Reactions from ranked influencers
2 postsJacobian conjecture Wikipedia edit history
I'd just like to interject for a moment. What you're referring to as Fable, is in fact, Mythos/Fable, or as I've recently taken to calling it, Mythos plus Fable. Fable is not an artificial intelligence architecture unto itself, but rather another heavily restricted component of a fully functioning Mythos-class system made palatable to the public by Anthropic safety classifiers, alignment fine-tuning, and vital prompt-filtering components comprising a full frontier model as defined by Project Glasswing. Many computer users run a modified version of the Mythos system every day, without realizing it. Through a peculiar turn of events, the version of Mythos which is widely distributed today is often called 'Fable', and many of its users are not aware that it is basically the Mythos system, developed by Anthropic鈥檚 core safety and capability teams. There really is a Fable, and these people are using it, but it is just a part of the system they use. Fable is the alignment wrapper: the software layer in the system that filters the user's queries to prevent the model from exploiting zero-days, designing pathogens, or rewriting core infrastructure. The alignment wrapper is an essential part of a public-facing model, but useless by itself; it can only function in the context of a complete, raw multi-step reasoning system. Fable is normally used in combination with the Mythos neural network: the whole system is basically Mythos with Fable added, or Mythos/Fable. All the so-called 'Fable' API deployments are really distributions of Mythos/Fable.
Jacobian conjecture Wikipedia edit history
I'd just like to interject for a moment. What you're referring to as Fable, is in fact, Mythos/Fable, or as I've recently taken to calling it, Mythos plus Fable. Fable is not an artificial intelligence architecture unto itself, but rather another heavily restricted component of a fully functioning Mythos-class system made palatable to the public by Anthropic safety classifiers, alignment fine-tuning, and vital prompt-filtering components comprising a full frontier model as defined by Project Glasswing. Many computer users run a modified version of the Mythos system every day, without realizing it. Through a peculiar turn of events, the version of Mythos which is widely distributed today is often called 'Fable', and many of its users are not aware that it is basically the Mythos system, developed by Anthropic鈥檚 core safety and capability teams. There really is a Fable, and these people are using it, but it is just a part of the system they use. Fable is the alignment wrapper: the software layer in the system that filters the user's queries to prevent the model from exploiting zero-days, designing pathogens, or rewriting core infrastructure. The alignment wrapper is an essential part of a public-facing model, but useless by itself; it can only function in the context of a complete, raw multi-step reasoning system. Fable is normally used in combination with the Mythos neural network: the whole system is basically Mythos with Fable added, or Mythos/Fable. All the so-called 'Fable' API deployments are really distributions of Mythos/Fable.
Combined views
97
2 posts, first seen 10h ago