Comment by randomblock1
3 days ago
Google says that writing and conciseness was a big focus for Argon. I suspect it's also been told to to write in wiki style, which usually gets to the point pretty quickly.
3 days ago
Google says that writing and conciseness was a big focus for Argon. I suspect it's also been told to to write in wiki style, which usually gets to the point pretty quickly.
Synthesis of data by LLMs is really good today. I have tried it several times, e.g. for obituaries, or profiles of known people. Nevertheless, I would suggest to double-check everything (Jev, another agents) as one LLM is not enough. Its knowledge can be capped somewhere you don’t want.
Fiction similar to Fakeipedia? Skill for unslop and a little detailed prompt is good enough to generate texts that 98 pct don’t recognize from human-written ones.
I've just made the code public: https://github.com/darajava/fakeipedia
Awesome. Thanks. Let’s add some events from Ted Chiang and Jorge Luis Borges books.