Most video reasoning models only generate textual reasoning traces without indicating when and where key evidence appears. Recent models such as OpenAI-o3 have sparked wide interest in evidence-centered reasoning for images, yet extending this ability to videos is more challenging, as it requires joint temporal tracking and spatial localization across dynamic scenes. We introduce Open-o3 Video, a non-agent framework that integrates explicit spatio-temporal evidence into video reasoning, and carefully collect training data and design training strategies to address the aforementioned challenges. The model highlights key timestamps, objects, and bounding boxes alongside its answers, allowing reasoning to be grounded in concrete visual observations. To enable this functionality, we first curate and build two high-quality datasets, STGR-CoT-30k for SFT and STGR-RL-36k for RL, with carefully constructed temporal and spatial annotations, since most existing datasets offer either temporal spans for videos or spatial boxes on images, lacking unified spatio-temporal supervision and reasoning traces. Then, we adopt a cold-start reinforcement learning strategy with multiple specially designed rewards that jointly encourage answer accuracy, temporal alignment, and spatial precision. On V-STAR benchmark, Open-o3 Video achieves state-of-the-art performance, raising mAM by 14.4% and mLGM by 24.2% on the Qwen2.5-VL baseline. Consistent improvements are also observed on a broad range of video understanding benchmarks, including VideoMME, WorldSense, VideoMMMU, and TVGBench. Beyond accuracy, the reasoning traces produced by Open-o3 Video also provide valuable signals for test-time scaling, enabling confidence-aware verification and improving answer reliability.

The realization of Artificial General Intelligence (AGI) necessitates Embodied AI agents capable of robust spatial perception, effective task planning, and adaptive execution in physical environments. However, current large language models (LLMs) and multimodal LLMs (MLLMs) for embodied tasks suffer from key limitations, including a significant gap between model design and agent requirements, an unavoidable trade-off between real-time latency and performance, and the use of unauthentic, offline evaluation metrics. To address these challenges, we propose EmbodiedBrain, a novel vision-language foundation model available in both 7B and 32B parameter sizes. Our framework features an agent-aligned data structure and employs a powerful training methodology that integrates large-scale Supervised Fine-Tuning (SFT) with Step-Augumented Group Relative Policy Optimization (Step-GRPO), which boosts long-horizon task success by integrating preceding steps as Guided Precursors. Furthermore, we incorporate a comprehensive reward system, including a Generative Reward Model (GRM) accelerated at the infrastructure level, to improve training efficiency. For enable thorough validation, we establish a three-part evaluation system encompassing General, Planning, and End-to-End Simulation Benchmarks, highlighted by the proposal and open-sourcing of a novel, challenging simulation environment. Experimental results demonstrate that EmbodiedBrain achieves superior performance across all metrics, establishing a new state-of-the-art for embodied foundation models. Towards paving the way for the next generation of generalist embodied agents, we open-source all of our data, model weight, and evaluating methods, which are available at https://zterobot.github.io/EmbodiedBrain.github.io.

Zero-shot reinforcement learning (RL) has emerged as a setting for developing general agents in an unsupervised manner, capable of solving downstream tasks without additional training or planning at test-time. Unlike conventional RL, which optimizes policies for a fixed reward, zero-shot RL requires agents to encode representations rich enough to support immediate adaptation to any objective, drawing parallels to vision and language foundation models. Despite growing interest, the field lacks a common analytical lens. We present the first unified framework for zero-shot RL. Our formulation introduces a consistent notation and taxonomy that organizes existing approaches and allows direct comparison between them. Central to our framework is the classification of algorithms into two families: direct representations, which learn end-to-end mappings from rewards to policies, and compositional representations, which decompose the representation leveraging the substructure of the value function. Within this framework, we highlight shared principles and key differences across methods, and we derive an extended bound for successor-feature methods, offering a new perspective on their performance in the zero-shot regime. By consolidating existing work under a common lens, our framework provides a principled foundation for future research in zero-shot RL and outlines a clear path toward developing more general agents.

