trajectory-reflection

Tag

Cards List
#trajectory-reflection

@blc_16: If you want to understand why RL struggles with long-horizon agent tasks, this is a good explanation. The core issue is…

X AI KOLs Timeline · 5d ago

The post explains why Reinforcement Learning struggles with long-horizon tasks due to sparse rewards and highlights GEPA, a method that uses trajectory-level textual reflection to preserve richer feedback signals for optimization.

0 favorites 0 likes
← Back to home

Submit Feedback