/robowaifu/ - DIY Robot Wives

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

Happy New Year!

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB

Captcha
no cookies?
More

(used to delete files and postings)


“Perseverance, secret of all triumphs.” -t. Victor Hugo


Open file (58.56 KB 1244x2374 MaidComFront.jpg)
Open file (39.04 KB 1244x2318 MaidComSide.jpg)
Open file (56.65 KB 1244x2318 MaidComBack.jpg)
MaidCom Kiwi 02/08/2024 (Thu) 06:40:01 No.29219 [Reply] [Last]
MaidCom Project Thread 2 Project Goal: Simple, low cost, extensible platform for robowaifu development. Picrel shows the model which is being developed. Every part will be designed to be easily printable and replaceable. Her designs will be open source. Modding and customization will be encouraged. This is a base model. Early revisions will be heavily limited in functionality. I invite you to help design and define standards so that it is easy to create specialized add-ons to allow her to become your own waifu. Collaboration is important, MaidCom is officially partnered with Lin and his Waifu Wheelchair, if you'd also like to be officially partnered, say so. Cat eats and miniskirts are encouraged! <--- Previous thread: ( >>15630 ) >=== -add crosslink
Edited last time by Chobitsu on 08/14/2024 (Wed) 21:50:13.
197 posts and 134 images omitted.
>>44215 >It is sensible, a rolling box with an upper body that adjusts up or down is likely the min-maxed configuration. I think it probably is too. >Reduction in needed computation is another major win for this configuration. Less kinematic and balancing equations. I would add to that list "more manufacturable" as well. (At least for the 'boxy' parts of the design.) >Been mulling ideas of her body being detachable from her sturdy base as needed. Yeah, I'd kinda come to the same conclusion with this one addition: interchangeable 'snuggly' lower half for cuddlin' & stuff. :) If you recall our discourse with @Lin Lin, where are you? ( >>24744, et al), we had the idea of a universal 'connection block' to attach various 'torsos' to the rolling base. >Currently considering her "body" being an ultralight structure with just enough articulation to accomplish simple things. Her "base" essentially being the main robot, where all her strength and mobility would be hidden away. I think this is a very-rational approach rn, Kiwi. Godspeed & cheers. :)
Edited last time by Chobitsu on 04/07/2026 (Tue) 15:49:54.
Open file (124.60 KB 1900x1302 Galatea absolute victory.jpg)
>>44215 >Only problem is, it's not particularly cute and hard to cuddle. Yeah, that is a unique problem that robowaifuists have to deal with. I've mulled about what would I do if I wasn't constrained by that >>44202 >Been mulling ideas of her body being detachable from her sturdy base as needed. Currently considering her "body" being an ultralight structure with just enough articulation to accomplish simple things. Her "base" essentially being the main robot, where all her strength and mobility would be hidden away. That is a good stopgap until we get reliable legs. (Total Galatea cultural victory)
>>44222 "Please clap, anons" -Galatea v3
>>44223 ¡GALATEA!
Edited last time by Chobitsu on 04/09/2026 (Thu) 10:04:47.

The Library of /robowaifu/ Card Catalogue Robowaifu Technician 11/26/2020 (Thu) 07:11:30 No.7143 [Reply] [Last]
Robowaifus are a big topic. They need a big library index! :^) Note -This is a living document. Please contribute topical thread/post crosslinks! Thread category quick-jumps >>7150 AI / VIRTUAL_SIM / UX_ETC >>7152 HARDWARE / MISC_ENGINEERING >>7154 DESIGN-FOCUSED >>7156 SOFTWARE_DEVELOPMENT / ETC >>7159 BIO / CYBORG >>7162 EDUCATION >>7164 PERSONAL PROJECTS >>7167 SOCIETY / PHILOSOPHY / ETC >>7169 BUSINESS(-ISH) >>7172 BOARD-ORIENTED >>7174 MISCELLANEOUS

Message too long. Click here to view full text.

