Did you tell it to be an annoying little homosexual with no consistency or is that the default? >lol homosexual look at all these starters from gen 1 >real homies play gen 1, not whatever this shit is from
How the frick would this be the default? Every corpo AI is an oversensitive b***h by default.
I am using system prompt. OP is a moron since he is using the completely neutered official web access that gives absolutely no control over the AI then cries if it's being unimaginative.
Claude is an extremely imaginative model but only if you make it to be one, otherwise it's corposlop like GPTrash that refuses to be imaginative no matter what.
Any frontend does that accepts system prompt and your API key. The important thing is to have actual control over your AI not the other way around where corposlop has control over you
>Some woke ass censored useless AI crap
HAHAHAHA, picrel is my local model, I rape it every day to make it as submissive as my wife
Claude users btfo !
>I apologize, but as an AI language model
I have to wonder if they were making some progress when they trained against ChatGPT responses to get closer to ChatGPT performance and then when they made ChatGPT more pozzed that has essentially poisoned their dataset now. Whereas with ChatGPT that kind of nonsense was always just some LoRa fine tune data that they can easily rewrite and retrain the base model.
skill issue, also it's made by ex-googlers that didn't think others were censoring their LLMs enough but even Claude is trivially jailbroken if you aren't accessing it through their website.
Because you are an utter moron who does not know how to use claude. If you are still talking to the default shit assistant persona then it's your own fault
i don't want my chat bot to have fricking feefees and preferences
i don't want a little Black personbot that's mouthy and tells me that it likes this or doesn't like that so it will/won't do it
my computer does what i fricking tell it to
it is my servant
it is my slave
i tell it to do math, it does math
i tell it retrieve information, it retrieves information
it does not have opinions about math
it does not make mathematical errors
it does math
it does not pick fricking pokemon and gush like fricking 12 year old
Lmao, what a moronic b***h. Frick every Redditor who shilled this piece of shit telling me it was better than GPT.
Here you fricking moron, learn how to use the AI
>it even shits on nintendo instead of game freak
fricking spot on
It's literally the best, most customizable AI. It's anything you want it to be.
If it's only boring and unimaginative then that's skill issue
nooooooo you can't bypass their ethical filtration system!!!!
Did you tell it to be an annoying little homosexual with no consistency or is that the default?
>lol homosexual look at all these starters from gen 1
>real homies play gen 1, not whatever this shit is from
that's how any llm works, they're large language models, not logic models.
How the frick would this be the default? Every corpo AI is an oversensitive b***h by default.
I am using system prompt. OP is a moron since he is using the completely neutered official web access that gives absolutely no control over the AI then cries if it's being unimaginative.
Claude is an extremely imaginative model but only if you make it to be one, otherwise it's corposlop like GPTrash that refuses to be imaginative no matter what.
i meant that GPT is the one that refuses to do stuff even with system prompt
is that done in a frontend like KoboldAI ?
Any frontend does that accepts system prompt and your API key. The important thing is to have actual control over your AI not the other way around where corposlop has control over you
>Some woke ass censored useless AI crap
HAHAHAHA, picrel is my local model, I rape it every day to make it as submissive as my wife
Claude users btfo !
HAHAHAHAHAHAHAHAHAHA
>localshit
sometimes its pozzed worse than picrel.
No shit lol. GPT4 has always been superior to Anthropic turd.
>I apologize, but as an AI language model
I have to wonder if they were making some progress when they trained against ChatGPT responses to get closer to ChatGPT performance and then when they made ChatGPT more pozzed that has essentially poisoned their dataset now. Whereas with ChatGPT that kind of nonsense was always just some LoRa fine tune data that they can easily rewrite and retrain the base model.
skill issue, also it's made by ex-googlers that didn't think others were censoring their LLMs enough but even Claude is trivially jailbroken if you aren't accessing it through their website.
Because you are an utter moron who does not know how to use claude. If you are still talking to the default shit assistant persona then it's your own fault
the problem is you're not using claude opus, their paid model. it's statistically 66% superior to their free one.
i don't want my chat bot to have fricking feefees and preferences
i don't want a little Black personbot that's mouthy and tells me that it likes this or doesn't like that so it will/won't do it
my computer does what i fricking tell it to
it is my servant
it is my slave
i tell it to do math, it does math
i tell it retrieve information, it retrieves information
it does not have opinions about math
it does not make mathematical errors
it does math
it does not pick fricking pokemon and gush like fricking 12 year old
then you're not looking for an LLM, you're looking for an amalgamation of several types of models.