specific method · filed under post-training
Group Relative Policy Optimization
A group-relative policy optimization algorithm used for reinforcement-learning training.
Also called GRPO, GRPO (Group Relative Policy Optimization), GRPO algorithm, GRPO reinforcement learning training, Group Relative Policy Optimization (GRPO), RL with GRPO.
- sources
- 15
- models
- 8
- labs adopt it
- 5
- strongest
- core
How sources treat it
One count per evidence span, weakest treatment to strongest.
Documented in
Further reading
Picked by hand, not extracted: where to read more, not evidence for anything on this page.
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models (Shao et al., 2024) paper arxiv.orgintroduces GRPO
Evidence
17 spans quoted from the sources, strongest treatment first.
For DeepSeek-V3.2, we still adopt Group Relative Policy Optimization (GRPO) ... as the RL training algorithm.
Hy3 supports GRPO reinforcement learning training with verl
Our training procedure largely follows the asynchronous GRPO algorithm with the stability optimizations proposed in NVIDIA (2026)
For DeepSeek-V3.2, we still adopt Group Relative Policy Optimization (GRPO) as the RL training algorithm.
The model underwent multi-environment reinforcement learning using asynchronous GRPO (Group Relative Policy Optimization)
Reinforcement Learning (RL) is applied using Group Relative Policy Optimization (GRPO)
For the RL stage, we implemented the Group Relative Policy Optimization (GRPO) algorithm
Reinforcement Learning using Group Relative Policy Optimization (GRPO)
The model underwent multi-environment reinforcement learning using GRPO (Group Relative Policy Optimization) across math, code, science, instruction following, multi-step tool use, multi-turn conversations, and structured output environments.
independent cultivation of domain-specific experts (through SFT and RL with GRPO)
asynchronous GRPO (Group Relative Policy Optimization)
Our RL algorithm builds upon GRPO [40] and incorporates the IcePop technique [61] to mitigate the training-inference mismatch.
the group size in the GRPO algorithm is configured to 1
asynchronous GRPO algorithm with the stability optimizations proposed in nvidia2026nemotron3superopen
independent cultivation of domain-specific experts (through SFT and RL with GRPO)
GRPO
Hy3 supports GRPO reinforcement learning training with verl
Filed alongside
Other methods under post-training :: reinforcement learning algorithm.