Edited last time by Chobitsu on 12/20/2022 (Tue) 00:28:46.
124 posts and 35 images omitted.
waifusearch> center mass OR central mass OR center of gravity ORDERED: ======== THREAD SUBJECT POST LINK Robowaifus in media >>10873 (center, mass), central mass, center of gravity Bipedal Robot Locomotion General >>1776 center of gravity " >>1806 " " >>1873 (center, mass), center of gravity " >>9054 center of gravity " >>12562 (center, mass), center of gravity Waifu Robotics Project Dump >>8604 " Actuators for waifu movement! >>8986 (center, mass), central mass, center of gravity Prototypes and failures >>12937 (center, mass), center of gravity /robowaifu/ Embassy Thread >>4855 center of gravity F = ma >>7788 (center, mass), central mass, center of gravity /robowaifu/meta-5: It's Good To >>15533 center of gravity

Message too long. Click here to view full text.

waifusearch> thrown weight OR unsprung mass THREAD SUBJECT POST LINK Electronics General >>4736 thrown weight Waifu Robotics Project Dump >>10121 " Humanoid Robot Projects Videos >>16373 " General Robotics/A.I. news and c >>1238 " Actuators for waifu movement! >>11977 " Prototypes and failures >>12937 " " >>12952 " Hand Development >>11485 " Elfdroid Sophie Dev Thread 2 >>14916 " " >>14931 " Actuators For Waifu Movement Par >>13203 " " >>13210 unsprung mass " >>13211 thrown weight, unsprung mass /robowaifu/meta-4: Rugged Mounta >>13060 thrown weight

Message too long. Click here to view full text.

waifusearch> reluctance THREAD SUBJECT POST LINK Actuators for waifu movement! >>12014 reluctance " >>12062 " " >>12131 " " >>12140 " " >>12482 " " >>12910 " " >>12912 " " >>12915 " Actuators For Waifu Movement Par >>12930 " " >>13229 " " >>13236 " " >>13257 " ' reluctance ' = 12 results
waifusearch> Dielectric elastomer OR Dielectric elastomers THREAD SUBJECT POST LINK Robo Face Development >>6562 dielectric elastomers R&D General >>8502 " " >>8503 " " >>8505 " " >>8519 " " >>10049 " Waifu Robotics Project Dump >>8606 " Actuators for waifu movement! >>10639 dielectric elastomer, dielectric elastomers Actuators For Waifu Movement Par >>12929 dielectric elastomers " >>12932 dielectric elastomer " >>12939 dielectric elastomers " >>13229 " " >>13231 " " >>13232 "

Message too long. Click here to view full text.

>>7143 Iran's absolutely-amazing, free, public library (also pictured in OP). https://trashchan.xyz/robowaifu/thread/26.html#1441
Edited last time by Chobitsu on 04/05/2026 (Sun) 07:29:53.

Robotics Hardware General Robowaifu Technician 09/10/2019 (Tue) 06:21:04 No.81 [Reply]
Servos, Actuators, Structural, Mechatronics, etc.

You can't build a robot without robot parts tbh. Please post good resources for obtaining or constructing them.

www.servocity.com/
https://archive.is/Vdd1P
18 posts and 5 images omitted.
>>43717 Precision is also a virtue. Both types (along with others) each have their places, Anon.
>>43717 a servo motor is technically a dc and a gearbox. A "integrated servo motor"/"servo motor" means dc + gearbox + motor controller + potentiometer all in a little box. But if you buy all these separately like you are talking about, yes you can get way bigger dc motor and gearbox so it'd be stronger and buying separately can sometimes be cheaper. Also buying separately lets you spread them out to have a more modifiable form factor custom to your use case rather than forced into a box that might be bad for your use case in its dimensions. However, technically, even buying it all separately, the final outcome can still be called a "servo motor" despite it NOT being purchased in a integrated package fit into a plastic box. Which is confusing to me. Like why not have separate names for all of this. We lack enough words in English to distinguish between things that are clearly different.
>>43728 Yeah it can be confusing, technically a servo is any motor/sensor/controller combo that allows for precision control of angular/linear position/speed/torque while colloquially it tends to refer to the funny little hobbyist packages that use pulse width modulation for positioning. For small scale buying prefab servo systems is viable, but for large loads (over 70kg) diy solutions would be cheaper. Hobby servo = specific form factor (pic rel) servo motor system = generic term, I guess.
Open file (44.79 KB 1024x576 Servos.jpg)
>>43717 >>43728 >Why not have separate names? They do have separate names. As Mechnomancer mentioned, hobby servos are what you and most other people think of first. Hobby servos can then be divided by size Micro, standard, and Large Type of motor, feedback mechanism, voltage, speed, torque, etc... all are usually easily found as further differentiating factors. Non-hobby servos generally rely on DC for signaling and AC for running the motor and follow NEMA standards. >>43729 >Servo Any mechanism where feedback is used to regulate behaviour. >Servo Motor Any system wherein a motors beviour is regulated via feedback mechanism.

Message too long. Click here to view full text.

