Showing cs.LGShow all
3 papers · 1 filter
cs.LG2026
Goal Retrieval versus Action-Value Ordering: Contrastive Critics under Best-of- Selection
Ayushman Singh, Siddharth Aphale
Goal-conditioned action generators can produce many plausible actions. Best-of-\(K\) selection then uses a critic to rank those actions for the current state and goal. Standard con…
cs.LG2026
Per-Context Reverse Curricula for Sparse-Reward Reinforcement Learning
Siddharth Aphale, Ayushman Singh
Sparse-reward reinforcement learning often fails because rollouts from the unassisted evaluation start rarely reach later task stages. Reverse curricula address this by starting so…
cs.LG2026
SFT Overtraining Predicts Rank Inversion via Entropy Collapse Under RLVR
Siddharth Aphale, Kelly Liu
The standard heuristic of selecting the SFT checkpoint with the highest pass@1 for GRPO can fail when SFT compresses the rollout distribution. For binary rewards, the expected with…