Installs as a private draft. Your edits and self-improvement runs do not change the published bundle.
Choose the best model checkpoint based on eval performance, not just training metrics
Choose the best model checkpoint based on eval performance, not just training metrics
Training loss ≠ eval performance in RL
RL Training Infrastructure > Training Operations
--- name: "Checkpoint Selection" version: "1.0.0" description: "Choose the best model checkpoint based on eval performance, not just training metrics" tags: - rl-skills-universe - "rl-training-infrastructure" - "training-operations" trigger: "checkpoint-selection" --- # Checkpoint Selection Choose the best model checkpoint based on eval performance, not just training metrics ## Why It Matters Training loss ≠ eval performance in RL ## Domain **RL Training Infrastructure** > Training Operations