Content filter too strict
complete
K
Kai Bogdanski
I experienced that a little tough love towards LLMs works wonders e.g.: calling them "idiot" or using phrases like "why the fuck". Since recently this gets blocked by the content filter.
I can understand that the content filter is necessary to prevent NSFW content and stuff but this is quite too much in my opinion and impedes my freedom of expression in some way.
Sometimes phrases not at all harmful are flagged by the content filter. This is very annoying!
Please loosen the content filter.
S
Siddhartha
complete
We have reduced the scope of the filters.
A
Amanda Davies
Yep, I'm getting frustrated with asking a simple question or saying a simple phrase like 'train wreck' and it puts the breaks on and lose the momentum of my conversation.
Fix it yesterday.
yifan deng
Yes, I can barely do any story now, it's so incredible that I have to start looking for new tools again, but I would rather go back to the way things were
yifan deng
In fact, it's completely harmless, it's just creating virtual content, LLMs have their own censorship mechanisms, so why add to them?
sheepsy XD
I liked it better when Merlin didn't have so many restrictions, I think it was what did it better than chatgpt, the ias that don't allow a little more freedom in the content get boring in the end, but if it's for answers, I used Merlin more because it allowed me to create more explicit content, I suppose I'd better go back to chatgpt now that they took away from Merlin what made it fun
Olga
I used the chatbots to help me improve my translation of a fantasy story, and now I can't send any excerpts that have bickering or even remote mentions of battles, deaths, etc., even in a fictional setting and within the PG rating. It's ridiculous how such a great tool became completely unusable overnight.
Olga
I can't even send hunting scenes or describe carnovores... great.
S
SurReality Releases
yeah.. I hadn't noticed it being a problem at all really until last night a bit and today it's made the entire experience unusable. Please either change this back to how it was, or let me know if you wont because my continued subscription hinges on that answer... not paying to be treated like a child while having my reason for the program (a free flowing conversation) rendered impossible.
Alfred Penniworth
LLM has built-in protection against such content. The filter only breaks the work with it. It is better, in principle, to return as it was, without a "filter".