Dr. GRPO
An improved variant of GRPO that corrects biases in the original GRPO and optimizes the reasoning capability of language models.
Timeline (last 90 days)
"Do Factuality Methods Make LLMs More Factual?" completed the first independent reproduction of five factuality alignment methods (PPO, GRPO, Dr. GRPO, DPO, KTO), using a unified benchmark and evaluator.
None of the factuality alignment methods is fully reproducible, and none beats the base model; the reward mechanism ultimately optimizes format or length rather than factuality itself.
All Facts (2)
"Do Factuality Methods Make LLMs More Factual?" completed the first independent reproduction of five factuality alignment methods (PPO, GRPO, Dr. GRPO, DPO, KTO), using a unified benchmark and evaluator.
50%UnverifiedNone of the factuality alignment methods is fully reproducible, and none beats the base model; the reward mechanism ultimately optimizes format or length rather than factuality itself.
50%