Robotic-assisted surgery (RAS) introduces complex challenges that current surgical error detection methods struggle to address effectively due to limited training data and methodological constraints. Therefore, we construct MERP (Multi-class Error in Robotic Prostatectomy), a comprehensive dataset for error detection in robotic prostatectomy with frame-level annotations featuring six clinically aligned error categories. In addition, we propose CARES (Collaborative Agentic Reasoning for Error Detection in Surgery), a novel zero-shot clinically-informed and risk-stratified agentic reasoning architecture for multi-class surgical error detection. CARES implements adaptive generation of medically informed, error-specific Chain-of-Thought (CoT) prompts across multiple expertise levels. The framework employs risk-aware routing to assign error task to expertise-matched reasoning pathways based on complexity and clinical impact. Subsequently, each pathway decomposes surgical error analysis into three specialized agents with temporal, spatial, and procedural analysis. Each agent analyzes using dynamically selected prompts tailored to the assigned expertise level and error type, generating detailed and transparent reasoning traces. By incorporating clinically informed reasoning from established surgical assessment guidelines, CARES enables zero-shot surgical error detection without prior training. Evaluation demonstrates superior performance with 54.3 mF1 on RARP and 52.0 mF1 on MERP datasets, outperforming existing zero-shot approaches by up to 14% while remaining competitive with trained models. Ablation studies demonstrate the effectiveness of our method. The dataset and code will be publicly available.

Vision-language models (VLMs) have made significant strides in reasoning, yet they often struggle with complex multimodal tasks and tend to generate overly verbose outputs. A key limitation is their reliance on chain-of-thought (CoT) reasoning, despite many tasks benefiting from alternative topologies like trees or graphs. To address this, we introduce STELAR-Vision, a training framework for topology-aware reasoning. At its core is TopoAug, a synthetic data pipeline that enriches training with diverse topological structures. Using supervised fine-tuning and reinforcement learning, we post-train Qwen2VL models with both accuracy and efficiency in mind. Additionally, we propose Frugal Learning, which reduces output length with minimal accuracy loss. On MATH-V and VLM-S2H, STELAR-Vision improves accuracy by 9.7% over its base model and surpasses the larger Qwen2VL-72B-Instruct by 7.3%. On five out-of-distribution benchmarks, it outperforms Phi-4-Multimodal-Instruct by up to 28.4% and LLaMA-3.2-11B-Vision-Instruct by up to 13.2%, demonstrating strong generalization. Compared to Chain-Only training, our approach achieves 4.3% higher overall accuracy on in-distribution datasets and consistently outperforms across all OOD benchmarks. We have released datasets, and code will be available.

LLMs have been shown to perform well in machine translation (MT) with the use of in-context learning (ICL), rivaling supervised models when translating into high-resource languages (HRLs). However, they lag behind when translating into low-resource language (LRLs). Example selection via similarity search and supervised fine-tuning help. However the improvements they give are limited by the size, quality and diversity of existing parallel datasets. A common technique in low-resource MT is synthetic parallel data creation, the most frequent of which is backtranslation, whereby existing target-side texts are automatically translated into the source language. However, this assumes the existence of good quality and relevant target-side texts, which are not readily available for many LRLs. In this paper, we present \textsc{TopXGen}, an LLM-based approach for the generation of high quality and topic-diverse data in multiple LRLs, which can then be backtranslated to produce useful and diverse parallel texts for ICL and fine-tuning. Our intuition is that while LLMs struggle to translate into LRLs, their ability to translate well into HRLs and their multilinguality enable them to generate good quality, natural-sounding target-side texts, which can be translated well into a high-resource source language. We show that \textsc{TopXGen} boosts LLM translation performance during fine-tuning and in-context learning. Code and outputs are available at https://github.com/ArmelRandy/topxgen.

Transforming unstructured text into structured data is a complex task, requiring semantic understanding, reasoning, and structural comprehension. While Large Language Models (LLMs) offer potential, they often struggle with handling ambiguous or domain-specific data, maintaining table structure, managing long inputs, and addressing numerical reasoning. This paper proposes an efficient system for LLM-driven text-to-table generation that leverages novel prompting techniques. Specifically, the system incorporates two key strategies: breaking down the text-to-table task into manageable, guided sub-tasks and refining the generated tables through iterative self-feedback. We show that this custom task decomposition allows the model to address the problem in a stepwise manner and improves the quality of the generated table. Furthermore, we discuss the benefits and potential risks associated with iterative self-feedback on the generated tables while highlighting the trade-offs between enhanced performance and computational cost. Our methods achieve strong results compared to baselines on two complex text-to-table generation datasets available in the public domain.

Accurate evaluation of procedural communication compliance is essential in simulation-based training, particularly in safety-critical domains where adherence to compliance checklists reflects operational competence. This paper explores a lightweight, deployable approach using prompt-based inference with open-source large language models (LLMs) that can run efficiently on consumer-grade GPUs. We present Prompt-and-Check, a method that uses context-rich prompts to evaluate whether each checklist item in a protocol has been fulfilled, solely based on transcribed verbal exchanges. We perform a case study in the maritime domain with participants performing an identical simulation task, and experiment with models such as LLama 2 7B, LLaMA 3 8B and Mistral 7B, running locally on an RTX 4070 GPU. For each checklist item, a prompt incorporating relevant transcript excerpts is fed into the model, which outputs a compliance judgment. We assess model outputs against expert-annotated ground truth using classification accuracy and agreement scores. Our findings demonstrate that prompting enables effective context-aware reasoning without task-specific training. This study highlights the practical utility of LLMs in augmenting debriefing, performance feedback, and automated assessment in training environments.

