Generative Zero-Shot Learning approach (GZSL) has demonstrated significant potential in 3D point cloud semantic segmentation tasks. GZSL leverages generative models like GANs or VAEs to synthesize realistic features (real features) of unseen classes. This allows the model to label unseen classes during testing, despite being trained only on seen classes. In this context, we introduce the Generalized Zero-Shot Learning based-upon Mixture-of-Experts (GZSL-MoE) model. This model incorporates Mixture-of-Experts layers (MoE) to generate fake features that closely resemble real features extracted using a pre-trained KPConv (Kernel Point Convolution) model on seen classes. The main contribution of this paper is the integration of Mixture-of-Experts into the Generator and Discriminator components of the Generative Zero-Shot Learning model for 3D point cloud semantic segmentation, applied to the COVERED dataset (CollabOratiVE Robot Environment Dataset) for Human-Robot Collaboration (HRC) environments. By combining the Generative Zero-Shot Learning model with Mixture-of- Experts, GZSL-MoE for 3D point cloud semantic segmentation provides a promising solution for understanding complex 3D environments, especially when comprehensive training data for all object classes is unavailable. The performance evaluation of the GZSL-MoE model highlights its ability to enhance performance on both seen and unseen classes. Keywords Generalized Zero-Shot Learning (GZSL), 3D Point Cloud, 3D Semantic Segmentation, Human-Robot Collaboration, COVERED (CollabOratiVE Robot Environment Dataset), KPConv, Mixture-of Experts

Large Language Models (LLMs) are increasingly deployed across diverse domains but often exhibit disparities in how they handle real-life queries. To systematically investigate these effects within various disability contexts, we introduce \textbf{AccessEval (Accessibility Evaluation)}, a benchmark evaluating 21 closed- and open-source LLMs across 6 real-world domains and 9 disability types using paired Neutral and Disability-Aware Queries. We evaluated model outputs with metrics for sentiment, social perception, and factual accuracy. Our analysis reveals that responses to disability-aware queries tend to have a more negative tone, increased stereotyping, and higher factual error compared to neutral queries. These effects show notable variation by domain and disability type, with disabilities affecting hearing, speech, and mobility disproportionately impacted. These disparities reflect persistent forms of ableism embedded in model behavior. By examining model performance in real-world decision-making contexts, we better illuminate how such biases can translate into tangible harms for disabled users. This framing helps bridges the gap between technical evaluation and user impact, reinforcing importance of bias mitigation in day-to-day applications. Our dataset is publicly available at: https://huggingface.co/datasets/Srikant86/AccessEval

Classical search engines using indexing methods in data infrastructures primarily allow keyword-based queries to retrieve content. While these indexing-based methods are highly scalable and efficient, due to a lack of an appropriate evaluation dataset and a limited understanding of semantics, they often fail to capture the user's intent and generate incomplete responses during evaluation. This problem also extends to domain-specific search systems that utilize a Knowledge Base (KB) to access data from various research infrastructures. Research infrastructures (RIs) from the environmental and earth science domain, which encompass the study of ecosystems, biodiversity, oceanography, and climate change, generate, share, and reuse large volumes of data. While there are attempts to provide a centralized search service using Elasticsearch as a knowledge base, they also face similar challenges in understanding queries with multiple intents. To address these challenges, we proposed an automated method to curate a domain-specific evaluation dataset to analyze the capability of a search system. Furthermore, we incorporate the Retrieval of Augmented Generation (RAG), powered by Large Language Models (LLMs), for high-quality retrieval of environmental domain data using natural language queries. Our quantitative and qualitative analysis of the evaluation dataset shows that LLM-based systems for information retrieval return results with higher precision when understanding queries with multiple intents, compared to Elasticsearch-based systems.

Homonyms are words with identical spelling but distinct meanings, which pose challenges for many generative models. When a homonym appears in a prompt, diffusion models may generate multiple senses of the word simultaneously, which is known as homonym duplication. This issue is further complicated by an Anglocentric bias, which includes an additional translation step before the text-to-image model pipeline. As a result, even words that are not homonymous in the original language may become homonyms and lose their meaning after translation into English. In this paper, we introduce a method for measuring duplication rates and conduct evaluations of different diffusion models using both automatic evaluation utilizing Vision-Language Models (VLM) and human evaluation. Additionally, we investigate methods to mitigate the homonym duplication problem through prompt expansion, demonstrating that this approach also effectively reduces duplication related to Anglocentric bias. The code for the automatic evaluation pipeline is publicly available.

Multimodal agents have demonstrated strong performance in general GUI interactions, but their application in automotive systems has been largely unexplored. In-vehicle GUIs present distinct challenges: drivers' limited attention, strict safety requirements, and complex location-based interaction patterns. To address these challenges, we introduce Automotive-ENV, the first high-fidelity benchmark and interaction environment tailored for vehicle GUIs. This platform defines 185 parameterized tasks spanning explicit control, implicit intent understanding, and safety-aware tasks, and provides structured multimodal observations with precise programmatic checks for reproducible evaluation. Building on this benchmark, we propose ASURADA, a geo-aware multimodal agent that integrates GPS-informed context to dynamically adjust actions based on location, environmental conditions, and regional driving norms. Experiments show that geo-aware information significantly improves success on safety-aware tasks, highlighting the importance of location-based context in automotive environments. We will release Automotive-ENV, complete with all tasks and benchmarking tools, to further the development of safe and adaptive in-vehicle agents.

