The PerAnsSumm 2025 challenge focuses on perspective-aware healthcare answer summarization (Agarwal et al., 2025). This work proposes a few-shot learning framework using a Snorkel-BART-SVM pipeline for classifying and summarizing open-ended healthcare community question-answering (CQA). An SVM model is trained with weak supervision via Snorkel, enhancing zero-shot learning. Extractive classification identifies perspective-relevant sentences, which are then summarized using a pretrained BART-CNN model. The approach achieved 12th place among 100 teams in the shared task, demonstrating computational efficiency and contextual accuracy. By leveraging pretrained summarization models, this work advances medical CQA research and contributes to clinical decision support systems.

Recent advances in Large Language Models (LLMs) have permitted the development of language-guided multi-robot systems, which allow robots to execute tasks based on natural language instructions. However, achieving effective coordination in distributed multi-agent environments remains challenging due to (1) misalignment between instructions and task requirements and (2) inconsistency in robot behaviors when they independently interpret ambiguous instructions. To address these challenges, we propose Instruction-Conditioned Coordinator (ICCO), a Multi-Agent Reinforcement Learning (MARL) framework designed to enhance coordination in language-guided multi-robot systems. ICCO consists of a Coordinator agent and multiple Local Agents, where the Coordinator generates Task-Aligned and Consistent Instructions (TACI) by integrating language instructions with environmental states, ensuring task alignment and behavioral consistency. The Coordinator and Local Agents are jointly trained to optimize a reward function that balances task efficiency and instruction following. A Consistency Enhancement Term is added to the learning objective to maximize mutual information between instructions and robot behaviors, further improving coordination. Simulation and real-world experiments validate the effectiveness of ICCO in achieving language-guided task-aligned multi-robot control. The demonstration can be found at https://yanoyoshiki.github.io/ICCO/.

Ensuring safe and effective use of AI requires understanding and anticipating its performance on novel tasks, from advanced scientific challenges to transformed workplace activities. So far, benchmarking has guided progress in AI, but it has offered limited explanatory and predictive power for general-purpose AI systems, given the low transferability across diverse tasks. In this paper, we introduce general scales for AI evaluation that can explain what common AI benchmarks really measure, extract ability profiles of AI systems, and predict their performance for new task instances, in- and out-of-distribution. Our fully-automated methodology builds on 18 newly-crafted rubrics that place instance demands on general scales that do not saturate. Illustrated for 15 large language models and 63 tasks, high explanatory power is unleashed from inspecting the demand and ability profiles, bringing insights on the sensitivity and specificity exhibited by different benchmarks, and how knowledge, metacognition and reasoning are affected by model size, chain-of-thought and distillation. Surprisingly, high predictive power at the instance level becomes possible using these demand levels, providing superior estimates over black-box baseline predictors based on embeddings or finetuning, especially in out-of-distribution settings (new tasks and new benchmarks). The scales, rubrics, battery, techniques and results presented here represent a major step for AI evaluation, underpinning the reliable deployment of AI in the years ahead. (Collaborative platform: https://kinds-of-intelligence-cfi.github.io/ADELE.)

Image captioning has been a longstanding challenge in vision-language research. With the rise of LLMs, modern Vision-Language Models (VLMs) generate detailed and comprehensive image descriptions. However, benchmarking the quality of such captions remains unresolved. This paper addresses two key questions: (1) How well do current VLMs actually perform on image captioning, particularly compared to humans? We built CapArena, a platform with over 6000 pairwise caption battles and high-quality human preference votes. Our arena-style evaluation marks a milestone, showing that leading models like GPT-4o achieve or even surpass human performance, while most open-source models lag behind. (2) Can automated metrics reliably assess detailed caption quality? Using human annotations from CapArena, we evaluate traditional and recent captioning metrics, as well as VLM-as-a-Judge. Our analysis reveals that while some metrics (e.g., METEOR) show decent caption-level agreement with humans, their systematic biases lead to inconsistencies in model ranking. In contrast, VLM-as-a-Judge demonstrates robust discernment at both the caption and model levels. Building on these insights, we release CapArena-Auto, an accurate and efficient automated benchmark for detailed captioning, achieving 94.3% correlation with human rankings at just $4 per test. Data and resources will be open-sourced at https://caparena.github.io.

As the open-weight AI landscape continues to proliferate-with model development, significant investment, and user interest-it becomes increasingly important to predict which models will ultimately drive innovation and shape AI ecosystems. Building on parallels with citation dynamics in scientific literature, we propose a framework to quantify how an open-weight model's influence evolves. Specifically, we adapt the model introduced by Wang et al. for scientific citations, using three key parameters-immediacy, longevity, and relative fitness-to track the cumulative number of fine-tuned models of an open-weight model. Our findings reveal that this citation-style approach can effectively capture the diverse trajectories of open-weight model adoption, with most models fitting well and outliers indicating unique patterns or abrupt jumps in usage.

