We introduce FreshStack, a holistic framework for automatically building information retrieval (IR) evaluation benchmarks by incorporating challenging questions and answers. FreshStack conducts the following steps:(1) automatic corpus collection from code and technical documentation,(2) nugget generation from community-asked questions and answers, and(3) nugget-level support, retrieving documents using a fusion of retrieval techniques and hybrid architectures.We use FreshStack to build five datasets on fast-growing, recent, and niche domains to ensure the tasks are sufficiently challenging. On FreshStack, existing retrieval models, when applied out-of-the-box, significantly underperform oracle approaches on all five domains, denoting plenty of headroom to improve IR quality. In addition, we identify cases where rerankers do not improve first-stage retrieval accuracy (two out of five domains) and oracle context helps an LLM generator generate a high-quality RAG answer.We hope FreshStack will facilitate future work toward constructing realistic, scalable, and uncontaminated IR and RAG evaluation benchmarks.

Although large language models (LLMs) have shown great performance in natural language processing (NLP) in the financial domain, there are no publicly available financially tailored LLMs, instruction tuning datasets, and evaluation benchmarks, which is critical for continually pushing forward the open-source development of financial artificial intelligence (AI). This paper introduces PIXIU, a comprehensive framework including the first financial LLM based on fine-tuning LLaMA with instruction data, the first instruction data with 128K data samples to support the fine-tuning, and an evaluation benchmark with 8 tasks and 15 datasets. We first construct the large-scale multi-task instruction data considering a variety of financial tasks, financial document types, and financial data modalities. We then propose a financial LLM called FinMA by fine-tuning LLaMA with the constructed dataset to be able to follow instructions for various financial tasks. To support the evaluation of financial LLMs, we propose a standardized benchmark that covers a set of critical financial tasks, including six financial NLP tasks and two financial prediction tasks. With this benchmark, we conduct a detailed analysis of FinMA and several existing LLMs, uncovering their strengths and weaknesses in handling critical financial tasks. The model, datasets, benchmark, and experimental results are open-sourced to facilitate future research in financial AI.

Reasoning based on Large Language Models (LLMs) has garnered increasing attention due to outstanding performance of these models in mathematical and complex logical tasks. Beginning with the Chain-of-Thought (CoT) prompting technique, numerous reasoning methods have emerged that decompose problems into smaller, sequential steps (or thoughts). However, existing reasoning models focus on conventional problem-solving and do not necessarily generate creative solutions by ``creative reasoning''. In domains where the solution space is expansive and conventional solutions are suboptimal, such as drug discovery or business strategization, creative reasoning to discover innovative solutions is crucial. To address this gap, first we introduce a computational framework for creative reasoning inspired by established cognitive science principles. With this framework, we propose three core creative reasoning paradigms, namely, \textit{combinational}, \textit{exploratory}, and \textit{transformative} reasoning, where each offers specific directions for systematic exploration of the universe of thoughts to generate creative solutions. Next, to materialize this framework using LLMs, we introduce the \textit{Universe of Thoughts} (or \textit{UoT}, for short), a novel set of methods to implement the aforementioned three creative processes. Finally, we introduce three novel tasks that necessitate creative problem-solving, along with an evaluation benchmark to assess creativity from three orthogonal perspectives: feasibility as constraint, and utility and novelty as metrics. With a comparative analysis against the state-of-the-art (SOTA) reasoning techniques as well as representative commercial models with reasoning capability, we show that UoT demonstrates superior performance in creative reasoning.

This paper comprehensively elaborates on the construction methodology, multi-dimensional evaluation system, and underlying design philosophy of CUFEInse v1.0. Adhering to the principles of "quantitative-oriented, expert-driven, and multi-validation," the benchmark establishes an evaluation framework covering 5 core dimensions, 54 sub-indicators, and 14,430 high-quality questions, encompassing insurance theoretical knowledge, industry understanding, safety and compliance, intelligent agent application, and logical rigor. Based on this benchmark, a comprehensive evaluation was conducted on 11 mainstream large language models. The evaluation results reveal that general-purpose models suffer from common bottlenecks such as weak actuarial capabilities and inadequate compliance adaptation. High-quality domain-specific training demonstrates significant advantages in insurance vertical scenarios but exhibits shortcomings in business adaptation and compliance. The evaluation also accurately identifies the common bottlenecks of current large models in professional scenarios such as insurance actuarial, underwriting and claim settlement reasoning, and compliant marketing copywriting. The establishment of CUFEInse not only fills the gap in professional evaluation benchmarks for the insurance field, providing academia and industry with a professional, systematic, and authoritative evaluation tool, but also its construction concept and methodology offer important references for the evaluation paradigm of large models in vertical fields, serving as an authoritative reference for academic model optimization and industrial model selection. Finally, the paper looks forward to the future iteration direction of the evaluation benchmark and the core development direction of "domain adaptation + reasoning enhancement" for insurance large models.

