---
title: "DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models"
description: "DeepSeekMath shows how targeted data and reinforcement learning can turn a compact open model into a strong mathematical reasoner. Read it for the data-selection pipeline behind its 120 billion math tokens and for GRPO, a PPO variant that removes the separate critic model by learning from relative rewards within groups of sampled answers."
canonical_url: "https://fanout.sh/daily/2026-08-02-deepseekmath-grpo"
md_url: "https://fanout.sh/daily/2026-08-02-deepseekmath-grpo.md"
last_updated: "2026-08-02"
access: "public"
---

# DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models

DeepSeekMath shows how targeted data and reinforcement learning can turn a compact open model into a strong mathematical reasoner. Read it for the data-selection pipeline behind its 120 billion math tokens and for GRPO, a PPO variant that removes the separate critic model by learning from relative rewards within groups of sampled answers.

## Paper details

- Authors: Shao et al.

- Venue: arXiv 2024

- Track: ML Math

- Difficulty: Hard

- Reading time: 55 min

- Original paper: https://arxiv.org/abs/2402.03300

## What you will learn

- How an iterative Common Crawl filtering pipeline expands a small high-quality math seed into a 120-billion-token training corpus.

- How GRPO estimates each answer's advantage by comparing its reward with the other answers sampled for the same problem.

- Why removing the learned critic reduces memory use while retaining PPO-style clipped policy updates for reinforcement learning.

## Continue exploring

[Explore GRPO](https://fanout.sh/knowledge-graph?node=ai-research%3Aconcept%3Agrpo): Open Fanout's GRPO node and connect the paper to PPO, group-relative advantages, critic-free reinforcement learning, and mathematical reasoning.

---
This representation contains public Fanout content only. Protected Pro lessons, account data, billing, checkout, and pricing are not included.

Browse the public content map: https://fanout.sh/sitemap.md