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

Reward Design, tagged.

Notes6TagReward DesignOldest22 Jul 2026Newest24 Sept 2026

Every article tagged reward design, newest first.

  1. C01
    ExecCritic: Learn to Test, Test to Improve
    Reinforcement LearningPost TrainingCode Generation+4
    · 16′
  2. C02
    RL from Execution Feedback (RLEF)
    Reinforcement LearningPost TrainingCode Generation+2
    · 16′
  3. C03
    Offline Multi-Step Tool-Use RL (SWiRL)
    Reinforcement LearningPost TrainingTool Use+2
    · 12′
  4. C04
    Error-Penalized Abstention Training
    Post TrainingReinforcement LearningReward Design+2
    · 22′
  5. C05
    Turn-Level Credit Assignment
    Reinforcement LearningPost TrainingAgentic Rl+2
    · 20′
  6. C06
    Trained Calibration
    Post TrainingCalibrationHonesty+1
    · 27′