Despite enthusiasm for Multi-Agent LLM Systems (MAS), their performance gains on popular benchmarks are often minimal. This gap highlights a critical need for a principled understanding of why MAS fail. Addressing this question requires systematic identification and analysis of failure patterns. We introduce MAST-Data, a comprehensive dataset of 1600+ annotated traces collected across 7 popular MAS frameworks. MAST-Data is the first multi-agent system dataset to outline the failure dynamics in MAS for guiding the development of better future systems. To enable systematic classification of failures for MAST-Data, we build the first Multi-Agent System Failure Taxonomy (MAST). We develop MAST through rigorous analysis of 150 traces, guided closely by expert human annotators andvalidated by high inter-annotator agreement (κ = 0.88). This process identifies 14 unique modes, clustered into 3 categories: (i) system design issues, (ii) inter-agent misalignment, and (iii) task verification. To enable scalable annotation, we develop an LLM-as-a-Judge pipeline with high agreement with human annotations. We leverage MAST and MAST-Data to analyze failure patterns across models (GPT4, Claude 3, Qwen2.5, CodeLlama) and tasks (coding, math, general agent), demonstrating improvement headrooms from better MAS design. Our analysis provides insights revealing that identified failures require more sophisticated solutions, highlighting a clear roadmap for future research. We publicly release our comprehensive dataset (MAST-Data), the MAST, and our LLM annotator to facilitate widespread research and development in MAS.

Pre-trained vision language models (VLMs), though powerful, typically lack training on decision-centric data, rendering them sub-optimal for decision-making tasks such as in-the-wild device control through Graphical User Interfaces (GUIs) when used off-the-shelf. While training with static demonstrations has shown some promise, we show that such methods fall short when controlling real GUIs due to their failure to deal with real world stochasticity and dynamism not captured in static observational data. This paper introduces a novel autonomous RL approach, called DigiRL, for training in-the-wild device control agents through fine-tuning a pre-trained VLM in two stages: offline and offline-to-online RL. We first build a scalable and parallelizable Android learning environment equipped with a VLM-based general-purpose evaluator and then identify the key design choices for simple and effective RL in this domain. We demonstrate the effectiveness of DigiRL using the Android-in-the-Wild (AitW) dataset, where our 1.5B VLM trained with RL achieves a 49.5\% absolute improvement -- from 17.7 to 67.2\% success rate -- over supervised fine-tuning with static human demonstration data. It is worth noting that such improvement is achieved without any additional supervision or demonstration data. These results significantly surpass not only the prior best agents, including AppAgent with GPT-4V (8.3\% success rate) and the 17B CogAgent trained with AitW data (14.4\%), but also our implementation of prior best autonomous RL approach based on filtered behavior cloning (57.8\%), thereby establishing a new state-of-the-art for digital agents for in-the-wild device control.

Social learning algorithms provide a model for the formation and propagation of opinions over social networks. However, most studies focus on the case in which agents share their information synchronously over regular intervals. In this work, we analyze belief convergence and steady-state learning performance for both traditional and adaptive formulations of social learning under asynchronous behavior by the agents, where some of the agents may decide to abstain from sharing any information with the network at some time instants. We also show how to recover the underlying graph topology from observations of the asynchronous network behavior.