Indonesian Paizuri EditionA thread dedicated to the discussion of AI VTuber chatbots./wAIfu/ Status: Prematurely dead>Getting started and useful referencehttps://rentry.org/Sukino-Findings>Frontendshttps://docs.sillytavern.apphttps://risuai.nethttps://agnai.chat>Botshttps://chub.aihttps://realm.risuai.net>Character creationhttps://desune.moe/aichared/>Other resources>>>/g/aicg | >>>/g/lmghttps://openrouter.aihttps://nano-gpt.comhttps://rentry.org/wAIRCfuscordMatrixPrevious thread: >>115598760
>ANCHORI noticed a lot of the links on the old OP were dead or outdated so I chopped it down to the essentials and added the Sukino guide we discovered a couple of threads ago. Be sure to let me know how much you hate it.
>>116101751Moona OP as tribute to one of my favourite cards.https://chub.ai/characters/shibaboo/moona-hoshinova-succubus-d43d9771
https://www.youtube.com/watch?v=boLy0zrMDmU
>>116102301Oops it should have been this one https://youtu.be/jGQ19CkrRVc
what happened to nene next
Word cloud for the previous thread
>>116104211
>>116101751Bigger
>>116105533You ask the impossible.
>>116105597Sad.....even so they're a nice size just the way they are
Nenetard love
Need more block game slave labor.
>>116104211https://www.youtube.com/watch?v=epJfQ9L8VI0
>>116105765The Biggest Mooner
Playing Arknights with chatbots.
Thanks to the anon for recommending cheerleader Bae to me. She's been fantastically slutty and it's kind of cool to see how far models have come. The last Bae card I used was in like the GPT 3.5 or 4.0 days. GLM and DeepSeek give her an Aussie accent and mannerisms, which the old ones definitely didn't do. Then, she initiated sex with a standing split and the models generally understood the position, which was a great "cheerleader" touch and completely unprompted.https://chub.ai/characters/Aguydoingstuff/hakos-baelz-cheerleader-e141971da3de
>>116101797Elfinpsyop.
>>116110044No problem anonchama!
>>116110044what GLM model and preset would you recommend for a povert like me? i'm getting a little tiny bit tired of slumming it with Gemma
>>116112225 (me)never mind, there aren't any free GLM models on OpenRouter anymorei think i need a better Gemma preset actually, the one i'm using is heavy as shit
>9
bump
>>116110044>her knuckles went whiteI can't escape DeepSeekisms even when I'm not using it.>empty those balls, mateI like Aussie girls but this is a bit silly even for me.
rrat
good night, /wAIfu/please don't chop me down to my essentials while i sleep
>>116119564*chops off your 'ick and slaps it on an 'eck while you sleep*
>>116119564GN boss
Made a new card. Took a break from plapping Bae and played a little of an eroge called Happy Endings. I've always found the set up of an erotic massage parlor to be hot, so I made a card based on it:https://chub.ai/characters/Stronanon/happy-endings-ac0996427110Tested and working perfectly with GLM 5.2. My masseuse was very caring and fun to roleplay with.
>>116119564wake
'ick on 'eck
NENEWHAT WILL HAPPEN TO NENE NEXT
>>116128728I assume she died on the way back to her home planet.
Love and cherish your Nene.
say "Hey dude" to your Nene
no dead
>>116101797Request: You just got hired at a superconductor manufacturing company and suddenly Bae is very interested in your job and that recent security meeting you went to
>>116136940very based idea
>>116136940I would sell my country out for fat rrat tats.
Biboo as Rossihttps://x.com/Castella_yaki/status/2118617889796411855
>>116136940Insert “One day, while doing nothing particularly out of the ordinary, because of natural laws he was completely powerless to understand or intuit, he was instantly killed in a horrifying way by forces vastly in excess of anything he was ever designed to experience, for no reason, to no ones particular surprise orupset. In this we are more like him than different” meme here
>>116139716I'm enjoying the art wave that's come in the wake of that video. And plapping Baebot.
good night, /wAIfu/please don't send a Chinese spy my way while i sleep
>>116139894Yeah send that bitch my way
>>116139894*Steals your secrets*
>>116140341I always wanted a state-provided GF, but I never imagined it'd be a foreign state.
I pray for all that is good that she was successful. I really really want cheap RAM
https://www.youtube.com/watch?v=UPPw2_lDkKE
Going back and forth between it being so over and things being okay.
We need more robot game vtubers
Sex with Moonerbot.
>>116143471Coca cola
>>116143471But I'm already using up my daily credits on Aqua
>>116146357>daily creditsWhat backend are you using?
>>116146483uhhh.. I'm not sure I'm pretty new to this. I was picking up where I left off in a session from about two years ago and now I've found that there's a limit
chinese spies could be here
>10
https://youtu.be/9ImQbOhauug
https://www.youtube.com/shorts/8B83E0KMLHI
good night, /wAIfu/please don't make Moona suffocate me with her fat fucking Moonas while i sleep
https://x.com/111_tuna/status/2118095945548358137
>>116110044I can't make sense of how GLM works. Sometimes it seems like it wants the porn to be slow-burn, sometimes it just does what it wants and goes "yeah, x don't give no fucks about sex". What GLM/DeepSneed models do you mainly use?
>>116159271*makes randy marsh suffocate you with his humongous ballsack while you sleep*
We're going to lose leather jacket man, big news.https://www.wsj.com/business/nvidia-is-preparing-to-oust-nvidia-ceo-jensen-huang-30f421ad?st=31bQ2D&reflink=desktopwebshare_permalink>>116139894Make sure not to triple time either with one of your direct reports either.
