ant.thinking imo is short term of "Anthropic" and thinking refer to reasoning process of the LLM. you can see the prefix "ant" repeat a lot behind the scenes of claude 3.5, it may refer to Anthropic
I"m pretty sure is an like a mixture of experts thing, where the encoded term 'antthinking' they created, so that when this trigger is mentioned that def is considered. Could it be "Artifact Neural Transformer?"
that is usually the way I have explained how 'AI' works but seeing more increasingly interesting info about how certain data scientists are saying that is how it used to work, but now we're not really sure how its working - I think over the next few months more info with be revealed.
Im going to create a system using python where GPT4o and 3.5 Sonnet will use antthinking prompts on each other to solve for "x" solution. This will either be great or terrible 😂
![Lp. Сердце Вселенной #1 НАЧАЛО ПУТЕШЕСТВИЯ [Новый сезон] • Майнкрафт](http://i.ytimg.com/vi/380Q7gLWGvk/mqdefault.jpg)
ant.thinking imo is short term of "Anthropic" and thinking refer to reasoning process of the LLM. you can see the prefix "ant" repeat a lot behind the scenes of claude 3.5, it may refer to Anthropic
Claude is brilliant I love it, it's not 100 percent visual basic friendly though. Close to it and other computing languages made it is it c++
Love the Delve hat. I wonder how many people "get" that hat?🤣
I"m pretty sure is an like a mixture of experts thing, where the encoded term 'antthinking' they created, so that when this trigger is mentioned that def is considered. Could it be "Artifact Neural Transformer?"
oooh interesting okay - I like that idea - silicon valley def loves their acronyms lol
Kinda sounds like the internal dialogue one would have when trying to make a decision.
When I see that prompt, it's hard to reconcile the idea of a language model being just a prediction of the next token.
that is usually the way I have explained how 'AI' works but seeing more increasingly interesting info about how certain data scientists are saying that is how it used to work, but now we're not really sure how its working - I think over the next few months more info with be revealed.
Im going to create a system using python where GPT4o and 3.5 Sonnet will use antthinking prompts on each other to solve for "x" solution. This will either be great or terrible 😂