The Mixture of Experts (MoE) architecture reduces the training and inference cost significantly compared to a dense model of equivalent capacity. Upcycling is an approach that initializes and trains an MoE model using a pre-trained dense model. While upcycling leads to initial performance gains, the training progresses slower than when trained from scratch, leading to suboptimal performance in the long term. We propose Drop-Upcycling - a method that effectively addresses this problem. Drop-Upcycling combines two seemingly contradictory approaches: utilizing the knowledge of pre-trained dense models while statistically re-initializing some parts of the weights. This approach strategically promotes expert specialization, significantly enhancing the MoE model's efficiency in knowledge acquisition. Extensive large-scale experiments demonstrate that Drop-Upcycling significantly outperforms previous MoE construction methods in the long term, specifically when training on hundreds of billions of tokens or more. As a result, our MoE model with 5.9B active parameters achieves comparable performance to a 13B dense model in the same model family, while requiring approximately 1/4 of the training FLOPs. All experimental resources, including source code, training data, model checkpoints and logs, are publicly available to promote reproducibility and future research on MoE.

Large Language Models (LLMs) have significantly advanced natural language processing with exceptional task generalization capabilities. Low-Rank Adaption (LoRA) offers a cost-effective fine-tuning solution, freezing the original model parameters and training only lightweight, low-rank adapter matrices. However, the memory footprint of LoRA is largely dominated by the original model parameters. To mitigate this, we propose LoRAM, a memory-efficient LoRA training scheme founded on the intuition that many neurons in over-parameterized LLMs have low training utility but are essential for inference. LoRAM presents a unique twist: it trains on a pruned (small) model to obtain pruned low-rank matrices, which are then recovered and utilized with the original (large) model for inference. Additionally, minimal-cost continual pre-training, performed by the model publishers in advance, aligns the knowledge discrepancy between pruned and original models. Our extensive experiments demonstrate the efficacy of LoRAM across various pruning strategies and downstream tasks. For a model with 70 billion parameters, LoRAM enables training on a GPU with only 20G HBM, replacing an A100-80G GPU for LoRA training and 15 GPUs for full fine-tuning. Specifically, QLoRAM implemented by structured pruning combined with 4-bit quantization, for LLaMA-3.1-70B (LLaMA-2-70B), reduces the parameter storage cost that dominates the memory usage in low-rank matrix training by 15.81$\times$ (16.95$\times$), while achieving dominant performance gains over both the original LLaMA-3.1-70B (LLaMA-2-70B) and LoRA-trained LLaMA-3.1-8B (LLaMA-2-13B). Code is available at https://github.com/junzhang-zj/LoRAM.

VLMs have the potential to revolutionize driver and in-cabin monitoring by offering a more holistic understanding of the driving scene. Rather than focusing on individual variables, VLMs are trained to describe the entire scene, considering all crucial elements. This comprehensive approach allows them to construct a coherent narrative around the scene, leading to a more thorough assessment of the driver's situation. Despite the potential benefits, there is a notable lack of scientific research exploring the application of VLMs in this field. We aim to conduct an initial exploration of how these systems perform in tasks such as distraction detection, drowsiness detection, and gaze estimation. By evaluating their performance, we hope to determine whether they can match or even surpass state-of-the-art models, or identify areas where they fall short. To achieve this, we will utilize data from the Driver Monitoring Dataset (DMD), which contains extensive material of drivers in various states of drowsiness and distraction containing drivers doing several actions that imply distraction like texting, having a phone call, drinking water, besides driving safely, as well as detailed gaze annotations. By integrating VLMs into DMS, we expect the model to: Have better scene comprehension, enabling it to provide detailed descriptions and respond to queries through Visual Question Answering (VQA) tasks.

In a blog post spotted by The Verge, the company this week published its response to President Trump's AI Action Plan. Announced at the end of February, the initiative saw the White House seek input from private industry, with the goal of eventually enacting policy that will work to "enhance America's position as an AI powerhouse" and enable innovation in the sector. "America's robust, balanced intellectual property system has long been key to our global leadership on innovation. In the same document, the company recommends the US maintain tight export controls on AI chips to China. It also says the US government should broadly adopt AI tools.

The automated future just lurched a few steps closer. Over the past few weeks, nearly all of the major AI firms--OpenAI, Anthropic, Google, xAI, Amazon, Microsoft, and Perplexity, among others--have announced new products that are focused not on answering questions or making their human users somewhat more efficient, but on completing tasks themselves. They are being pitched for their ability to "reason" as people do and serve as "agents" that will eventually carry out complex work from start to finish. Humans will still nudge these models along, of course, but they are engineered to help fewer people do the work of many. Last month, Anthropic launched Claude Code, a coding program that can do much of a human software developer's job but far faster, "reducing development time and overhead."