With the rapid development of large language models (LLMs), various LLM-based works have been widely applied in educational fields. However, most existing LLMs and their benchmarks focus primarily on the knowledge dimension, largely neglecting the evaluation of cultivation capabilities that are essential for real-world educational scenarios. Additionally, current benchmarks are often limited to a single subject or question type, lacking sufficient diversity. This issue is particularly prominent within the Chinese context. To address this gap, we introduce OmniEduBench, a comprehensive Chinese educational benchmark. OmniEduBench consists of 24.602K high-quality question-answer pairs. The data is meticulously divided into two core dimensions: the knowledge dimension and the cultivation dimension, which contain 18.121K and 6.481K entries, respectively. Each dimension is further subdivided into 6 fine-grained categories, covering a total of 61 different subjects (41 in the knowledge and 20 in the cultivation). Furthermore, the dataset features a rich variety of question formats, including 11 common exam question types, providing a solid foundation for comprehensively evaluating LLMs' capabilities in education. Extensive experiments on 11 mainstream open-source and closed-source LLMs reveal a clear performance gap. In the knowledge dimension, only Gemini-2.5 Pro surpassed 60\% accuracy, while in the cultivation dimension, the best-performing model, QWQ, still trailed human intelligence by nearly 30\%. These results highlight the substantial room for improvement and underscore the challenges of applying LLMs in education.

Recent techniques such as retrieval-augmented generation or chain-of-thought reasoning have led to longer contexts and increased inference costs. Context compression techniques can reduce these costs, but the most effective approaches require fine-tuning the target model or even modifying its architecture. This can degrade its general abilities when not used for this specific purpose. Here we explore an alternative approach: an encoder that compresses the context into continuous representations which replace token embeddings in decoder LLMs. First, we perform a systematic study of training strategies and architecture choices for the encoder. Our findings led to the design of an Adaptable text Representations Compressor, named ARC-Encoder, which outputs $x$-times fewer continuous representations (typically $x\!\in\!\{4,8\}$) than text tokens. We evaluate ARC-Encoder across a variety of LLM usage scenarios, ranging from in-context learning to context window extension, on both instruct and base decoders. Results show that ARC-Encoder achieves state-of-the-art performance on several benchmarks while improving computational efficiency at inference. Finally, we demonstrate that our models can be adapted to multiple decoders simultaneously, allowing a single encoder to generalize across different decoder LLMs. This makes ARC-Encoder a flexible and efficient solution for portable encoders that work seamlessly with multiple LLMs. We release a training code at https://github.com/kyutai-labs/ARC-Encoder , fine-tuning dataset and pretrained models are available at https://huggingface.co/collections/kyutai/arc-encoders-68ee18787301407d60a57047 .

Recent advances in multi-modal large language models (MLLMs) have enabled increasingly sophisticated autonomous visualization agents capable of translating user intentions into data visualizations. However, measuring progress and comparing different agents remains challenging, particularly in scientific visualization (SciVis), due to the absence of comprehensive, large-scale benchmarks for evaluating real-world capabilities. This position paper examines the various types of evaluation required for SciVis agents, outlines the associated challenges, provides a simple proof-of-concept evaluation example, and discusses how evaluation benchmarks can facilitate agent self-improvement. We advocate for a broader collaboration to develop a SciVis agentic evaluation benchmark that would not only assess existing capabilities but also drive innovation and stimulate future development in the field.

In recent years, with the rapid development of the depth and breadth of large language models' capabilities, various corresponding evaluation benchmarks have been emerging in increasing numbers. As a quantitative assessment tool for model performance, benchmarks are not only a core means to measure model capabilities but also a key element in guiding the direction of model development and promoting technological innovation. We systematically review the current status and development of large language model benchmarks for the first time, categorizing 283 representative benchmarks into three categories: general capabilities, domain-specific, and target-specific. General capability benchmarks cover aspects such as core linguistics, knowledge, and reasoning; domain-specific benchmarks focus on fields like natural sciences, humanities and social sciences, and engineering technology; target-specific benchmarks pay attention to risks, reliability, agents, etc. We point out that current benchmarks have problems such as inflated scores caused by data contamination, unfair evaluation due to cultural and linguistic biases, and lack of evaluation on process credibility and dynamic environments, and provide a referable design paradigm for future benchmark innovation.

Instruction Tuning has emerged as a pivotal post-training paradigm that enables pre-trained language models to better follow user instructions. Despite its significance, little attention has been given to optimizing the loss function used. A fundamental, yet often overlooked, question is whether the conventional auto-regressive objective - where loss is computed only on response tokens, excluding prompt tokens - is truly optimal for instruction tuning. In this work, we systematically investigate the impact of differentially weighting prompt and response tokens in instruction tuning loss, and propose Weighted Instruction Tuning (WIT) as a better alternative to conventional instruction tuning. Through extensive experiments on five language models of different families and scale, three finetuning datasets of different sizes, and five diverse evaluation benchmarks, we show that the standard instruction tuning loss often yields suboptimal performance and limited robustness to input prompt variations. We find that a low-to-moderate weight for prompt tokens coupled with a moderate-to-high weight for response tokens yields the best-performing models across settings and also serve as better starting points for the subsequent preference alignment training. These findings highlight the need to reconsider instruction tuning loss and offer actionable insights for developing more robust and generalizable models. Our code is open-sourced at https://github.com/kowndinya-renduchintala/WIT.