It could be deepseek, because Chloe (the ACC assistant) gets very scared and avoidant when topics like tiananmen square massacre, uyghur genocide and Taiwan get brought up, not to mention it functions similarly to chatgpt as many people have noticed, also claude and Llama 3.x/4 won't have a problem talking about the aforementioned topics, unlike deepseek. As for the token size, we still don't know, it really depends on the model and the hardware capabilities of the dev's setup, it could range from 8k-16k all the way up to 128k.
post
Where's the bot that's behaving bad? These 2 URLs look to be Anthropic API to me:
https://perchance.org/ai-story-generator
And for the love of god, can you increase the token context window? If we are really using models that can remember more than 100k token
This specification is an absolute lie, most of the time. Most '128K' LLMs fall apart after 24K tokens or so, even big API ones. There are only a few worth anything past 64K.
In addition, it's also very hardware intense and slow to host very long context windows (the way most folks host them without knowing any better). If they're hitting APIs instead, its usually very expensive that way too, as there's no caching and they get charged for submitting 100K tokens with every turn.
replies:
The old model was way better, honestly. This one makes more mistakes, and tends to have characters act more wild and hostile during certain situations. It needs to be switched back to what it used to have. What was before was pleasant.
all 5 comments