H
Howardism
Howardism · Vol. 03Plate I · No. 01

Reward Design, tagged.

Notes5TagReward DesignOldest22 Jul 2026Newest17 Aug 2026

Every article tagged reward design, newest first.

  1. C01
    RL from Execution Feedback (RLEF)
    Reinforcement LearningPost TrainingCode Generation+2
    · 15′
  2. C02
    Offline Multi-Step Tool-Use RL (SWiRL)
    Reinforcement LearningPost TrainingTool Use+2
    · 12′
  3. C03
    Error-Penalized Abstention Training
    Post TrainingReinforcement LearningReward Design+2
    · 22′
  4. C04
    Turn-Level Credit Assignment
    Reinforcement LearningPost TrainingAgentic Rl+2
    · 19′
  5. C05
    Trained Calibration
    Post TrainingCalibrationHonesty+1
    · 17′