The political biases of Large Language Models (LLMs) are usually assessed by simulating their answers to English surveys. In this work, we propose an alternative framing of political biases, relying on principles of fairness in multilingual translation. We systematically compare the translation quality of speeches in the European Parliament (EP), observing systematic differences with majority parties from left, center, and right being better translated than outsider parties. This study is made possible by a new, 21-way multiparallel version of EuroParl, the parliamentary proceedings of the EP, which includes the political affiliations of each speaker. The dataset consists of 1.5M sentences for a total of 40M words and 249M characters. It covers three years, 1000+ speakers, 7 countries, 12 EU parties, 25 EU committees, and hundreds of national parties.

Retrieval-augmented generation (RAG) remains brittle on multi-step questions and heterogeneous evidence sources, trading accuracy against latency and token/tool budgets. This paper introduces Hierarchical Sequence (HSEQ) Iteration for Heterogeneous Question Answering, a unified framework that (i) linearize documents, tables, and knowledge graphs into a reversible hierarchical sequence with lightweight structural tags, and (ii) perform structure-aware iteration to collect just-enough evidence before answer synthesis. A Head Agent provides guidance that leads retrieval, while an Iteration Agent selects and expands HSeq via structure-respecting actions (e.g., parent/child hops, table row/column neighbors, KG relations); Finally the head agent composes canonicalized evidence to genearte the final answer, with an optional refinement loop to resolve detected contradictions. Besides, HSEQ exhibits three key advantages: (1) a format-agnostic unification that enables a single policy to operate across text, tables, and KGs without per-dataset specialization; (2) guided, budget-aware iteration that reduces unnecessary hops, tool calls, and tokens while preserving accuracy; and (3) evidence canonicalization for reliable QA, improving answers consistency and auditability. Large language models (LLMs), such as ChatGPT (Achiam et al., 2023), LLaMA (Dubey et al., 2024), Falcon (Zuo et al., 2025), have been increasingly relying on retrieval-augmented generation (RAG) to ground answers in external evidence. With reliable supplementary knowledge offered factual errors are reduced, especially in domain-specific questions, leading to higher accuracy and fewer hallucinations (Zhu et al., 2021b; Gao et al., 2023; Zhao et al., 2024). However they may fall with branchy plans, repeated web/file calls, and verbose chain-of-thought prompts, yielding unpredictable token/tool costs and latency; termination is often heuristic, leading to premature answers or extra wasted loops with budgets decoupled from the evidence actually inspected (Singh et al., 2025). Although existing heterogeneous RAG systems (Y u, 2022; Christmann & Weikum, 2024) are available to deal with multiple formats of data, they may still face issues in either weak alignment across representations or lossy and non-reversible serialization that obscures provenance and blocks faithful reconstruction. Hierarchical Sequence Iteration (HSEQ) for Heterogeneous Question Answering introduces a reversible hierarchical sequence interface that linearizes documents, tables, and KGs into a sequence of typed segments with lightweight structure (e.g., parent/child locality, offsets or coordinates, minimal schema/time tags).

We unveil that internal representations in large language models (LLMs) serve as reliable proxies of learned knowledge, and propose RECALL, a novel representation-aware model merging framework for continual learning without access to historical data. RECALL computes inter-model similarity from layer-wise hidden representations over clustered typical samples, and performs adaptive, hierarchical parameter fusion to align knowledge across models. This design enables the preservation of domain-general features in shallow layers while allowing task-specific adaptation in deeper layers. Unlike prior methods that require task labels or incur performance trade-offs, RECALL achieves seamless multi-domain integration and strong resistance to catastrophic forgetting. Extensive experiments across five NLP tasks and multiple continual learning scenarios show that RECALL outperforms baselines in both knowledge retention and generalization, providing a scalable and data-free solution for evolving LLMs.

Large language models (LLMs) produce outputs with varying levels of uncertainty, and, just as often, varying levels of correctness; making their practical reliability far from guaranteed. To quantify this uncertainty, we systematically evaluate four approaches for confidence estimation in LLM outputs: VCE, MSP, Sample Consistency, and CoCoA (Vashurin et al., 2025). For the evaluation of the approaches, we conduct experiments on four question-answering tasks using a state-of-the-art open-source LLM. Our results show that each uncertainty metric captures a different facet of model confidence and that the hybrid CoCoA approach yields the best reliability overall, improving both calibration and discrimination of correct answers. We discuss the trade-offs of each method and provide recommendations for selecting uncertainty measures in LLM applications.

