Recent advancements in large language models (LLMs) have significantly advanced the automation of software development tasks, including code synthesis, program repair, and test generation. More recently, researchers and industry practitioners have developed various autonomous LLM agents to perform end-to-end software development tasks. These agents are equipped with the ability to use tools, run commands, observe feedback from the environment, and plan for future actions. However, the complexity of these agent-based approaches, together with the limited abilities of current LLMs, raises the following question: Do we really have to employ complex autonomous software agents? To attempt to answer this question, we build Agentless -- an agentless approach to automatically solve software development problems. Compared to the verbose and complex setup of agent-based approaches, Agentless employs a simplistic two-phase process of localization followed by repair, without letting the LLM decide future actions or operate with complex tools. Our results on the popular SWE-bench Lite benchmark show that surprisingly the simplistic Agentless is able to achieve both the highest performance (27.33%) and lowest cost (\$0.34) compared with all existing open-source software agents! Furthermore, we manually classified the problems in SWE-bench Lite and found problems with exact ground truth patch or insufficient/misleading issue descriptions. As such, we construct SWE-bench Lite-S by excluding such problematic issues to perform more rigorous evaluation and comparison. Our work highlights the current overlooked potential of a simple, interpretable technique in autonomous software development. We hope Agentless will help reset the baseline, starting point, and horizon for autonomous software agents, and inspire future work along this crucial direction.

The development of autonomous agents increasingly relies on Multimodal Language Models (MLMs) to perform tasks described in natural language with GUI environments, such as websites, desktop computers, or mobile phones. Existing benchmarks for MLM agents in interactive environments are limited by their focus on a single environment, lack of detailed and generalized evaluation methods, and the complexities of constructing tasks and evaluators. To overcome these limitations, we introduce Crab, the first agent benchmark framework designed to support cross-environment tasks, incorporating a graph-based fine-grained evaluation method and an efficient mechanism for task and evaluator construction. Our framework supports multiple devices and can be easily extended to any environment with a Python interface. Leveraging Crab, we developed a cross-platform Crab Benchmark-v0 comprising 100 tasks in computer desktop and mobile phone environments. We evaluated four advanced MLMs using different single and multi-agent system configurations on this benchmark. The experimental results demonstrate that the single agent with GPT-4o achieves the best completion ratio of 35.26%. All framework code, agent code, and task datasets are publicly available at https://github.com/camel-ai/crab.

Large language models have demonstrated their capabilities in storyline creation and human-like character role-playing. Current language model agents mainly focus on reasonable behaviors from the level of individuals, and their behaviors might be hard to constraint on the level of the whole storyline. In this paper we introduce IBSEN, a director-actor coordinate agent framework that generates drama scripts and makes the plot played by agents more controllable. The director agent writes plot outlines that the user desires to see, instructs the actor agents to role-play their characters, and reschedules the plot when human players participate in the scenario to ensure the plot is progressing towards the objective. To evaluate the framework, we create a novel drama plot that involves several actor agents and check the interactions between them under the instruction of the director agent. Evaluation results show that our framework could generate complete, diverse drama scripts from only a rough outline of plot objectives, meanwhile maintaining the characteristics of characters in the drama. Our codes and prompts are available at https://github.com/OpenDFM/ibsen.

The sustainable foraging problem is a dynamic environment testbed for exploring the forms of agent cognition in dealing with social dilemmas in a multi-agent setting. The agents need to resist the temptation of individual rewards through foraging and choose the collective long-term goal of sustainability. We investigate methods of online learning in Neuro-Evolution and Deep Recurrent Q-Networks to enable agents to attempt the problem one-shot as is often required by wicked social problems. We further explore if learning temporal dependencies with Long Short-Term Memory may be able to aid the agents in developing sustainable foraging strategies in the long term. It was found that the integration of Long Short-Term Memory assisted agents in developing sustainable strategies for a single agent, however failed to assist agents in managing the social dilemma that arises in the multi-agent scenario.

AI agents are an exciting new research direction, and agent development is driven by benchmarks. Our analysis of current agent benchmarks and evaluation practices reveals several shortcomings that hinder their usefulness in real-world applications. First, there is a narrow focus on accuracy without attention to other metrics. As a result, SOTA agents are needlessly complex and costly, and the community has reached mistaken conclusions about the sources of accuracy gains. Our focus on cost in addition to accuracy motivates the new goal of jointly optimizing the two metrics. We design and implement one such optimization, showing its potential to greatly reduce cost while maintaining accuracy. Second, the benchmarking needs of model and downstream developers have been conflated, making it hard to identify which agent would be best suited for a particular application. Third, many agent benchmarks have inadequate holdout sets, and sometimes none at all. This has led to agents that are fragile because they take shortcuts and overfit to the benchmark in various ways. We prescribe a principled framework for avoiding overfitting. Finally, there is a lack of standardization in evaluation practices, leading to a pervasive lack of reproducibility. We hope that the steps we introduce for addressing these shortcomings will spur the development of agents that are useful in the real world and not just accurate on benchmarks.

