Recent advancements in long-context language modeling have attracted significant attention, yet their practical applications often suffer from suboptimal context utilization. To efficiently address this issue, we introduce the Structured Packing for Long Context, SPLiCe, a method that uses retrieval to collate mutually relevant documents into long training samples. We demonstrate that SPLiCe improves performance on long-context tasks, particularly by achieving perfect accuracy on the synthetic Needle in the Haystack benchmark, and effectively mitigating the ‘lost-in-the-middle’ phenomenon often observed in large language models. Notably, these long-context capabilities also extend to realistic downstream tasks, such as Qasper, across multiple model sizes—3B, 7B, and 13B—and are achieved with only brief fine-tuning on 2-6 billion tokens. We supplement these results with a detailed analysis of SPLiCe, examining the impact of hyperparameter choices, the different mixtures and proportions of SPLiCe-generated training data, and the choice of the retriever. We also study the transfer of long-context utilization skills between the modalities. An intriguing finding from our analysis is that training on a corpus of code can enhance performance on natural language tasks.

Deep neural networks often forget previously learned information when trained sequentially on new objectives, a phenomenon known as catastrophic forgetting. Existing generative strategies combat this issue by randomly sampling rehearsal examples from a generative model. Such an approach contradicts buffer-based approaches where sampling strategy plays an important role. We propose to bridge this gap and benefit from the combination of DDPM trained on the previous task and the classifier guidance technique to actively generate rehearsal examples specifically designed to minimize forgetting in the currently trained classifier. Our experimental results show that GUIDE significantly reduces catastrophic forgetting, outperforming conventional random sampling approaches and surpassing recent state-of-the-art methods in continual learning with generative replay, and buffer-based rehearsal.

Self-supervision has the potential to transform reinforcement learning (RL), paralleling the breakthroughs it has enabled in other areas of machine learning. While self-supervised learning in other domains aims to find patterns in a fixed dataset, self-supervised goal-conditioned reinforcement learning (GCRL) agents discover *new* behaviors by learning from the goals achieved during unstructured interaction with the environment. However, these methods have failed to see similar success, both due to a lack of data from slow environment simulations as well as a lack of stable algorithms. We take a step toward addressing both of these issues by releasing a high-performance codebase and benchmark (`JaxGCRL`) for self-supervised GCRL, enabling researchers to train agents for millions of environment steps in minutes on a single GPU. By utilizing GPU-accelerated replay buffers, environments, and a stable contrastive RL algorithm, we reduce training time by up to $22\times$. Additionally, we assess key design choices in contrastive RL, identifying those that most effectively stabilize and enhance training performance. With this approach, we provide a foundation for future research in self-supervised GCRL, enabling researchers to quickly iterate on new ideas and evaluate them in diverse and challenging environments. Code: [https://anonymous.4open.science/r/JaxGCRL-2316/README.md](https://anonymous.4open.science/r/JaxGCRL-2316/README.md)

Large Language Models (LLMs) and Vision Language Models (VLMs) possess extensive knowledge and exhibit promising reasoning abilities, however, they still struggle to perform well in complex, dynamic environments. Real-world tasks require handling intricate interactions, advanced spatial reasoning, long-term planning, and continuous exploration of new strategies—areas in which we lack effective methodologies for comprehensively evaluating these capabilities. To address this gap, we introduce BALROG, a novel benchmark designed to assess the agentic capabilities of LLMs and VLMs through a diverse set of challenging games. Our benchmark incorporates a range of existing reinforcement learning environments with varying levels of difficulty, including tasks that are solvable by non-expert humans in seconds to extremely challenging ones that may take years to master (e.g., the NetHack Learning Environment). We devise fine-grained metrics to measure performance and conduct an extensive evaluation of several popular open-source and closed-source LLMs and VLMs. Our findings indicate that while current models achieve partial success in the easier games, they struggle significantly with more challenging tasks. Notably, we observe severe deficiencies in vision-based decision-making, as several models perform worse when visual representations of the environments are provided. We release BALROG as an open and user-friendly benchmark to facilitate future research and development in the agentic community. Code and Leaderboard at balrogai.com

This paper presents a novel approach to premise selection, a crucial reasoning task in automated theorem proving. Traditionally, symbolic methods that rely on extensive domain knowledge and engineering effort are applied to this task. In contrast, this work demonstrates that contrastive training with the transformer architecture can achieve higher-quality retrieval of relevant premises, without the knowledge or feature engineering overhead. Our method, Magnushammer, outperforms the most advanced and widely used automation tool in interactive theorem proving called Sledgehammer. On the PISA and miniF2f benchmarks Magnushammer achieves $59.5\%$ (against $38.3\%$) and $34.0\%$ (against $20.9\%$) success rates, respectively. By combining Magnushammer with a language-model-based automated theorem prover, we further improve the state-of-the-art proof success rate from $57.0\%$ to $71.0\%$ on the PISA benchmark using $4$x fewer parameters. Moreover, we develop and open source a novel dataset for premise selection, containing textual representations of (proof state, relevant premise) pairs. To the best of our knowledge, this is the largest available premise selection dataset, and the first dataset of this kind for the Isabelle proof assistant.