This paper presents our system built for the WASSA-2024 Cross-lingual Emotion Detection Shared Task. The task consists of two subtasks: first, to assess an emotion label from six possible classes for a given tweet in one of five languages, and second, to predict words triggering the detected emotions in binary and numerical formats. Our proposed approach revolves around fine-tuning quantized large language models, specifically Orca~2, with low-rank adapters (LoRA) and multilingual Transformer-based models, such as XLM-R and mT5. We enhance performance through machine translation for both subtasks and trigger word switching for the second subtask. The system achieves excellent performance, ranking 1st in numerical trigger words detection, 3rd in binary trigger words detection, and 7th in emotion detection.

While large language models (LLMs) show promise for various tasks, their performance in compound aspect-based sentiment analysis (ABSA) tasks lags behind fine-tuned models. However, the potential of LLMs fine-tuned for ABSA remains unexplored. This paper examines the capabilities of open-source LLMs fine-tuned for ABSA, focusing on LLaMA-based models. We evaluate the performance across four tasks and eight English datasets, finding that the fine-tuned Orca~2 model surpasses state-of-the-art results in all tasks. However, all models struggle in zero-shot and few-shot scenarios compared to fully fine-tuned ones. Additionally, we conduct error analysis to identify challenges faced by fine-tuned models.

As multimodal large language models advance rapidly, the automation of mobile tasks has become increasingly feasible through the use of mobile-use agents that mimic human interactions from graphical user interface. To further enhance mobile-use agents, previous studies employ demonstration learning to improve mobile-use agents from human demonstrations. However, these methods focus solely on the explicit intention flows of humans (e.g., step sequences) while neglecting implicit intention flows (e.g., personal preferences), which makes it difficult to construct personalized mobile-use agents. In this work, to evaluate the I ntention A lignment R ate between mobile-use agents and humans, we first collect Mo-bileIAR, a dataset containing human-intent-aligned actions and ground-truth actions. This enables a comprehensive assessment of the agents' understanding of human intent. Then we propose IFRAgent, a framework built upon I ntention Flow R ecognition from human demonstrations. IFRAgent analyzes explicit intention flows from human demonstrations to construct a query-level vector library of standard operating procedures (SOP), and analyzes implicit intention flows to build a user-level habit repository. IFRAgent then leverages a SOP extractor combined with retrieval-augmented generation and a query rewriter to generate personalized query and SOP from a raw ambiguous query, enhancing the alignment between mobile-use agents and human intent. Experimental results demonstrate that IFRAgent outperforms baselines by an average of 6.79% (32.06% relative improvement) in human intention alignment rate and improves step completion rates by an average of 5.30% (26.34% relative improvement).

Large language models (LLMs) are increasingly being applied to black-box optimization tasks, from program synthesis to molecule design. Prior work typically leverages in-context learning to iteratively guide the model towards better solutions. Such methods, however, often struggle to balance exploration of new solution spaces with exploitation of high-reward ones. Recently, test-time training (TTT) with synthetic data has shown promise in improving solution quality. However, the need for hand-crafted training data tailored to each task limits feasibility and scalability across domains. To address this problem, we introduce MiGrATe-a method for online TTT that uses GRPO as a search algorithm to adapt LLMs at inference without requiring external training data. MiGrATe operates via a mixed-policy group construction procedure that combines on-policy sampling with two off-policy data selection techniques: greedy sampling, which selects top-performing past completions, and neighborhood sampling (NS), which generates completions structurally similar to high-reward ones. Together, these components bias the policy gradient towards exploitation of promising regions in solution space, while preserving exploration through on-policy sampling. We evaluate MiGrATe on three challenging domains-word search, molecule optimization, and hypothesis+program induction on the Abstraction and Reasoning Corpus (ARC)-and find that it consistently outperforms both inference-only and TTT baselines, demonstrating the potential of online TTT as a solution for complex search tasks without external supervision.

Large language models (LLMs) have revolutionized artificial intelligence by enabling complex reasoning capabilities. While recent advancements in reinforcement learning (RL) have primarily focused on domain-specific reasoning tasks (e.g., mathematics or code generation), real-world reasoning scenarios often require models to handle diverse and complex environments that narrow-domain benchmarks cannot fully capture. To address this gap, we present InternBootcamp, an open-source framework comprising 1000+ domain-diverse task environments specifically designed for LLM reasoning research. Our codebase offers two key functionalities: (1) automated generation of unlimited training/testing cases with configurable difficulty levels, and (2) integrated verification modules for objective response evaluation. These features make InternBootcamp fundamental infrastructure for RL-based model optimization, synthetic data generation, and model evaluation. Although manually developing such a framework with enormous task coverage is extremely cumbersome, we accelerate the development procedure through an automated agent workflow supplemented by manual validation protocols, which enables the task scope to expand rapidly. % With these bootcamps, we further establish Bootcamp-EVAL, an automatically generated benchmark for comprehensive performance assessment. Evaluation reveals that frontier models still underperform in many reasoning tasks, while training with InternBootcamp provides an effective way to significantly improve performance, leading to our 32B model that achieves state-of-the-art results on Bootcamp-EVAL and excels on other established benchmarks. In particular, we validate that consistent performance gains come from including more training tasks, namely \textbf{task scaling}, over two orders of magnitude, offering a promising route towards capable reasoning generalist.