Explicit communication is often valued for its directness during interaction. Implicit communication, on the other hand, is indirect in that its communicative content must be inferred. Implicit communication is considered more desirable in teaming situations that requires reduced interruptions for improved fluency. In this paper, we investigate another unique advantage of implicit communication: its ability to manipulate the perception of object or behavior of interest. When communication results in the perception of an object or behavior to deviate from other information (about the object or behavior) available via observation, it introduces a discrepancy between perception and observation. We show that such a discrepancy in visual perception can benefit human-robot interaction in a controlled manner and introduce an approach referred to as active shadowing (ASD). Through user studies, we demonstrate the effectiveness of active shadowing in creating a misaligned perception of the robot's behavior and its execution in the real-world, resulting in more efficient task completion without sacrificing its understandability. We also analyze conditions under which such visual manipulation is effective.

Offline multi-agent reinforcement learning (MARL) leverages static datasets of experience to learn optimal multi-agent control. However, learning from static data presents several unique challenges to overcome. In this paper, we focus on coordination failure and investigate the role of joint actions in multi-agent policy gradients with offline data, focusing on a common setting we refer to as the 'Best Response Under Data' (BRUD) approach. By using two-player polynomial games as an analytical tool, we demonstrate a simple yet overlooked failure mode of BRUD-based algorithms, which can lead to catastrophic coordination failure in the offline setting. Building on these insights, we propose an approach to mitigate such failure, by prioritising samples from the dataset based on joint-action similarity during policy learning and demonstrate its effectiveness in detailed experiments. More generally, however, we argue that prioritised dataset sampling is a promising area for innovation in offline MARL that can be combined with other effective approaches such as critic and policy regularisation. Importantly, our work shows how insights drawn from simplified, tractable games can lead to useful, theoretically grounded insights that transfer to more complex contexts. A core dimension of offering is an interactive notebook, from which almost all of our results can be reproduced, in a browser.

While numerous works have assessed the generative performance of language models (LMs) on tasks requiring Theory of Mind reasoning, research into the models' internal representation of mental states remains limited. Recent work has used probing to demonstrate that LMs can represent beliefs of themselves and others. However, these claims are accompanied by limited evaluation, making it difficult to assess how mental state representations are affected by model design and training choices. We report an extensive benchmark with various LM types with different model sizes, fine-tuning approaches, and prompt designs to study the robustness of mental state representations and memorisation issues within the probes. Our results show that the quality of models' internal representations of the beliefs of others increases with model size and, more crucially, with fine-tuning. We are the first to study how prompt variations impact probing performance on theory of mind tasks. We demonstrate that models' representations are sensitive to prompt variations, even when such variations should be beneficial. Finally, we complement previous activation editing experiments on Theory of Mind tasks and show that it is possible to improve models' reasoning performance by steering their activations without the need to train any probe.

With the remarkable advancements of large language models (LLMs), LLM-based agents have become a research hotspot in human-computer interaction. However, there is a scarcity of benchmarks available for LLM-based mobile agents. Benchmarking these agents generally faces three main challenges: (1) The inefficiency of UI-only operations imposes limitations to task evaluation. (2) Specific instructions within a singular application lack adequacy for assessing the multi-dimensional reasoning and decision-making capacities of LLM mobile agents. (3) Current evaluation metrics are insufficient to accurately assess the process of sequential actions. To this end, we propose Mobile-Bench, a novel benchmark for evaluating the capabilities of LLM-based mobile agents. First, we expand conventional UI operations by incorporating 103 collected APIs to accelerate the efficiency of task completion. Subsequently, we collect evaluation data by combining real user queries with augmentation from LLMs. To better evaluate different levels of planning capabilities for mobile agents, our data is categorized into three distinct groups: SAST, SAMT, and MAMT, reflecting varying levels of task complexity. Mobile-Bench comprises 832 data entries, with more than 200 tasks specifically designed to evaluate multi-APP collaboration scenarios. Furthermore, we introduce a more accurate evaluation metric, named CheckPoint, to assess whether LLM-based mobile agents reach essential points during their planning and reasoning steps.

Modeling crowds has many important applications in games and computer animation. Inspired by the emergent following effect in real-life crowd scenarios, in this work, we develop a method for implicitly grouping moving agents. We achieve this by analyzing local information around each agent and rotating its preferred velocity accordingly. Each agent could automatically form an implicit group with its neighboring agents that have similar directions. In contrast to an explicit group, there are no strict boundaries for an implicit group. If an agent's direction deviates from its group as a result of positional changes, it will autonomously exit the group or join another implicitly formed neighboring group. This implicit grouping is autonomously emergent among agents rather than deliberately controlled by the algorithm. The proposed method is compared with many crowd simulation models, and the experimental results indicate that our approach achieves the lowest congestion levels in some classic scenarios. In addition, we demonstrate that adjusting the preferred velocity of agents can actually reduce the dissimilarity between their actual velocity and the original preferred velocity. Our work is available online.