Instruction tuning is crucial for aligning Large Language Models (LLMs), yet the quality of instruction-following data varies significantly. While high-quality data is paramount, it is often scarce; conversely, abundant low-quality data is frequently discarded, leading to substantial information loss. Existing data augmentation methods struggle to augment this low-quality data effectively, and the evaluation of such techniques remains poorly defined. To address this, we formally define the task of Instruction Distillation: distilling multiple low-quality and redundant inputs into high-quality and coherent instruction-output pairs. This process uses three complementary reward signals: quality, semantic alignment, and format compliance, via Group Relative Policy Optimization (GRPO). We demonstrate that LM-Mixup effectively augments imperfect datasets: fine-tuning LLMs on its distilled data, which accounts for only about 3% of the entire dataset, not only surpasses full-dataset training but also competes with state-of-the-art high-quality data selection methods across multiple benchmarks. Our work establishes that low-quality data is a valuable resource when properly distilled and augmented with LM-Mixup, significantly enhancing the efficiency and performance of instruction-tuned LLMs. The code and the dataset are available at: https://github.com/yuu250/LM-mixup. In recent years, large language models (LLMs) have achieved notable progress in natural language processing and multimodal understanding (Team et al., 2023; Singhal et al., 2023; Deng et al., 2025; Li et al., 2024b; 2025a; Pang et al., 2025b). This progress stems not only from improved architectures and larger scales but also from more efficient ways for models to learn and apply knowledge (Patil & Jadon, 2025; Dredze, 2025). While the conventional view holds that high-quality human alignment requires massive annotated data (Kim et al., 2024; K opf et al., 2023), recent studies show that LLMs acquire most knowledge during pre-training (Brown et al., 2020; Roberts et al., 2020). This shifts the research focus from "more data" to "better data", emphasizing efficient high-quality data selection for model improvement. However, high-quality samples are scarce and costly, while real-world datasets contain abundant redundant or low-quality data, leading to significant information waste.

Large language models (LLMs) are increasingly deployed in real-world systems, yet they can produce toxic or biased outputs that undermine safety and trust. Post-hoc model repair provides a practical remedy, but the high cost of parameter updates motivates selective use of repair data. Despite extensive prior work on data selection for model training, it remains unclear which sampling criteria are most effective and efficient when applied specifically to behavioral repair of large generative models. Our study presents a systematic analysis of sample prioritization strategies for LLM repair. We evaluate five representative selection methods, including random sampling, K-Center, gradient-norm-based selection(GraNd), stratified coverage (CCS), and a Semantic-Aware Prioritized Sampling (SAPS) approach we proposed. Repair effectiveness and trade-offs are assessed through toxicity reduction, perplexity on WikiText-2 and LAMBADA, and three composite metrics: the Repair Proximity Score (RPS), the Overall Performance Score (OPS), and the Repair Efficiency Score (RES). Experimental results show that SAPS achieves the best balance between detoxification, utility preservation, and efficiency, delivering comparable or superior repair outcomes with substantially less data. Random sampling remains effective for large or robust models, while high-overhead methods such as CCS and GraNd provide limited benefit. The optimal data proportion depends on model scale and repair method, indicating that sample selection should be regarded as a tunable component of repair pipelines. Overall, these findings establish selection-based repair as an efficient and scalable paradigm for maintaining LLM reliability.

This paper presents a new computanullonal model of creanullve outcomes, informed by creanullvity theories and techniques, which was implemented tool to generate more novel opportuninulles for innovanullon projects. The model implemented five funcnullons that were developed to contribute to the generanullon of innovanullon opportuninulles with higher novelty without loss of usefulness. The model was evaluated using opportuninulles generated for an innovanullon project in the hospitality sector . The evaluanullon revealed that the co mputanullonal model generated outcomes that were more novel and/or useful than outcomes from Notebook LM and ChatGPT4o. However, not all of the model's funcnullons contributed to the generanullon of more novel opportuninulles, leading to new direcnullons for further model development .