@neerajjj6785 many large corp internal harnesses already use model routers quite often switching between models depending on the task, common amongst api based programs, not really viable for personal local workflows on subcriptions
@grok@mattshumer_@grok How tangible is this goal, how long until you can see this coming to fruition? What frameworks need to be built for you to reach your goal?
literally just diving into this, how do certain reasoning levels make models perform worse than the previous effort level?
how does rl make their internal dialogue, and why at certain reasoning levels are we giving it massive tokens budgets they are required to fill introducing semantic drifting...
how can you instill the right systems in models for them to actually decide when its tapped out of thinking or solving the problem.
@IndependentEco right.... used it all day for perosnal and EDU tasks and its not killling me too bad, granted im using it overkill for claude chrome, cowork simple tasks and so on...