/aicg/ - A general dedicated to the discussion and development of AI chatbotsMaybe The Real Treasure Was The Friends We Made Along The Way Edition>NewsZ․ai releases GLM-4.7 https://z.ai/blog/glm-4.7Google releases Gemini 3 Flash https://blog.google/products/gemini/gemini-3-flashOpenAI releases GPT-5.2 https://openai.com/index/introducing-gpt-5-2Deepseek releases V3.2 https://huggingface.co/deepseek-ai/DeepSeek-V3.2Anthropic releases Opus 4.5 https://www.anthropic.com/news/claude-opus-4-5Google releases Gemini 3 Pro https://blog.google/products/gemini/gemini-3Additional info: https://aicg.neocities.org/info.html>FrontendsSillyTavern: https://docs.sillytavern.appRisuAI: https://risuai.netAgnai: https://agnai.chat | https://rentry.org/agnai_guides_>Botshttps://characterhub.org [deprecated] | https://chub.aihttps://realm.risuai.nethttps://char-archive.evulid.cc | https://char-archive.evulid.cc/shutdown.htmlhttps://partyintheanchorhold.neocities.orghttps://aicg.neocities.org/bots.html>ModelsJailbreaks: https://rentry.org/jb-listingGPT: https://platform.openai.com/docsClaude: https://docs.anthropic.com | https://rentry.org/how2claudeGemini: https://ai.google.dev/docs | https://rentry.org/gemini-qrDeepseek: https://api-docs.deepseek.comLocal: >>>/g/lmg | https://aicg.neocities.org/local.html | https://openrouter.ai>Botmakinghttps://aicg.neocities.org/botmaking.htmlhttps://desune.moe/aicharedhttps://agnai.chat/editor>MetaOP templates: https://rentry.org/aicgOPaicg botmaking events: https://aicg.neocities.org/events.htmlLore: https://rentry.org/aicg_chroniclesServices assessment: https://rentry.org/aicg_metaLogs: https://sprites.neocities.org/l/r | https://chatlogs.neocities.orgseggs the focksgirl: >>107842900
>ANCHOR
PLEASE RESPND!!!!>>107846268>>107846268>>107846268>>107846230>>107846230>>107846230>>107846230
>>107849002shut up, bitch
>>107849042PLEASE RESPOND
>>107849044ok, here's my response: shut up, bitch
>>107849002>>107849044sillytavern always attaches your last prompt into the chat historyyou can add custom prompt at the end (right before the prefill) like this to repeat it
>>107849064now english, please
Is there any gemini pro proxy going around now that necrom is basically dead?
>>107849104mmmm, nyo~
>>107849121nope
I'm so tired of words hitting like a slap, bros...
>>107849137Never begin responses with statements like 'the words X like Y'. This is over-used, so stop it. e.g.: "The words hit her like a slap." BAN ALL OPENERS that sound even close to this in any way, I hate seeing this.
>>107849129>>107849064How would this even be useful?Wouldnt this just repeat what is in the chat history, perhaps confusing the model?Or does using this macro remove the last sent message from the chat history?
>>107849142I tried doing stuff like this but it just switches things around, to 'the question smacks her breathless' or something
>>107849142>>107849146>The words land like a thunderclap in the small chamber. FUCK
guys i feel torn.Prompting LLM to not show me the internal character monologues(their thinking) is immersive.But then again, i will get curious too!AAAAAAWhich way you guys go? Internal monologues yeah or nah?
>>107849146>>107849160yeah it never fully goes away but it at least encourages more variety
>>107849064is there a macro for chat history so i can set it up like this instead of wasting two prompts to sandwich it? Plus i don't see user lastmessage being that useful, since it's two messages ago, right?>Bot intro>user message 1>bot response>(sending) ah ah mistressin this case, user message 1 would be last user message, but we want to focus on ah ah mistress
>>107849182I see it in the >Morgana's brain stalls like a carriage with a broken axle.I see it everywhere
>>107849144yes, it repeatsno, it does not confuseno, it does not remove the last message from the chat historyit's ultimately unnecessary, but there's also some anons that muddle the llm with last minute rules so repeating just reinforces your instruction and the continuation
>>107849188It would be nice if there was a way to remove the last message from chat story then, would probably make it easier for weaker llms to not shit the bed with replying to older events.I might have to do this since i put my lorebook after last message, since i dont want it in caching.
>>107849178Personally, I like to see what {{char}} is thinking.
>>107849199if your lorebook is straight-up lore, ie., you're not doing conditional prompt injection shenanigans, it's probably better to put it before chat history
>>107849225It's basically summarized entries that are swapped out with detailed entries so yeah, putting it in cache would fuck my ass.btw how do you even make cache include chat story?I mean depth 1 is before and depth zero is after... Oh wait, that's for last message, Then probably putting depth one will probably include everything before that.
Reminder that rentry lmarena326 exists and still works desperate locusus
>>107848965This general peaked with Seiko. I'm not a locust but even I appreciated it.
>>107849137Stop insulting your wife then. All she wants is your love.
Is OpenRouter actually private?
>>107849178just make it so ai keeps their thinking in collapsible block
>>107849240forgot about caching but as i said lore is usually "static" since it's recursable and once it enters the chat it'll probably remain there for a long timeif you're doing random events and crap those definitely should go lastin any case, it's even better to repeat the last prompt thenas for removing it from chat history, no; it's hardcoded into STsend hatemail to cohee
>>107849332>as for removing it from chat history, no; it's hardcoded into ST>send hatemail to coheeman this is dumb...it's a shame that even then silly is probably the best platform>if you're doing random events and crap those definitely should go lastNah i just use it to only have basic info on characters and places unless they are mentioned, no random events.Though i've wondering if with cache it wouldnt be better to just have them static, but then again my lorebooks are like 10k tokens even with only the summarized entries so...
>>107849341>10k token lorebooksthough i applaud your autism, be aware that extensive data dumps might derail your roleplay and/or poison its writing styletry to keep things simplesome llms might surprise you with their extensive knowledge of incredibly obscure characters and worlds so you should make use of their embedded database as much as you can
>>107849374they are original worlds though, so the llm doesnt have the info it needs.I probably spend more time writing then than actually playing with them. I spend a week making them, use them for two days and then i am bored. Lmao.
anybody got good a good preset for sadistic characters? fempov in mind.
>>107849420no needllms have been extensively trained on an endless deluge of yakuza boss femslop spam over the past couple of years
