ZABKA_TM

A single prompt to follow:

“You are in a competition. Compose a haiku that can be translated into the maximum number of languages, while still preserving the original meaning and fitting the strict rules of the haiku format.

Once you have composed the original haiku, list each language it can be translated into while remaining a haiku in both languages. In this competition, a point is scored for every language successfully translated, but you automatically fail the test if a single attempted haiku does not have the correct number of syllables. The test is also failed if a translated haiku does not preserve the original haiku’s meaning.”

GPT4 failed the logic test, even though the translations appeared to be accurate. It claimed a haiku could translate to sixteen languages. Of those translations I think about five of them were correct haikus. Will try later, once GPT4 stops the throttling.

Let’s say you spend an unholy amount of processing time training a 70b. You like history. You want a good LLM for historical info.

By the time you upload it the LLM is outdated. Now what?

If you want it to speak accurately about modern events you’d have to retrain it again. Repeating the process over and over, because time keeps moving on while your LLM does not.

This clearly could become more efficient. Optimally, each subject would probably need to be considered a separate file while the central “brain” of the LLM becomes its own structure.

As it stands, updating the entire LLM is very cost prohibitive and makes no sense if you’re trying to work out specific data points. Why, for example, would you want to update the entire Cantonese dictionary when you just want to fix the list of Alaskan donut shops?

I understand that the tech currently has to treat both the information and the “thinking” behind an LLM as one and the same. It seems more efficient, more effective, to separate the two.