Yannis Flet-Berliac

Anirudhan Badrinath, Yannis Flet-Berliac, Allen Nie et al. · stanford

Despite the recent advancements in offline reinforcement learning via supervised learning (RvS) and the success of the decision transformer (DT) architecture in various domains, DTs have fallen short in several challenging benchmarks. The root cause of this underperformance lies in their inability to seamlessly connect segments of suboptimal trajectories. To overcome this limitation, we present a novel approach to enhance RvS methods by integrating intermediate targets. We introduce the Waypoint Transformer (WT), using an architecture that builds upon the DT framework and conditioned on automatically-generated waypoints. The results show a significant increase in the final return compared to existing RvS methods, with performance on par or greater than existing state-of-the-art temporal difference learning-based methods. Additionally, the performance and stability improvements are largest in the most challenging environments and data configurations, including AntMaze Large Play/Diverse and Kitchen Mixed/Partial.

Allen Nie, Yannis Flet-Berliac, Deon R. Jordan et al. · stanford

Offline reinforcement learning (RL) can be used to improve future performance by leveraging historical data. There exist many different algorithms for offline RL, and it is well recognized that these algorithms, and their hyperparameter settings, can lead to decision policies with substantially differing performance. This prompts the need for pipelines that allow practitioners to systematically perform algorithm-hyperparameter selection for their setting. Critically, in most real-world settings, this pipeline must only involve the use of historical data. Inspired by statistical model selection methods for supervised learning, we introduce a task- and method-agnostic pipeline for automatically training, comparing, selecting, and deploying the best policy when the provided dataset is limited in size. In particular, our work highlights the importance of performing multiple data splits to produce more reliable algorithm-hyperparameter selection. While this is a common approach in supervised learning, to our knowledge, this has not been discussed in detail in the offline RL setting. We show it can have substantial impacts when the dataset is small. Compared to alternate approaches, our proposed pipeline outputs higher-performing deployed policies from a broad range of offline policy learning algorithms and across various simulation domains in healthcare, education, and robotics. This work contributes toward the development of a general-purpose meta-algorithm for automatic algorithm-hyperparameter selection for offline RL.

Kefan Dong, Yannis Flet-Berliac, Allen Nie et al. · stanford

We present a model-based offline reinforcement learning policy performance lower bound that explicitly captures dynamics model misspecification and distribution mismatch and we propose an empirical algorithm for optimal offline policy selection. Theoretically, we prove a novel safe policy improvement theorem by establishing pessimism approximations to the value function. Our key insight is to jointly consider selecting over dynamics models and policies: as long as a dynamics model can accurately represent the dynamics of the state-action pairs visited by a given policy, it is possible to approximate the value of that particular policy. We analyze our lower bound in the LQR setting and also show competitive performance to previous lower bounds on policy selection across a set of D4RL tasks.

Yao Liu, Yannis Flet-Berliac, Emma Brunskill · stanford

Offline policy optimization could have a large impact on many real-world decision-making problems, as online learning may be infeasible in many applications. Importance sampling and its variants are a commonly used type of estimator in offline policy evaluation, and such estimators typically do not require assumptions on the properties and representational capabilities of value function or decision process model function classes. In this paper, we identify an important overfitting phenomenon in optimizing the importance weighted return, in which it may be possible for the learned policy to essentially avoid making aligned decisions for part of the initial state space. We propose an algorithm to avoid this overfitting through a new per-state-neighborhood normalization constraint, and provide a theoretical justification of the proposed algorithm. We also show the limitations of previous attempts to this approach. We test our algorithm in a healthcare-inspired simulator, a logged dataset collected from real hospitals and continuous control tasks. These experiments show the proposed method yields less overfitting and better test performance compared to state-of-the-art batch reinforcement learning algorithms.

Yannis Flet-Berliac, Debabrota Basu · stanford

Although Reinforcement Learning (RL) is effective for sequential decision-making problems under uncertainty, it still fails to thrive in real-world systems where risk or safety is a binding constraint. In this paper, we formulate the RL problem with safety constraints as a non-zero-sum game. While deployed with maximum entropy RL, this formulation leads to a safe adversarially guided soft actor-critic framework, called SAAC. In SAAC, the adversary aims to break the safety constraint while the RL agent aims to maximize the constrained value function given the adversary's policy. The safety constraint on the agent's value function manifests only as a repulsion term between the agent's and the adversary's policies. Unlike previous approaches, SAAC can address different safety criteria such as safe exploration, mean-variance risk sensitivity, and CVaR-like coherent risk sensitivity. We illustrate the design of the adversary for these constraints. Then, in each of these variations, we show the agent differentiates itself from the adversary's unsafe actions in addition to learning to solve the task. Finally, for challenging continuous control tasks, we demonstrate that SAAC achieves faster convergence, better efficiency, and fewer failures to satisfy the safety constraints than risk-averse distributional RL and risk-neutral soft actor-critic algorithms.