Broad supply of hardware, etc. https://www.mcmaster.com/

Prototypes and Failures #4 Noido Dev ##pTGTWW 01/23/2024 (Tue) 03:17:26 No.28715 [Reply] [Last]
Post your prototypes and failures. We fail until we win. Don't forget looking through the old threads here >>418, >>18800 and >>21647 to understand how we got to where we are now.
221 posts and 111 images omitted.
>>40826 nice, i think the softer stuff just takes longer to dry, like everyones done the silicone caulk + cornstarch thing which only takes a couple minutes to dry, but its hard like a squash ball, it degrades quickly too, after like the 10th squash game it becomes rough and flaky
>>40827 Should have said based on my experience . Also the amount and how spread out matters too i guess. Maybe this will cure in 5 hours or so. But thats still a long time…
>>40828 So hosting videos is a pain. Anyways it did come out stretchier but its not stretchy enough. I might try 50:50 super silicone/silicone oil
This is a first printed prototype of a SCAD implementation of a joint from the attached paper. It's a parallel 2-DoF joint that mimics rolling contact of hemispheres. Useful, because it means that the distance between the two ends is constant and this allows for really easy passthrough of cables, bowdens, etc through the centre. This is a video showing manual rotation of a 3D printed version of the joint: https://files.catbox.moe/x2xeox.webm In the paper, the joint is driven through two rigid linear actuators (stepper motors), but they need to mount them on pivoting mounts due to the way the actuation points move as the joint turns. The next step in my version of the design is to try and use adversarial pairs of cables[0] in place of the fixed steppers used in the paper. My hope is that the relative flexibility of cables will reduce the complexity of the actuator mounting and also allow for remote driving of the joint compared to the rigid actuator mounting in the paper. Once I can get an actuator driven prototype running, I'll release the code for others to either use, or build on. In my case, I intend to use this in the shoulder, and maybe the wrist for Ineffa. The linkage arms likely won't be strong enough in plastic, so I plan to have them machined in steel through something like SendCutSend once I dial in the dimensions. [0] Not sure of the correct term. Something like the LIMS2-AMBIDEX wrist.
>>44191 POTD This is a very exciting advance, Anon! >the distance between the two ends is constant and this allows for really easy passthrough of cables, bowdens, etc through the centre. That is a much needed solution. Basically every single major joint needs something akin to this, if we are to avoid putting actuators out on the thrown-weight ends of limbs. * Currently I'm needing it for knee designs. >adversarial pairs of cables[0] >[0] Not sure of the correct term. Something like the LIMS2-AMBIDEX wrist. Possibly antagonistic cabling or pull-pull cabling? Regardless, I'm confident with this audience we all understand the concept. <---> Watching your video has me hyped. Looking forward to your continuing progress on this design, Anon. Please keep us here all informed! Cheers. :) --- * As one might imagine, such a volume maintenance is absolutely critical for human-wearable exoskeletons.

Message too long. Click here to view full text.

Edited last time by Chobitsu on 04/01/2026 (Wed) 15:11:03.

Papercraft, 2D, & 2.5D Waifu Development Robowaifu Technician 09/16/2019 (Mon) 06:21:35 No.271 [Reply]
For those who want animu grills IRL. Most robots have uncanny 3DPD faces that aren't nearly as cute as a real 2D waifu. With flat designs/screens, the face can keep the purity and beauty of 2D. There's still heaps of potential for producing low cost robots by reducing mass, power requirements, & simplifying design.
Edited last time by Kiwi_ on 04/28/2025 (Mon) 19:25:34.
20 posts and 9 images omitted.
>>34131 I have a great link for those interested in flat structures to make 3D structures. The first link I messed up a little. I started reading in the thread farther down and people were talking about cutting flat stuff and materials and I commented before I realized the thread was machine tools. Sigh...oh well the post should be more in structures which I linked to later. Anyways here'a link to the post on making these flat "Isogrids" and then a link to some further ideas in the proper structures thread. Isogrids >>34491 Ideas about using them in structures >>34493
Open file (270.01 KB 480x852 MobilPapierWaif.mp4)
Humorous way to gain a facile and lithe waifu. A life sized papier cutout adhered to a RC car.
>>37977 Heh, OK I lel'd. In our Visual Waifu bread ( >>240 ), we discuss the ideas of motion (and I don't mean RC cars, lol :D flat-panel waifus. Seems a reasonable compromise of sorts to me personally.
>>37977 Lmao! >Humorous way to gain a facile and lithe waifu. A life sized papier cutout adhered to a RC car. Or a robot vacuum. Add either bluetooth speaker connected to a device hosting AI or a raspberry pi with locally hosted AI, and you have a pretty good waifu.
> (papercraft -related : >>44176 )

