Divide-and-Conquer CoT: RL for Reducing Latency via Parallel Reasoning

arXiv:2601.23027 · cs.LG · Submitted 2026-01-30 · Read on arXiv

cs.LG

Submitted: 2026-01-30

Updated: 2026-10-05

Code: https://github.com/amahankali10/DC_CoT_RL_for_Low_Latenc

Project page: https://moonshotai.github.io/Kimi-K2/thinking.h

Terminology

• Accuracy — 4× in this paper · explained in 1 episode
• Benchmark — 1× in this paper · explained in 2 episodes

Sources

• GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning
• MathArena: Evaluating LLMs on Uncontaminated Math Competitions
• Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads
• The Era of Agentic Organization: Learning to Organize with Language Models
• The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models
• Multi-Agent Collaboration via Evolving Orchestration
• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
• Improving Factuality and Reasoning in Language Models through Multiagent Debate
• Single-agent or Multi-agent Systems? Why Not Both?
• OpenThoughts: Data Recipes for Reasoning Models
• Don't Overthink it. Preferring Shorter Thinking Chains for Improved LLM Reasoning
• JustRL: Scaling a 1.5B LLM with a Simple RL Recipe
• Skywork Open Reasoner 1 Technical Report
• BroRL: Scaling Reinforcement Learning via Broadened Exploration
• OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automation
• Learning to Keep a Promise: Scaling Language Model Decoding Parallelism with Learned Asynchronous Decoding
• The Art of Scaling Reinforcement Learning Compute for LLMs
• Towards a Science of Scaling Agent Systems
• Fast Inference from Transformers via Speculative Decoding
• Solving Quantitative Reasoning Problems with Language Models

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks