RL in Name Only? Analyzing the Structural Assumptions in RL post-training for LLMs

Reinforcement learning-based post-training of large language models (LLMs) has recently gained attention, particularly following the release of DeepSeek R1, which applied GRPO for fine-tuning. Amid the growing hype around improved reasoning abilities attributed to RL post-training, we critically examine the formulation and assumptions underlying these methods. We start by highlighting the popular structural assumptions made in modeling LLM training as a Markov Decision Process (MDP), and show how they lead to a degenerate MDP that doesn't quite need the RL/GRPO apparatus. The two critical structural assumptions include (1) making the MDP states be just a concatenation of the actions-with states becoming the context window and the actions becoming the tokens in LLMs and (2) splitting the reward of a state-action trajectory uniformly across the trajectory. Through a comprehensive analysis, we demonstrate that these simplifying assumptions make the approach effectively equivalent to an outcome-driven supervised learning. Our experiments on benchmarks including GSM8K and Countdown using Qwen-2.5 base models show that iterative supervised fine-tuning, incorporating both positive and negative samples, achieves performance comparable to GRPO-based training. We will also argue that the structural assumptions indirectly incentivize the RL to generate longer sequences of intermediate tokens-which in turn feeds into the narrative of "RL generating longer thinking traces." While RL may well be a very useful technique for improving the reasoning abilities of LLMs, our analysis shows that the simplistic structural assumptions made in modeling the underlying MDP render the popular LLM RL frameworks and their interpretations questionable.

Policy Search using Dynamic Mirror Descent MPC for Model Free Off Policy RL

Recent works in Reinforcement Learning (RL) combine model-free (Mf)-RL algorithms with model-based (Mb)-RL approaches to get the best from both: asymptotic performance of Mf-RL and high sample-efficiency of Mb-RL. Inspired by these works, we propose a hierarchical framework that integrates online learning for the Mb-trajectory optimization with off-policy methods for the Mf-RL. In particular, two loops are proposed, where the Dynamic Mirror Descent based Model Predictive Control (DMD-MPC) is used as the inner loop to obtain an optimal sequence of actions. These actions are in turn used to significantly accelerate the outer loop Mf-RL. We show that our formulation is generic for a broad class of MPC based policies and objectives, and includes some of the well-known Mb-Mf approaches. Based on the framework we define two algorithms to increase sample efficiency of Off Policy RL and to guide end to end RL algorithms for online adaption respectively. Thus we finally introduce two novel algorithms: Dynamic-Mirror Descent Model Predictive RL(DeMoRL), which uses the method of elite fractions for the inner loop and Soft Actor-Critic (SAC) as the off-policy RL for the outer loop and Dynamic-Mirror Descent Model Predictive Layer(DeMo Layer), a special case of the hierarchical framework which guides linear policies trained using Augmented Random Search(ARS). Our experiments show faster convergence of the proposed DeMo RL, and better or equal performance compared to other Mf-Mb approaches on benchmark MuJoCo control tasks. The DeMo Layer was tested on classical Cartpole and custom-built Quadruped trained using Linear Policy.