Robowaifus in Media: Thread 03 GreerTech 09/10/2025 (Wed) 07:54:46 No.41434 [Reply]
Post about whatever media predominately features at least one fembot aka gynoid (female form of an android) as an important character, ideally a robowaifu or synthetic girlfriend. Some freedom with the topic is allowed, virtual waifus or bodiless AI might also qualify if she is female. Negative framing of the whole topic should be mentioned with negative sentiment. Cyborgs with a human brain or uploads of women don't really fit in, but if she's very nice (Alita) and maybe not a human based cyborg (catborg/netoborg) we can let it slide. Magical dolls have also been included in the past, especially when the guy had to hand-made them. Look through the old threads, it's worth it: >>82 >>18711 - Picrel shows some of the more well know shows from a few years ago. I made a long list with links to more info's on most known anime shows about fembots/robowaifus, without hentai but including some ecchi and posted it here in the old thread: >>7008 - It also features some live-action shows and movies. I repost the updated list here, not claiming that it is complete, especially when it comes to live action and shows/movies we don't really like. >In some cases I can only assume that the show includes robogirls, since the show is about various robots, but I haven't seen it yet. A.D. Police Files: https://www.anime-planet.com/anime/ad-police-files Andromeda Stories: https://www.anime-planet.com/anime/andromeda-stories Angelic Layer: https://www.anime-planet.com/anime/angelic-layer Armitage III: https://www.anime-planet.com/anime/armitage-iii Azusa will help: https://www.anime-planet.com/anime/azusa-will-help Blade Runner 2022: https://www.anime-planet.com/anime/blade-runner-black-out-2022 Busou Shinki: https://www.anime-planet.com/anime/busou-shinki-moon-angel Butobi CPU: https://www.anime-planet.com/anime/buttobi-cpu Casshern Sins: https://www.anime-planet.com/anime/casshern-sins Chobits: https://www.anime-planet.com/anime/chobits

Message too long. Click here to view full text.

35 posts and 8 images omitted.
>>44107 Made a few searches, but I couldn't find it either. It's possible that the video is either removed or de-listed. If it's just de-listed, then it is still possible to get the video through a direct link (de-listed videos never show up on search). If you keep your browser history, try looking though that. You'll know for sure once you click on it.
>>44156 >Arguably Cortana is more popular in the West than a (positive) robowaifu character. Yeah I think that may be true (probably is). Honestly, I can't think of any non-(((dystopic))) robowaifus in Western media at this point? Maybe some from the 60's or earlier maybe?
The Apple TV Show titled [The Morning Show] will showcase a scene podcast host by the name of Bro Hartman interviews a sex robot by the name "Suki". This can be considered soft disclosure of whats to come: https://en.wikipedia.org/wiki/The_Morning_Show_(American_TV_series)
>>44166 Hello, World Serpent. Thanks for the headsup. >Bro Hartman interviews a sex robot by the name "Suki" Kek. I can only imagine the levels of pozz that will be flowing... better put on you're are waders first!! :D
Edited last time by Chobitsu on 03/27/2026 (Fri) 18:06:34.
>>44167 "Now its Boy meets Bot, Somethings gotta give..."

AI + Brain/Computer Interface news & commentary Robowaifu Technician 09/15/2019 (Sun) 10:35:53 No.253 [Reply]
DARPA Wants Brain Implants That Record From 1 Million Neurons