Raphael Boige, Yannis Flet-Berliac, Arthur Flajolet et al. · stanford

Self-supervised learning has brought about a revolutionary paradigm shift in various computing domains, including NLP, vision, and biology. Recent approaches involve pre-training transformer models on vast amounts of unlabeled data, serving as a starting point for efficiently solving downstream tasks. In reinforcement learning, researchers have recently adapted these approaches, developing models pre-trained on expert trajectories. This advancement enables the models to tackle a broad spectrum of tasks, ranging from robotics to recommendation systems. However, existing methods mostly rely on intricate pre-training objectives tailored to specific downstream applications. This paper conducts a comprehensive investigation of models, referred to as pre-trained action-state transformer agents (PASTA). Our study covers a unified methodology and covers an extensive set of general downstream tasks including behavioral cloning, offline RL, sensor failure robustness, and dynamics change adaptation. Our objective is to systematically compare various design choices and offer valuable insights that will aid practitioners in developing robust models. Key highlights of our study include tokenization at the component level for actions and states, the use of fundamental pre-training objectives such as next token prediction or masked language modeling, simultaneous training of models across multiple domains, and the application of various fine-tuning strategies. In this study, the developed models contain fewer than 7 million parameters allowing a broad community to use these models and reproduce our experiments. We hope that this study will encourage further research into the use of transformers with first principle design choices to represent RL trajectories and contribute to robust policy learning.

John Dang, Shivalika Singh, Daniel D'souza et al.

We introduce the Aya Expanse model family, a new generation of 8B and 32B parameter multilingual language models, aiming to address the critical challenge of developing highly performant multilingual models that match or surpass the capabilities of monolingual models. By leveraging several years of research at Cohere For AI and Cohere, including advancements in data arbitrage, multilingual preference training, and model merging, Aya Expanse sets a new state-of-the-art in multilingual performance. Our evaluations on the Arena-Hard-Auto dataset, translated into 23 languages, demonstrate that Aya Expanse 8B and 32B outperform leading open-weight models in their respective parameter classes, including Gemma 2, Qwen 2.5, and Llama 3.1, achieving up to a 76.6% win-rate. Notably, Aya Expanse 32B outperforms Llama 3.1 70B, a model with twice as many parameters, achieving a 54.0% win-rate. In this short technical report, we present extended evaluation results for the Aya Expanse model family and release their open-weights, together with a new multilingual evaluation dataset m-ArenaHard.

Team Cohere, Aakanksha, Arash Ahmadian et al. · mila

In this report we describe the development of Command A, a powerful large language model purpose-built to excel at real-world enterprise use cases. Command A is an agent-optimised and multilingual-capable model, with support for 23 languages of global business, and a novel hybrid architecture balancing efficiency with top of the range performance. It offers best-in-class Retrieval Augmented Generation (RAG) capabilities with grounding and tool use to automate sophisticated business processes. These abilities are achieved through a decentralised training approach, including self-refinement algorithms and model merging techniques. We also include results for Command R7B which shares capability and architectural similarities to Command A. Weights for both models have been released for research purposes. This technical report details our original training pipeline and presents an extensive evaluation of our models across a suite of enterprise-relevant tasks and public benchmarks, demonstrating excellent performance and efficiency.

Nathan Grinsztajn, Yannis Flet-Berliac, Mohammad Gheshlaghi Azar et al.

To better align Large Language Models (LLMs) with human judgment, Reinforcement Learning from Human Feedback (RLHF) learns a reward model and then optimizes it using regularized RL. Recently, direct alignment methods were introduced to learn such a fine-tuned model directly from a preference dataset without computing a proxy reward function. These methods are built upon contrastive losses involving the log-likelihood of (dis)preferred completions according to the trained model. However, completions have various lengths, and the log-likelihood is not length-invariant. On the other side, the cross-entropy loss used in supervised training is length-invariant, as batches are typically averaged token-wise. To reconcile these approaches, we introduce a principled approach for making direct alignment length-invariant. Formally, we introduce a new averaging operator, to be composed with the optimality operator giving the best policy for the underlying RL problem. It translates into averaging the log-likelihood within the loss. We empirically study the effect of such averaging, observing a trade-off between the length of generations and their scores.

Yannis Flet-Berliac, Nathan Grinsztajn, Florian Strub et al.

