agents
fact
neutral
Training on successful agent trajectories can introduce noisy supervision because successful trajectories may contain ineffective, redundant, or risky steps
task success does not guarantee high-quality supervision: successful trajectories may still contain ineffective, redundant, or risky steps. Directly using such trajectories for SFT can introduce noisy supervision and encourage models to imitate undesirable problem-solving behaviors.
Computation and Language30 Aug 2026