# Beyond Token-Level Policy Gradients for Complex Reasoning with Large Language Models

arXiv: https://arxiv.org/abs/2602.14386 · HTML: https://mlsanity.com/paper/2602.14386/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning.
- Builds on 19 of its 28 references that are in the map.
- First posted on arXiv 16 Feb 2026.

## Placed in

- LLM reasoning / [Test-time compute and efficient thinking](https://mlsanity.com/lr/test-time-compute-and-efficient-thinking/)

## Builds on (most cited first, 19 of 19)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://mlsanity.com/paper/2503.14476/) (2025)
- [Group Sequence Policy Optimization](https://mlsanity.com/paper/2507.18071/) (2025)
- [DeepSeek-V3 Technical Report](https://mlsanity.com/paper/2412.19437/) (2024)
- [VAPO: Efficient and Reliable Reinforcement Learning for Advanced Reasoning Tasks](https://mlsanity.com/paper/2504.05118/) (2025)
- [MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention](https://mlsanity.com/paper/2506.13585/) (2025)
- [Hydra: Sequentially-Dependent Draft Heads for Medusa Decoding](https://mlsanity.com/paper/2402.05109/) (2024)
- [A Survey on Human Preference Learning for Large Language Models](https://mlsanity.com/paper/2406.11191/) (2024)
- [Benchmarking and Improving Large Vision-Language Models for Fundamental Visual Graph Understanding and Reasoning](https://mlsanity.com/paper/2412.13540/) (2024)
- [Multi-Token Prediction Needs Registers](https://mlsanity.com/paper/2505.10518/) (2025)
- [Thinking in Character: Advancing Role-Playing Agents with Role-Aware Reasoning](https://mlsanity.com/paper/2506.01748/) (2025)
- [Policy Gradient with Active Importance Sampling](https://mlsanity.com/paper/2405.05630/) (2024)
- [Tokenization Constraints in LLMs: A Study of Symbolic and Arithmetic Reasoning Limits](https://mlsanity.com/paper/2505.14178/) (2025)

## Abstract (arXiv)

Existing policy-gradient methods for auto-regressive language models typically select subsequent tokens one at a time as actions in the policy. While effective for many generation tasks, such an approach may not fully capture the structure of complex reasoning tasks, where a single semantic decision is often realized across multiple tokens--for example, when defining variables or composing equations. This introduces a potential mismatch between token-level optimization and the inherently block-level nature of reasoning in these settings. To bridge this gap, we propose Multi-token Policy Gradient Optimization (MPO), a framework that treats sequences of K consecutive tokens as unified semantic actions. This block-level perspective enables our method to capture the compositional structure of reasoning trajectories and supports optimization over coherent, higher-level objectives. Experiments on mathematical reasoning and coding benchmarks show that MPO outperforms standard token-level policy gradient baselines, highlight the limitations of token-level policy gradients for complex reasoning, motivating future research to look beyond token-level granularity for reasoning-intensive language tasks.