Verified75% confidenceFactExact time
GRPO was proposed by the DeepSeek team in 2024 and is a lightweight variant of PPO.
3
Sources
75%
Confidence
Long-term
Relevance
7/17/2026
First Seen
Sources
Related Claims
UnverifiedGRPO emerged in 2024 and was adopted by DeepSeek-R1, further simplifying the reference model dependency on top of DPO81% similarUnverifiedGrok-1 was released in open-source form in March 2024, with 314 billion parameters, using a Mixture of Experts (MoE) architecture.62% similarUnverifiedFlamingo was proposed by DeepMind in 2022, and BLIP-2 was proposed by Salesforce in 2023.62% similarVerifiedDeepSeek was founded in 2023 by High-Flyer, a quantitative private equity giant.58% similarUnverifiedSince 2024, further simplified alignment algorithms such as ORPO, SimPO, and KTO have emerged.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/542330API
curl https://kongchang.com/api/v1/knowledge/claims/542330MCP
get_claim(id=542330)