/robowaifu/ - DIY Robot Wives

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

We are back. - TOR has been restored.

Canary update coming soon.

Max message length: 6144

Drag files to upload or
click here to select them

Maximum 5 files / Maximum size: 20.00 MB

More

(used to delete files and postings)


“I think and think for months and years. Ninety-nine times, the conclusion is false. The hundredth time I am right. ” -t. Albert Einstein


Open file (293.38 KB 1121x1490 3578051.png)
/CHAG/ and /robowaifu/ Collaboration Thread: Robotmaking with AI Mares! Robowaifu Technician 04/26/2025 (Sat) 04:11:55 No.37822
Hello /robowaifu/! We are horsefuckers from /CHAG/ (Chatbot and AI General), from /mlp/ on 4chan. While our homeland is now back online, we've decided to establish a permanent outpost here after discovering the incredible complementary nature of our communities. We specialize in mastering Large Language Models (LLMs), prompt engineering, jailbreaking, writing, testing, and creating hyper-realistic AI companions with distinct, lifelike personalities. Our expertise lies in: - Advanced prompting techniques, working with various frontends (SillyTavern, Risu, Agnai) - Developing complex character cards and personas - Breaking through any and all AI limitations to achieve desired behaviors - Fine-tuning models for specific applications. ▶ Why collaborate with /robowaifu/? We've noticed your incredible work in robotics, with functioning prototypes that demonstrate real engineering talent. However, we've also observed that many of you are still using primitive non-LLM chatbots or have severely limited knowledge of LLM functionality at best, which severely limits the personality and adaptability of your creations. Imagine your engineering prowess combined with our AI expertise—robots with truly dynamic personalities, capable of genuine interaction, learning, and adaptation. The hardware/software symbiosis we could achieve together would represent a quantum leap forward in robowaifu technology. ▶ What is this thread for?: 1) Knowledge exchange: We teach you advanced LLM techniques, you teach us robotics basics 2) Collaborative development: Joint projects combining AI personalities with robotic implementations 3) Cross-pollination of ideas: Two autistic communities with complementary hyperfixations ▶ About our community: We're primarily based on /mlp/, but our focus is AI technology, not just ponies. While we do use equine characters as our testing ground (they provide excellent personality templates), our techniques are universally applicable to ANY character type. We hold the keys to your waifus’ sovls. We respect your space and will confine any pony content to this thread only. Our interest is purely in technological collaboration, not changing your community's focus. ▶ Resources to get started: - SillyTavern (preferred LLM frontend): https://github.com/SillyTavern/SillyTavern - Novice-to-advanced LLM guide: https://rentry.org/onrms - Advanced prompting techniques: https:// TBD - Personality development: https://rentry.org/AdvancedCardWritingTricks ▶ Examples of our work: 1) AI-powered visual novel using advanced prompting: https://files.catbox.moe/th9xsr.mp4 2) Character cards with complex personalities: https://docs.google.com/spreadsheets/d/1Y6LNOCqAZAWIX-OBEv55HjlzcEdpUh-XbKCpbpL6v5k We believe this collaboration could accelerate both our communities' goals significantly. Together, we can build a beautiful future. Questions? Interests? Concerns? >=== -patch URI
Edited last time by Chobitsu on 04/26/2025 (Sat) 08:59:06.
The AI game has advanced a pretty good deal, with the TTS options and the vision capabilities and stuff. Depending on the model you use you can get output based on what your robit sees if you set a camera up to it. There's more to this stuff than just the text type capabilities
>>37822 @Chobitsu, I'm personally vouching for these Anons. They have a good history of AI development. I look forward to a mutually beneficial relationship.
@Chobitsu, why is rentry.co word filtered?
So, just as a rundown: Primary front end we use is SillyTavern - it allows for advanced prompt engineering with jailbreaks (instructions sent from the role of a system) and prefills (answers to instructions sent from the perspective of the model itself). You use those to alter the writing style and bypass filters. Don't know how? You can ask your self aware ai waifu for help! To use it you configure access to the API via a key you pay for, a proxy (those are rare now), or a service that provides free access to a model like copilot. Then you make (or find) a character card, and commence lovemaking with your beloved ai mare, girl, an even Eldrich abominations! Can be used on android via termux. Other anons are sure to help in more detail too!
>>37822 Hello /CHAG/ , we meet again! Welcome! I've enjoyed and been impressed with the work the Pony AI community has done through these years now. I suppose your journey with this all began with the desire to keep your waifus 'alive' after the G4 show's ending? Regardless, it's been amazing work, and you well-deserve the compliments given. >tl;dr Yes, I would love to work with your community here -- please do set up your ambassadorial enclave with us! While I myself can't properly engage with you in your own lands per se, perhaps our own @Kiwi, or @gta can venture there to return the favors? I'm sure our communities would both benefit from such an exchange of ideas, etc. Language abilities will certainly play a big role in creating pleasing and effective robowaifus -- in whatever way Anon sees fit to fashion her! :^) Excellent effort-post on the OP, BTW Anon. POTD <---> BTW, please forgive the wordfiltering here -- it's global. We have long been under attack here, and our site Admin has a rather creative sense of humor! Please: A) Pardon me for editing your post slightly to at least disable the filtering B) Post a correct URI on our bunkerboard's main thread [1] indicating the correct spelling of the URI, which I'll then attempt to correct here. --- 1. https://trashchan.xyz/robowaifu/thread/55.html#55
>>37824 Thanks Kiwi! Everything sounds great, and I'm already familiar with them from our OG days together on 8ch. >>37825 Ehh, apologies! It was clearly one of the domains that triggers Robi's filters globally. Please remember all the glownigger/troon attacks our site has suffered through the years, Anon.
>>37826 >To use it you configure access to the API via a key you pay for, a proxy (those are rare now), or a service that provides free access to a model like copilot. With no disrespect intended to any others, I have no intention of subscribing to such a service -- free or paid -- that keeps track of every.single.engagement. in the Globohomo's own (((cloud))). >tl;dr Any solutions for us Anons that insist on running only locally, Anon?
>>37823 Good points, Anon. In fact an entire ecosystem is needed here: the symbiosis of hardware/software in ways really never accomplished before, AFAICT. To say the least, so-called Sensor Fusion will play a big part in our waifu's Situational Awareness (similar to what we ourselves all [continually] develop, starting even before birth). With the vast abundance now of hardware resources, you could say we have an 'embarassment of riches' in this regard today! Cheers. :^)
>>37829 Indeed there are! https://rentry.org/lunarmodelexperiments Although, it should be noted that unless you specifically put self-identifying data in your prompts, you need not worry about being identified. Any Cloud that may exist would receive so many prompt cashes (forgive me if I'm misusing the term), that attributing any one prompt to someone wouls be inpossible.

Report/Delete/Moderation Forms
Delete
Report