GRPO — Group Relative Policy Optimization
Lightbulb Partners's implementation of Group Relative Policy Optimization (GRPO) from DeepSeek-R1, a simplified alternative to PPO that removes the critic network for efficient agent policy training.
Lightbulb Partners's implementation of Group Relative Policy Optimization (GRPO) from DeepSeek-R1, a simplified alternative to PPO that removes the critic network for efficient agent policy training.