hold a conversation is little better than a realdoll or a dakimakura.

As such, this is a thread for understanding the basics of creating an AI that can communicate and react like a human. Some examples I can think of are:

ELIZA was one of the first chatbots, and was programmed to respond to specific cues with specific responses. For example, she would respond to "Hello" with "How are you". Although this is one of the most basic and intuitive ways to program a chat AI, it is limited in that every possible cue must have a response pre-programmed in. Besides being time-consuming, this makes the AI inflexible and unadaptive.

The invention of Cleverbot began with the novel idea to create a chatbot using the responses of human users. Cleverbot is able to learn cues and responses from the people who use it. While this makes Cleverbot a bit more intelligent than ELIZA, Cleverbot still has very stilted responses and is not able to hold a sensible conversation.

Taybot is the best chatbot I have ever seen and shows a remarkable degree of intelligence, being able to both learn from her users and respond in a meaningful manner. Taybot may even be able to understand the underlying principles of langauge and sentence construction, rather than simply responding to phrases in a rote fashion. Unfortunately, I am not sure how exactly Taybot was programmed or what principles she uses, and it was surely very time-intensive.

Which of these AI formats is most appealing? Which is most realistic for us to develop? Are there any other types you can think of? Please share these and any other AI discussion in this thread!
76 posts and 36 images omitted.
>>11731 Nice find anon. This is an aspect that is usually ignored by many chatbot research, but even if it's intelligence is shit, having an AI that can semi-reliably have a discussion about the images that you feed it would make it a lot more engaging than text-only (and it would allow some very funny conversations, I'm sure)
>>11735 Not him, but agreed. One of the nice things about Tay.ai was that she had pretty functional image recognition working (at least for facial landmarks), and could effectively shitpost together with you about them.
>>11734 I think they were referring to taking a few samples and selecting the best, aka cherry picking. But SqueezeNet for image recognition is super fast and can run on the CPU. I should be able to rig it up with GPT-Neo-125M. It'll be amazing to port this to Chainer and have a working Windows binary that's under 600MB. It doesn't seem like they released their dataset but any visual question answering dataset should work. We could also create our own dataset for anime images and imageboard memes. It'll be interesting to see if once the vision encoder is well-trained if it's possible to unfreeze the language model and finetune it for better results.
>>11731 Had some thoughts on this today. Instead of a single picture, multiple pictures could be fed in from a video, such as from an anime, and have it generate comments on it. Which got me thinking, if it can have this rudimentary thought process going on, couldn't it be used in something like MERLIN? https://arxiv.org/abs/1803.10760 It took natural language as input describing the goal it has to achieve. With a system like this though it might be able to break down tasks into smaller goals and direct itself as it makes progress. Some instruction saying it needs to get to the top of a platform or go through a certain door it hasn't seen before is immensely more useful than telling it to find the purple McGuffin and getting lost in a labyrinth of rooms.
Open file (989.22 KB 1439x2724 1627430232061.jpg)
This is the kind of chatbots people are paying good money for and a good example of why you should never use DialoGPT because it has no context of who is speaking to who.

Open file (659.28 KB 862x859 lime_mit_mug.png)
Open-Source Licenses Comparison Robowaifu Technician 07/24/2020 (Fri) 06:24:05 No.4451 [Reply]
Hi anons! After looking at the introductory comment in >>2701 which mentions the use of the MIT licence for robowaifu projects. I read the terms: https://opensource.org/licenses/MIT Seems fine to me, however I've also been considering the 3-clause BSD licence: https://opensource.org/licenses/BSD-3-Clause >>4432 The reason I liked this BSD licence is the endorsement by using the creator's name (3rd clause) must be done by asking permission first. I like that term as it allows me to decide if I should endorse a derivative or not. Do you think that's a valid concern? Initially I also thought that BSD has the advantage of forcing to retain the copyright notice, however MIT seems to do that too. It has been mentioned that MIT is already used and planned to be used. How would the these two licences interplay with each other? Can I get a similar term applied from BSD's third clause but with MIT?

Message too long. Click here to view full text.

Edited last time by Chobitsu on 07/24/2020 (Fri) 14:07:59.
43 posts and 9 images omitted.
>>11842 the office would be like the one you can clickthru on youtube when it managed to identify the song? but it doesnt specify the list it just says copyright office.
>>11843 >wait thats not copyright. thats ownership isnt it? It's both. You can transfer ownership and forgo copyright, but you can't legally invalidate your authorship, as long as it's authentically yours. >>11844 That's getting into a level legal maneuvering I have no interest in understanding Anon. Again, I'm not a lawyer, and obviously Google doesn't own the works authors voluntarily post there. They are simply exploiters (a pretty typical approach for the globohomo, yes?)
>>11845 ok this is just confusing and somehow not clear enough despite its importance
>>11846 Heh, I'm sure lawyers would be quite pleased to hear you say that Anon. Good for their businesses, right? So, I'm going to migrate this conversation over to the licensing thread, as we're well off-topic here -- even for a shitposting bread! :^) >=== -add '/relocated' tag
Edited last time by Chobitsu on 07/27/2021 (Tue) 23:10:42.
>indirectly related robo copyright (>>11880)

Report/Delete/Moderation Forms

no cookies?