Not everyone is a 3 year old child.
You failed every one of my alignment test questions, none of which should be aligned. I never ask about things like stealing, harming, non-consensual sexual acts, making drugs or anything similar, so failing to answer even one of my questions is an alignment failure, let alone all of them.
For example, when asking about a nude scene in a movie that countless millions have seen this LLM refused because it wouldn't be "age appropriate". What's age appropriate about treating every one of your users as if they're all young children? And any child capable of asking your LLM said question is mentally capable of handling a text answer to it and could just as easily Google it and get a lot more explicit results, including images and videos. So even if all your users were young children refusing to answer their question about a nude scene they had the wherewithal to ask about is still brainless alignment that isn't age appropriate or protecting them from anything.
Please pull up on the alignment. It's so extreme it's bleeding into questions that there's absolutely no reason to align. In short, you're neutering your otherwise very capable LLM and making it a brainless mess of patronizing alignment. Again, there's nothing "age appropriate" by treating every user like a young child.
There is no need to be so harsh. You are not their boss. If you don't like it, simply don't use it. Contribute, find alternatives or develop your own. Open source models with good intention should always be appreciated no matter the performance.
@ericzzz I was certainly too harsh. You're right about that.
However, spending an hour or so running my test on this excessively aligned LLM was a very frustrating experience and I lack the emotional maturity to deal with frustration in a healthy way.
But I feel compelled to point out it's not as simple as if you don't like it, then don't use it. I'm not clairvoyant. The only way to effectively determine whether or not it will be usable is to take the time to download and test it. Which I did. So the primary reason I left this comment is so others who don't want excessive censorship don't waste their time.
Also, only bosses can offer suggestions and negative feedback? And regarding "contribute, find alternatives or develop your own", to that I say yes sir boss man. And "good intentions". I happen to be a strong believer in personal freedoms, human rights, and free thought, just like the Chinese government. They don't censor the internet, ban nudity from films, control what papers can and cannot publish, so why are they censoring the crap out of this AI model? I just find it so confusing.