We experiment with a low-latency, end-to-end voice-to-voice communication model to optimize it for real-time conversational applications. By analyzing components essential to voice to voice (V-2-V) system viz. automatic speech recognition (ASR), text-to-speech (TTS), and dialog management, our work analyzes how to reduce processing time while maintaining high-quality interactions to identify the levers for optimizing V-2-V system. Our work identifies that TTS component which generates life-like voice, full of emotions including natural pauses and exclamations has highest impact on Real time factor (RTF). The experimented V-2-V architecture utilizes CSM1b has the capability to understand tone as well as context of conversation by ingesting both audio and text of prior exchanges to generate contextually accurate speech. We explored optimization of Residual Vector Quantization (RVQ) iterations by the TTS decoder which come at a cost of decrease in the quality of voice generated. Our experimental evaluations also demonstrate that for V-2-V implementations based on CSM most important optimizations can be brought by reducing the number of RVQ Iterations along with the codebooks used in Mimi.

Log analysis is crucial for monitoring system health and diagnosing failures in complex systems. Recent advances in large language models (LLMs) offer new opportunities for automated log analysis, leveraging their reasoning capabilities to perform tasks such as anomaly detection and failure prediction. However, general-purpose LLMs struggle to formulate structured reasoning workflows that align with expert cognition and deliver precise details of reasoning steps. To address these challenges, we propose LogReasoner, a coarse-to-fine reasoning enhancement framework designed to enable LLMs to reason log analysis tasks like experts. LogReasoner consists of two stages: (1) coarse-grained enhancement of expert thinking, where high-level expert thoughts are constructed from collected troubleshooting flowcharts and existing tasks to enable LLMs to formulate structured reasoning workflows and (2) fine-grained enhancement of specific steps, where we first fine-tune the LLM with task-specific stepwise solutions to enhance the LLM for instantiated reasoning, then employ the preference learning to calibrate the LLM's reasoning details from its mistakes, further strengthen the LLM's analytical granularity and correctness. We evaluate LogReasoner on four distinct log analysis tasks using open-source LLMs such as Qwen-2.5 and Llama-3. Experimental results show that LogReasoner significantly outperforms existing LLMs, achieving state-of-the-art performance and demonstrating its effectiveness in enhancing the reasoning capabilities of LLMs for log analysis.

Large language models (LLMs) often exhibit limited performance on domain-specific tasks due to the natural disproportionate representation of specialized information in their training data and the static nature of these datasets. Knowledge scarcity and temporal lag create knowledge gaps for domain applications. While post-training on domain datasets can embed knowledge into models, existing approaches have some limitations. Continual Pre-Training (CPT) treats all tokens in domain documents with equal importance, failing to prioritize critical knowledge points, while supervised fine-tuning (SFT) with question-answer pairs struggles to develop the coherent knowledge structures necessary for complex reasoning tasks. To address these challenges, we propose Reinforcement Learning from Augmented Generation (RLAG). Our approach iteratively cycles between sampling generations and optimizing the model through calculated rewards, effectively embedding critical and contextually coherent domain knowledge. We select generated outputs with the highest log probabilities as the sampling result, then compute three tailored reward metrics to guide the optimization process. To comprehensively evaluate domain expertise, we assess answer accuracy and the rationality of explanations generated for correctly answered questions. Experimental results across medical, legal, astronomy, and current events datasets demonstrate that our proposed method significantly outperforms baseline approaches. Our code and data are open sourced at https://github.com/ChaojunNie/RLAG.

Reinforcement learning (RL) has become a cornerstone in advancing large-scale pre-trained language models (LLMs). Successive generations, including GPT-o series, DeepSeek-R1, Kimi-K1.5, Grok 4, and GLM-4.5, have relied on large-scale RL training to enhance reasoning and coding capabilities. To meet the community's growing RL needs, numerous RL frameworks have been proposed. However, RL training remains computationally expensive, with rollout generation accounting for more than 90% of total runtime. In addition, its efficiency is often constrained by the long-tail distribution of rollout response lengths, where a few lengthy responses stall entire batches, leaving GPUs idle and underutilized. As model and rollout sizes continue to grow, this bottleneck increasingly limits scalability. To address this challenge, we propose Active Partial Rollouts in Reinforcement Learning (APRIL), which mitigates long-tail inefficiency. In the rollout phase, APRIL over-provisions rollout requests, terminates once the target number of responses is reached, and recycles incomplete responses for continuation in future steps. This strategy ensures that no rollouts are discarded while substantially reducing GPU idle time. Experiments show that APRIL improves rollout throughput by 22.5% on average (at most 44%) across commonly used RL algorithms (GRPO, DAPO, GSPO), accelerates convergence, and achieves 2.1% on average(at most 8%) higher final accuracy across tasks. Moreover, APRIL is both framework and hardware agnostic, already integrated into the slime RL framework, and deployable on NVIDIA and AMD GPUs alike. Taken together, this work unifies system-level and algorithmic considerations in proposing APRIL, with the aim of advancing RL training efficiency and inspiring further optimizations in RL systems. Our codebase is available at https://github.com/RLsys-Foundation/APRIL

We introduce Memory-QA, a novel real-world task that involves answering recall questions about visual content from previously stored multimodal memories. This task poses unique challenges, including the creation of task-oriented memories, the effective utilization of temporal and location information within memories, and the ability to draw upon multiple memories to answer a recall question. To address these challenges, we propose a comprehensive pipeline, Pensieve, integrating memory-specific augmentation, time- and location-aware multi-signal retrieval, and multi-memory QA fine-tuning. We created a multimodal benchmark to illustrate various real challenges in this task, and show the superior performance of Pensieve over state-of-the-art solutions (up to 14% on QA accuracy).