Really liking this /goal usage
- a main orchestrator with /goal which spawns subagents
- implementor agent -> passes work to a review agent
- loop till finish while keeping main orchestrator thread context light
New Anthropic research: Natural Language Autoencoders.
Models like Claude talk in words but think in numbers. The numbers—called activations—encode Claude’s thoughts, but not in a language we can read.
Here, we train Claude to translate its activations into human-readable text.
@DegeniusQ Hey qyl - could I check how would you rank funding arb opportunities?
I know high APY's will degenerate quickly, and want to have a more balanced filter, for longer lasting pairs
How do you handle this?
Also any chance I could dm you on this, thanks!
@DegeniusQ How does one go about to improve their execution?
Is this solely a latency issue? Or something along the lines of smart routing or so?
Thank you!