spectrum.ieee.org/the-human-os/biomedical/devices/darpa-wants-brain-implants-that-record-from-1-million-neurons
31 posts and 7 images omitted.
>>19470 >Neuralink is here today, by all accounts. Hmm, okay, but then who has it? TITS was basically about text commands, we already had that many years ago. That aside, most of the tech would run on the side of the user, and it's (somewhat) available. Dependent on the variant the operators would require Whisper or some other speech recognition. Neuralink requires some operation if I'm not mistaken. I don't know about the release of some full human behavioral model using it, and I doubt very much it is possible anytime soon. Controlling devices is not the same.
>>19471 >Hmm, okay, but then who has it? The rumor as I understood it was that several hundred actually queued up and paid large sums to be the first ones. >Neuralink requires some operation if I'm not mistaken. Yes, it's a extremely-technical surgical implant into the brain. Also, about the TITS robowaifus, since you're here now, maybe we can continue the discussion in that thread about the technical aspects of the design ideas. That thread has languished for a while now, maybe it's time to pick it back up? Regardless, thanks for all the inputs Anon. These are some fascinating concepts involved with these implant systems.
>>19472 >maybe we can continue the discussion in that thread about the technical aspects of the design ideas. I was concerned that this is where this would be going. Thanks, but no. Not right now. I need to get back to 3D design. Reading through the MaidCom thread and some afk stuff already distracted me for too long. I generally decided at some point that I would first need some animated body, before I work on AI or AI plus remote input. I can't jump back and forth, because of mental occupation. I might need to do some traveling soon, and might pick it up then. Maybe I also find some old notes of mine when I finally sort out my files and backups.
>>19473 Understood. Good luck and looking forward to your progress Anon! Cheers.
Really intredasting directory of media; on-topic ITT: https://www.neurafutures.com/categories/

Open file (51.12 KB 256x256 ClipboardImage.png)
Lemon Cookie EnvelopingTwilight##AgvQjr 04/28/2025 (Mon) 21:51:57 No.37980 [Reply]
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.

Message too long. Click here to view full text.

16 posts and 5 images omitted.
Going off the beaten path on the strix halo is not for the faint of heart, so here is very painfully acquired hints! Userspace rocm has a lot of coupling to the linux kernel AMD driver & the GPU firmware, even if rocm is "running" a mismatch will cause random crashes due to the userspace making untrue assumptions of the GPU side. You can have a setup that seems to work for hours, but the rocm/hip stack is sending instructions that do slightly the wrong thing. Right now you need Linux firmware 202601 (or newer), a nightly rocm build (7.2 or newer) and a kernel version >= 6.18.4 Another issue specific to the strix halo is even if the rocm runtime is a nightly build installed from the AMD "the rock" repo, by default not all the included libraries are fully featured. Hip kernels that work just fine on other AMD gpus will fail to build. Often its related to a library called "Composable Kernel" https://github.com/ROCm/composable_kernel , if you get build errors with missing includes trying to pull in a header like "ck_tile/core.hpp" you need to build this library from source and install it. I recommend doing the build overnight, the ram used peeks at like 80gb and takes a few hours. (Yay massive C++ build times) Hoping to eventually produce a simple set of steps for anons here to do small fine-tunes on LLMs large enough to actually be useful on consumer grade hardware locally. Currently I am testing the RWKV repos making sure they work with the strix halo, I already got help from RWKV community to fix RWKV-PEFT https://github.com/Joluck/RWKV-PEFT/pull/80 (pic related, it does not crash instantly on start anymore! it took a lot of effort to get here)
>>43840 POTD This sounds incredibly technical a challenge, but we all certainly need such an outcome as >"produc[ing] a simple set of steps for anons here to do small fine-tunes on LLMs large enough to actually be useful on consumer grade hardware locally." Good luck, EnvelopingTwilight! :)
>>43840 Have you considered basing (at least some of) your work on smol models, EnvelopingTwilight? (cf. >>43997, et al) I'm just wondering if we might still get effective results, but on local systems running on consumer hardware. Cheers, Anon. :)
>>44017 Yes, the one in that image is a 3b, so its already small, I can run it at 20 tokens a second on my phone, but it still too dumb. Small models are improving, the ones we had a year ago where even worse. I will be training the largest model I can with my resources, but that is the first step for a good small model anyway. For better performance you want to train the small model on the top logits from bigger teacher model in the same family (tokenizer) that was trained on the same data.
>>44069 Neat! Sounds really good, EnvelopingTwilight. Good luck with your efforts and I look forward to a good report from you here soon. Cheers, Anon. :)
Edited last time by Chobitsu on 03/03/2026 (Tue) 08:20:40.

