DeepSeek introduces GRPO (DeepSeekMath)
AIMade by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made
GRPO introduced (DeepSeekMath)
Why it matters
Group relative policy optimization without a value function later became the de facto standard for RL in reasoning models
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2402.03300Added in Week 35, 2026 · See that week
On the map — ⑫ Training Data line
- Previous stopTextbooks Are All You Need
- This record · stopGRPO
- Next stopRunning out of data?
In the model family tree — DeepSeek family · 12 versions
- Earlier versionThe first version in this branch
- This recordDeepSeekMath
- Later versionDeepSeek-Coder-V2