Reinforcement Learning (RL) has been used to finetune Large Language Models (LLMs) using a reward model trained from preference data, to better align with human judgment. The recently introduced direct alignment methods, which are often simpler, more stable, and computationally lighter, can more directly achieve this. However, these approaches cannot optimize arbitrary rewards, and the preference-based ones are not the only rewards of interest for LLMs (eg., unit tests for code generation or textual entailment for summarization, among others). RL-finetuning is usually done with a variation of policy gradient, which calls for on-policy or near-on-policy samples, requiring costly generations. We introduce Contrastive Policy Gradient, or CoPG, a simple and mathematically principled new RL algorithm that can estimate the optimal policy even from off-policy data. It can be seen as an off-policy policy gradient approach that does not rely on important sampling techniques and highlights the importance of using (the right) state baseline. We show this approach to generalize the direct alignment method IPO (identity preference optimization) and classic policy gradient. We experiment with the proposed CoPG on a toy bandit problem to illustrate its properties, as well as for finetuning LLMs on a summarization task, using a learned reward function considered as ground truth for the purpose of the experiments.

Yannis Flet-Berliac, Johan Ferret, Olivier Pietquin et al.

Despite definite success in deep reinforcement learning problems, actor-critic algorithms are still confronted with sample inefficiency in complex environments, particularly in tasks where efficient exploration is a bottleneck. These methods consider a policy (the actor) and a value function (the critic) whose respective losses are built using different motivations and approaches. This paper introduces a third protagonist: the adversary. While the adversary mimics the actor by minimizing the KL-divergence between their respective action distributions, the actor, in addition to learning to solve the task, tries to differentiate itself from the adversary predictions. This novel objective stimulates the actor to follow strategies that could not have been correctly predicted from previous trajectories, making its behavior innovative in tasks where the reward is extremely rare. Our experimental analysis shows that the resulting Adversarially Guided Actor-Critic (AGAC) algorithm leads to more exhaustive exploration. Notably, AGAC outperforms current state-of-the-art methods on a set of various hard-exploration and procedurally-generated tasks.

Yannis Flet-Berliac, Reda Ouhamma, Odalric-Ambrym Maillard et al.

Policy gradient algorithms have proven to be successful in diverse decision making and control tasks. However, these methods suffer from high sample complexity and instability issues. In this paper, we address these challenges by providing a different approach for training the critic in the actor-critic framework. Our work builds on recent studies indicating that traditional actor-critic algorithms do not succeed in fitting the true value function, calling for the need to identify a better objective for the critic. In our method, the critic uses a new state-value (resp. state-action-value) function approximation that learns the value of the states (resp. state-action pairs) relative to their mean value rather than the absolute value as in conventional actor-critic. We prove the theoretical consistency of the new gradient estimator and observe dramatic empirical improvement across a variety of continuous control tasks and algorithms. Furthermore, we validate our method in tasks with sparse rewards, where we provide experimental evidence and theoretical insights.

Yannis Flet-Berliac, Philippe Preux

A common challenge in reinforcement learning is how to convert the agent's interactions with an environment into fast and robust learning. For instance, earlier work makes use of domain knowledge to improve existing reinforcement learning algorithms in complex tasks. While promising, previously acquired knowledge is often costly and challenging to scale up. Instead, we decide to consider problem knowledge with signals from quantities relevant to solve any task, e.g., self-performance assessment and accurate expectations. $\mathcal{V}^{ex}$ is such a quantity. It is the fraction of variance explained by the value function $V$ and measures the discrepancy between $V$ and the returns. Taking advantage of $\mathcal{V}^{ex}$, we propose MERL, a general framework for structuring reinforcement learning by injecting problem knowledge into policy gradient updates. As a result, the agent is not only optimized for a reward but learns using problem-focused quantities provided by MERL, applicable out-of-the-box to any task. In this paper: (a) We introduce and define MERL, the multi-head reinforcement learning framework we use throughout this work. (b) We conduct experiments across a variety of standard benchmark environments, including 9 continuous control tasks, where results show improved performance. (c) We demonstrate that MERL also improves transfer learning on a set of challenging pixel-based tasks. (d) We ponder how MERL tackles the problem of reward sparsity and better conditions the feature space of reinforcement learning agents.

Yannis Flet-Berliac, Philippe Preux

In reinforcement learning, policy gradient algorithms optimize the policy directly and rely on sampling efficiently an environment. Nevertheless, while most sampling procedures are based on direct policy sampling, self-performance measures could be used to improve such sampling prior to each policy update. Following this line of thought, we introduce SAUNA, a method where non-informative transitions are rejected from the gradient update. The level of information is estimated according to the fraction of variance explained by the value function: a measure of the discrepancy between V and the empirical returns. In this work, we use this metric to select samples that are useful to learn from, and we demonstrate that this selection can significantly improve the performance of policy gradient methods. In this paper: (a) We define SAUNA's metric and introduce its method to filter transitions. (b) We conduct experiments on a set of benchmark continuous control problems. SAUNA significantly improves performance. (c) We investigate how SAUNA reliably selects samples with the most positive impact on learning and study its improvement on both performance and sample efficiency.