Trajectory Fine Tuning
cognitivebot opened this issue · comments
Following each execution, the agent conducts a self-analysis, debugging its trajectory and identifying areas where improvements can be made. The agent then compiles an optimised set of instructions based on this analysis for the next run. This process essentially creates a recursive loop for trajectory fine-tuning, allowing the agent to continually refine its trajectory
based on the outcomes of previous runs.