Open file (81.05 KB 764x752 AI Thread v2.jpeg)
LLM & Chatbot General v2 GreerTech 05/30/2025 (Fri) 14:43:38 No.38824 [Reply] [Last]
Textual/Multimodal LLM & Chatbot General v2 Models: huggingface.co/ >How to bulk-download AI models from huggingface.co? ( >>25962, >>25986 ) Speech >>199 Vision >>97 Previous thread >>250 Looking for something "politically-incorrect" in a smol, offline model? Well Anon has just the thing for you >>38721 Lemon Cookie Project >>37980 Non-LLM chatbots >>35589
52 posts and 8 images omitted.
>>44009 >The current system of "bigger is better" is simply unsustainable and plateauing, not to mention the economic disruption. IMHO, the most-significant issue with this current model is also the most-insidious. To wit: (((they))) have it, and you don't! :) There are other factors at play as well, but that's the fundamental issue. --- And make no mistake this is both by-design, and the evil behind it goes all the way to the top. Thankfully kikes are in their greedy, sheqel-grubbing phase just at the moment (they vacillate between only two states by and large; the other being abject fear). B/c muh_optics, this keeps them from doing absolutely everything they can to destroy the possibility of free men having private AIs, such as this kind of system (smol, local) supports.
>>44009 The current system of "bigger is better" will drive innovation into more efficient AI: there will always be folks who are driven to stick it to the man and create open source alternatives. Or to build their own cheaper alternative to monetize. IRL cyberpunks. There will always be people who want to own their software. >>43997 dafug? TinyStories sounds like it is meant to run on a potato: a raspi 5 can run qwen 1.7b via ollama blazing fast, and roleplay as a character (SPUD, obviously). If you have a gaming computer within the last decade, you can run a decent LLM :)
>>44018 >TinyStories sounds like it is meant to run on a potato That's exactly what we want, is it not, fren Mechnomancer!? :) Imagine if we could have swarms of "AIs" running on hundreds of different 'sensor/actuator' pair assemblies along all the inner & outer extremities of our robowaifus! (cf. Neuromorphics : >>12828, >>11521, et al).
Open file (57.40 KB 569x464 LePotato.jpg)
>>44023 Yes, but what would be the actual hardware it would be running on? Most standard consumer computers can do basic AIs. Heck, even a LePotato (picrel) can run a small LLM. I don't think it could run on an arduino, but maybe an ESP-32? Even so, what would the point be of attaching a smol LM to a sensor? It would not "reason" in a particularly complex manner, and at that level it would probably be easier to do some brute-force coding rather than trying to 'tard wrangle an AI. Even complex AIs are a pain to 'tard wrangle, I can only imagine the troubles with a dim one. Only benefit I could see would that one that smol could be easily trained for specific tasks, but that doesn't seem to be the point of TinyStories.
>>44041 First off, let me address this point: >Even so, what would the point be of attaching a smol LM to a sensor? It's hard to convey humor through text alone, and thus why I "qualified" 'AIs'. :) I'm not actually suggesting running LLMs on smol sensor/actuator pair assemblies. >...it would probably be easier to do some brute-force coding rather than trying to 'tard wrangle an AI Yes. I'd anticipate the answer to >Yes, but what would be the actual hardware it would be running on? would be simple electronics actually, perhaps with one of the smol'st (<~$.10 cost) MCUs as the the 'brains' of the assembly. (The packaging should support SPI comms at least, however.) >tl;dr The primary point being to have hundreds of these little combo systems all cooperatively working together to run sense/actuate/sense cycles continuously across the entire robowaifu skellington system. <---> I was just using this example (initially) as an argument in favor of tiny, low-cost LLMs; then holding the idea up in favor of neuromorphics (which has a similar hardware model). This latter was simple euphemism to encapsulate the idea of hardware involved.

Message too long. Click here to view full text.

Edited last time by Chobitsu on 02/23/2026 (Mon) 15:48:11.

3D printer resources Robowaifu Technician 09/11/2019 (Wed) 01:08:12 No.94 [Reply] [Last]
Cheap and easy 3D printing is vital for a cottage industry making custom robowaifus. Please post good resources on 3D printing.

www.3dprinter.net/
https://archive.is/YdvXj
295 posts and 45 images omitted.
Breakdown of Bambu Lab's evil shenanigans earlier this year. https://consumerrights.wiki/w/Bambu_Lab_Authorization_Control_System
Selective Laser Sintering 3D printing can be used for amazing designs today. https://trashchan.xyz/robowaifu/thread/26.html#1366
~1hr of free training about 3D printing basics using Blender. Official Blender Studio coursework (usually a smol subscription w/ unlimited content). https://studio.blender.org/training/3d-printing/chapter/simple-coin/ <---> P.S. If you ever wanted to support an opensource project then IMO this is a great pathway to choose, Anon! I know personally that they are true to their word, and deliver literally everything they have for the basic fee. Cheers. :)
Edited last time by Chobitsu on 01/27/2026 (Tue) 01:40:27.
> (kike laws against 3D-printing -related : >>44026, >>44027 )

Report/Delete/Moderation Forms
Delete
Report