Fine-tuning is a widespread technique that allows practitioners to transfer pre-trained capabilities, as recently showcased by the successful applications of foundation models. However, fine-tuning reinforcement learning (RL) models remains a challenge. This work conceptualizes one specific cause of poor transfer, accentuated in the RL setting by the interplay between actions and observations: forgetting of pre-trained capabilities. Namely, a model deteriorates on the state subspace of the downstream task not visited in the initial phase of fine-tuning, on which the model behaved well due to pre-training. This way, we lose the anticipated transfer benefits. We identify conditions when this problem occurs, showing that it is common and, in many cases, catastrophic. Through a detailed empirical analysis of the challenging NetHack and Montezuma’s Revenge environments, we show that standard knowledge retention techniques mitigate the problem and thus allow us to take full advantage of the pre-trained capabilities. In particular, in NetHack, we achieve a new state-of-the-art for neural models, improving the previous best score from $5$K to over $10$K points in the Human Monk scenario.

The ability of continual learning systems to transfer knowledge from previously seen tasks in order to maximize performance on new tasks is a significant challenge for the field, limiting the applicability of continual learning solutions to realistic scenarios. Consequently, this study aims to broaden our understanding of transfer and its driving forces in the specific case of continual reinforcement learning. We adopt SAC as the underlying RL algorithm and Continual World as a suite of continuous control tasks. We systematically study how different components of SAC (the actor and the critic, exploration, and data) affect transfer efficacy, and we provide recommendations regarding various modeling options. The best set of choices, dubbed ClonEx-SAC, is evaluated on the recent Continual World benchmark. ClonEx-SAC achieves 87% final success rate compared to 80% of PackNet, the best method in the benchmark. Moreover, the transfer grows from 0.18 to 0.54 according to the metric provided by Continual World.

Planning in large state spaces inevitably needs to balance the depth and breadth of the search. It has a crucial impact on the performance of a planner and most manage this interplay implicitly. We present a novel method Shoot Tree Search (STS), which makes it possible to control this trade-off more explicitly. Our algorithm can be understood as an interpolation between two celebrated search mechanisms: MCTS and random shooting. It also lets the user control the bias-variance trade-off, akin to TD(n), but in the tree search context. In experiments on challenging domains, we show that STS can get the best of both worlds consistently achieving higher scores.

Humans excel in solving complex reasoning tasks through a mental process of moving from one idea to a related one. Inspired by this, we propose Subgoal Search (kSubS) method. Its key component is a learned subgoal generator that produces a diversity of subgoals that are both achievable and closer to the solution. Using subgoals reduces the search space and induces a high-level search graph suitable for efficient planning. In this paper, we implement kSubS using a transformer-based subgoal module coupled with the classical best-first search framework. We show that a simple approach of generating $k$-th step ahead subgoals is surprisingly efficient on three challenging domains: two popular puzzle games, Sokoban and the Rubik's Cube, and an inequality proving benchmark INT. kSubS achieves strong results including state-of-the-art on INT within a modest computational budget.

Communication is compositional if complex signals can be represented as a combination of simpler subparts. In this paper, we theoretically show that inductive biases on both the training framework and the data are needed to develop a compositional communication. Moreover, we prove that compositionality spontaneously arises in the signaling games, where agents communicate over a noisy channel. We experimentally confirm that a range of noise levels, which depends on the model and the data, indeed promotes compositionality. Finally, we provide a comprehensive study of this dependence and report results in terms of recently studied compositionality metrics: topographical similarity, conflict count, and context independence.

Continual learning (CL) --- the ability to continuously learn, building on previously acquired knowledge --- is a natural requirement for long-lived autonomous reinforcement learning (RL) agents. While building such agents, one needs to balance opposing desiderata, such as constraints on capacity and compute, the ability to not catastrophically forget, and to exhibit positive transfer on new tasks. Understanding the right trade-off is conceptually and computationally challenging, which we argue has led the community to overly focus on catastrophic forgetting. In response to these issues, we advocate for the need to prioritize forward transfer and propose Continual World, a benchmark consisting of realistic and meaningfully diverse robotic tasks built on top of Meta-World as a testbed. Following an in-depth empirical evaluation of existing CL methods, we pinpoint their limitations and highlight unique algorithmic challenges in the RL setting. Our benchmark aims to provide a meaningful and computationally inexpensive challenge for the community and thus help better understand the performance of existing and future solutions. Information about the benchmark, including the open-source code, is available at https://sites.google.com/view/continualworld.