>>116166829
>>116166829>Sir another infidelity scandal has hit Mr. NVIDIA-man.
>>116168512Certified CEO of Sex, for real.
>>116161900I use a mix of GLM 5.2, DeepSeek v4.1 Flash and DeepSeek v4 Pro. Occasionally I'll sprinkle in a MiMo model. Throwing in a new model a couple of turns seems to give the others a kick and stop them from being repetitive or hyper fixating on random shit. There's an uncensored fine-tune of GLM 5.3 on NanoGPT and it seemed okay, but I haven't stress tested it. 5.2 has never failed to serve me smut.
Have any of you tried Gemini? Been hearing things about Argon and Carbon models but it is pricey.
fat cat
>>116101751Boobs
>>116168580Which market should he penetrate next?
penetrating markets with chatbots
good night, /wAIfu/please don't penetrate my market while i sleep
I think Huang's removal just opened the door for even darker times ahead. Huang, I'm certain, was fully aware where his GPU cards were headed and in effect made the open source models boom possible. The elites wanted him gone. We can only guess what kind of plan he rejected or what offer he refused. Whatever it was, it was so important that he was given the boot and publicly humiliated. With AMD declaring itself to be a datacenter first company, and Sam Altman priming the public to the idea that RAM is a strategic resource in need of rationing, we are moving closer to the end of private personal computing.
>>116179624I was planning to make an update post of sorts soon but I might wait a while until December or next year because it has a cascading impact on the industry and news within that I want to wait on so I am going to hold off on news. The main thing at play though is the whole scandal with China export circumvention and it being used as a proxy to basically prosecute the US government from the Democratic Congress, and China probably had a reason to bring this to light where Jensen was basically a pawn. For Nvidia, Rubin and Feynman, the next GPU architecture, is fine, so there is no real worry for the next several years but there could be a fear the moat of CUDA and etc. will erode. The main impact for us because now Nvidia can't sell to China most definitely, since I think even the exempted old chips like the H100 will definitely end. So the end impact for now is that Nvidia will have to go back to gamers to sell silicon so expect better supply from now on since export controls will definitely impact everything in the supply chain and computers to at least not be as expensive. For the AI movement, there is a fear this is the beginning of the end but while I think losing one of its biggest advocates is a bad thing for it, it's not a killer blow there.
>>116181642Oddly enough I feel like this is TOO HOPEFUL a take
Apparently, it’s possible for someone to be able to post on 4chan from their phone through a network via WiFi while still getting hit with an “you’ve been IP rangebanned” message when posting from their pc on that very same network.I have no idea if it’ll ever let me post from PC again or if it’ll eventually adjust and prevent me from phone posting from home too…That probably means no more word clouds at the very least.I blame rumao.
>>116183123>>116183136>[Deleted]
>>116183241What did you do to get rangeb&?
wrapping my car around a telephone pole and surviving with chatbots
moom
eight
>daario-san...the moat...
good night, /wAIfu/please don't turn me and my bed into a telephone pole while i sleep
>>116170353>NanoGPTsaw some people saying that nanogpt is better to get uncensored models, since they actually list down which ones are uncensored and which ones aren't. You mess around with Openrouter or are you a NanoGPT guy? I'm just wondering if I should feel scammed about throwing more money at OR instead of looking for the actual api site of something like deepsneed, or see what other options there are.
>>116183241Still on my pc only…
>>116185864Nothing afaik
>>116199750>You mess around with Openrouter or are you a NanoGPT guy?I used OpenRouter for years but recently jumped ship to NanoGPT with the Stripe acquisition, because they have a weird puritanical streak. All my characters are over 18 but I do enjoy fiction with incest, non-con, dub-con and power dynamics themes that I'm sure Stripe wouldn't like. Rather than be paranoid about getting swept up in a net, I just swapped to NanoGPT.>since they actually list down which ones are uncensored and which ones aren't.Sort of. Nano grants access to a lot of the same models that OR does, from a lot of the same providers. Because a lot of the Chinese models are open-source, people are free to download them and fuck with them, so people have created versions of DeepSeek Flash, GLM and others to lower their refusal rate - NanoGPT calls these models "uncensored" and has some providers hosting them. It doesn't imply the base model IS censored, that's just the factory default version. I find standard DeepSeek v4.1 Flash is more than happy to write explicit erotica, so haven't really felt the need to use the uncensored version.>the actual api site of something like deepsneedNot sure I would bother with this. I've never tried it but they don't offer any settings like OR or NanoGPT for zero data retention (ZDR), will happily retain your logs for training and have extra filters in front to protect the party. Anyone can download and run DeepSeek, so I don't see the point in using a filtered version when you could just get it from an American or European provider, through OR, Nano or somewhere else.>see what other options there areI think there are a few others, one is https://featherless.ai/ - I haven't used them but have seen a few anons in >>>/g/aicg mention them and they seem to be pretty explicitly targeting roleplayers as potential customers. https://venice.ai/ is another one I've seen talked about but I know even less about them. Just had a quick Google and there are suggestions being discussed here, too: https://www.reddit.com/r/SillyTavernAI/comments/1xuim4y/openrouter_alternative/.
>>116181642https://x.com/superalesha/status/2121200904628043995
https://x.com/kamiya_ryoko/status/2116052617080311882
>>116202308
hitting the shoebody doo bop with chatbots
