>>717
that was correct reaction as for me this AI was also so dumb that I was laughing all the time. when it was coming up with most ridiculous ideas.


But on the other hand it is reasonable as it has no clue what the hell is going on and why it is happening
If you ask it to do something more reasonable it does it much better also it does better if you give it more freedom.

it can write in the tone of original writer and it does it all the time which is probably why the timmy story ended so childish as I did not give it any good reference
but the problem is that it has pretty limited context memory like somewhere about 2000 tokens (pieces of words) and if you give it too much info it goes ADHD.
Like in the text above it can be seen how it starts reasonably well but then it starts using pervious text as reference and wanders into something where it pretty much starts repeating itself


But you can't feed it a big story and get some sensible output, I think for that purpose you woud need to do a training to create LORA (finetune AI) and then it will be quite good in that topic.

what you actually can do is to feed it some text and make it act like character in that description
so you could put part of the story with example how characters are acting then it will either continue it or you can interact with those characters yourself
it is more or less what replika does

later I may try to put some stories from gc and ask it to continue them

>>716
well it started reasonably good, but then wandered into that primary school student style
I made more attempts and in other situations it just got off topic or stuck in the infinite loops.