Durgesh Kalwar

Upasana Biswas, Durgesh Kalwar, Subbarao Kambhampati et al.

Mutual adaptation is a central challenge in human-AI teaming, as humans naturally adjust their strategies in response to an AI agent's behavior. Existing approaches attempt to approximate human behavior by diversifying training partners; however, these partners are typically static and fail to capture the adaptive nature of human teammates. When agents are trained jointly in standard multi-agent settings, they often converge to opaque coordination strategies that work only with their co-trained partners, leading to poor generalization. To model adaptive human behavior, we formulate human-AI teaming as an Interactive Partially Observable Markov Decision Process (I-POMDP). We propose NestRL, a nested training regime that learns the solution to a finite-level I-POMDP by training agents at each level against adaptive agents from the level below. This exposes agents to adaptive behavior while preventing emergence of opaque coordination strategies. We provide theoretical analysis showing that NestRL agents avoid convergence to partner-specific strategies, and validate this empirically in the Overcooked domain against state-of-the-art baselines. NestRL achieves higher task performance with both unseen adaptive agents and real human teammates, while exhibiting significantly greater adaptability over the course of interaction.

Durgesh Kalwar, Omkar Shelke, Somjit Nath et al.

Improving sample efficiency is a key challenge in reinforcement learning, especially in environments with large state spaces and sparse rewards. In literature, this is resolved either through the use of auxiliary tasks (subgoals) or through clever exploration strategies. Exploration methods have been used to sample better trajectories in large environments while auxiliary tasks have been incorporated where the reward is sparse. However, few studies have attempted to tackle both large scale and reward sparsity at the same time. This paper explores the idea of combining exploration with auxiliary task learning using General Value Functions (GVFs) and a directed exploration strategy. We present a way to learn value functions which can be used to sample actions and provide directed exploration. Experiments on navigation tasks with varying grid sizes demonstrate the performance advantages over several competitive baselines.

Durgesh Kalwar, Omkar Shelke, Harshad Khadilkar

We consider the inventory management problem, where the goal is to balance conflicting objectives such as availability and wastage of a large range of products in a store. We propose a reinforcement learning (RL) approach that utilises General Value Functions (GVFs) to derive domain-backed inventory replenishment policies. The inventory replenishment decisions are modelled as a sequential decision making problem, which is challenging due to uncertain demand and the existence of aggregate (cross-product) constraints. In existing literature, GVFs have primarily been used for auxiliary task learning. We use this capability to train GVFs on domain-critical characteristics such as prediction of stock-out probability and wastage quantity. Using this domain expertise for more effective exploration, we train an RL agent to compute the inventory replenishment quantities for a large range of products (up to 6000 in the reported experiments), which share aggregate constraints such as the total weight/volume per delivery. Additionally, we show that the GVF predictions can be used to provide additional domain-backed insights into the decisions proposed by the RL agent. Finally, since the environment dynamics are fully transferred, the trained GVFs can be used for faster adaptation to vastly different business objectives (for example, due to the start of a promotional period or due to deployment in a new customer environment).

Durgesh Kalwar, Vineeth B. S

We consider the problem of designing a sequential decision making agent to maximize an unknown time-varying function which switches with time. At each step, the agent receives an observation of the function's value at a point decided by the agent. The observation could be corrupted by noise. The agent is also constrained to take safe decisions with high probability, i.e., the chosen points should have a function value greater than a threshold. For this switching environment, we propose a policy called Adaptive-SafeOpt and evaluate its performance via simulations. The policy incorporates Bayesian optimization and change point detection for the safe sequential optimization problem. We observe that a major challenge in adapting to the switching change is to identify safe decisions when the change point is detected and prevent attraction to local optima.

Subbarao Kambhampati, Kaya Stechly, Karthik Valmeekam et al.

Intermediate token generation (ITG), where a model produces output before the solution, has been proposed as a method to improve the performance of language models on reasoning tasks. These intermediate tokens have been called "reasoning traces" or even "thoughts" -- implicitly anthropomorphizing the model, implying these tokens resemble steps a human might take when solving a challenging problem.In this paper, we present evidence that this anthropomorphization isn't a harmless metaphor, and instead is quite dangerous -- it confuses the nature of these models and how to use them effectively, and leads to questionable research.

Soumya Rani Samineni, Durgesh Kalwar, Karthik Valmeekam et al.

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.

Soumya Rani Samineni, Durgesh Kalwar, Vardaan Gangal et al.

Reinforcement Learning with Verifiable Rewards (RLVR)-based post-training of Large Language Models (LLMs) has been shown to improve accuracy on reasoning tasks and continues to attract significant attention. Existing RLVR methods, however, typically treat all tokens uniformly without accounting for token-level advantages. These methods primarily evaluate performance based on final answer correctness or Pass@K accuracy, and yet make claims about RL post-training leading to improved reasoning traces. This motivates our investigation into the effect of RL post-training on intermediate tokens which are not directly incentivized. To study this, we design an experimental setup using the GRPO algorithm with Qwen-2.5-0.5B model on the GSM8K dataset. We introduce trace coherence, a First-Order Logic (FOL)-based measure to capture the consistency of reasoning steps by identifying errors in the traces. We distinguish between trace validity and trace coherence, noting that the former implies logical soundness while the latter measures local coherence via lack of errors. Our results show that RL post-training overall improves trace coherence with the most significant gains on problems where the base model fails but the RL model succeeds. Surprisingly, RL enhances local coherence without necessarily producing valid or correct solutions. This highlights a crucial distinction: improved local coherence in reasoning steps does not guarantee final answer correctness. We argue that claims of improved reasoning via RL must be examined with care, as these may be based on improved trace coherence, which may not translate into fully valid mathematical proofs.

Durgesh Kalwar, Mayank Baranwal, Harshad Khadilkar

In today's data-sensitive landscape, distributed learning emerges as a vital tool, not only fortifying privacy measures but also streamlining computational operations. This becomes especially crucial within fully decentralized infrastructures where local processing is imperative due to the absence of centralized aggregation. Here, we introduce DYNAWEIGHT, a novel framework to information aggregation in multi-agent networks. DYNAWEIGHT offers substantial acceleration in decentralized learning with minimal additional communication and memory overhead. Unlike traditional static weight assignments, such as Metropolis weights, DYNAWEIGHT dynamically allocates weights to neighboring servers based on their relative losses on local datasets. Consequently, it favors servers possessing diverse information, particularly in scenarios of substantial data heterogeneity. Our experiments on various datasets MNIST, CIFAR10, and CIFAR100 incorporating various server counts and graph topologies, demonstrate notable enhancements in training speeds. Notably, DYNAWEIGHT functions as an aggregation scheme compatible with any underlying server-level optimization algorithm, underscoring its versatility and potential for widespread integration.