Unsupervised skill discovery carries the promise that an intelligent agent can learn reusable skills through autonomous, reward-free environment interaction.

NeurIPS Rebuttal for "Retrieval-Augmented Generation for Knowledge-Intensive NLP T asks" We thank reviewers for their thoughtful, detailed reviews. "information retrieval strategy to improve the the generation Pre-trained seq2seq models have only become available in the last year (T5, BART) or two (GPT2). We study two RAG models. RAG-Sequence's formulation is similar to REALM, but RAG-Token is novel and Further, we explore novel decoding strategies for these models. "contribution [...] is not very specific, since R1 suggested that "A figure or example about P AG-Sequence Model and P AG-Token Model is needed", and R3 mentions "description of the model is quite concise (due to space restrictions)".

Our understanding of learning input-output relationships with neural nets has improved rapidly in recent years, but little is known about the convergence of the underlying representations, even in the simple case of linear autoencoders (LAEs).

Retrieval-augmented generation (RAG) has emerged as a promising solution to mitigate the limitations of large language models (LLMs), such as hallucinations and outdated information. However, it remains unclear how LLMs handle knowledge conflicts arising from different augmented retrieved passages, especially when these passages originate from the same source and have equal trustworthiness. In this work, we conduct a comprehensive evaluation of LLM-generated answers to questions that have varying answers based on contradictory passages from Wikipedia, a dataset widely regarded as a high-quality pre-training resource for most LLMs. Specifically, we introduce WikiContradict, a benchmark consisting of 253 high-quality, human-annotated instances designed to assess the performance of LLMs in providing a complete perspective on conflicts from the retrieved documents, rather than choosing one answer over another, when augmented with retrieved passages containing real-world knowledge conflicts. We benchmark a diverse range of both closed and open-source LLMs under different QA scenarios, including RAG with a single passage, and RAG with 2 contradictory passages. Through rigorous human evaluations on a subset of WikiContradict instances involving 5 LLMs and over 3,500 judgements, we shed light on the behaviour and limitations of these models. For instance, when provided with two passages containing contradictory facts, all models struggle to generate answers that accurately reflect the conflicting nature of the context, especially for implicit conflicts requiring reasoning. Since human evaluation is costly, wealso introduce an automated model that estimates LLM performance using a strong open-source language model, achieving an F-score of 0.8. Using this automated metric, we evaluate more than 1,500 answers from seven LLMs across all WikiContradict instances.

Benefiting from the inductive biases learned from large-scale datasets, open-vocabulary semantic segmentation (OVSS) leverages the power of vision-language models, such as CLIP, to achieve remarkable progress without requiring task-specific training. However, due to CLIP's pre-training nature on image-text pairs, it tends to focus on global semantic alignment, resulting in suboptimal performance when associating fine-grained visual regions with text. This leads to noisy and inconsistent predictions, particularly in local areas. We attribute this to a dispersed bias stemming from its contrastive training paradigm, which is difficult to alleviate using CLIP features alone. To address this, we propose a structure-aware feature rectification approach that incorporates instance-specific priors derived directly from the image. Specifically, we construct a region adjacency graph (RAG) based on low-level features (e.g., colour and texture) to capture local structural relationships and use it to refine CLIP features by enhancing local discrimination. Extensive experiments show that our method effectively suppresses segmentation noise, improves region-level consistency, and achieves strong performance on multiple open-vocabulary segmentation benchmarks.

Vision-language models (VLMs) have achieved strong performance in visual question answering (VQA), yet they remain constrained by static training data. Retrieval-Augmented Generation (RAG) mitigates this limitation by enabling access to up-to-date, culturally grounded, and multilingual information; however, multilingual multimodal RAG remains largely underexplored. We introduce M4-RAG, a massive-scale benchmark covering 42 languages and 56 regional dialects and registers, comprising over 80,000 culturally diverse image-question pairs for evaluating retrieval-augmented VQA across languages and modalities. To balance realism with reproducibility, we build a controlled retrieval environment containing millions of carefully curated multilingual documents relevant to the query domains, approximating real-world retrieval conditions while ensuring consistent experimentation. Our systematic evaluation reveals that although RAG consistently benefits smaller VLMs, it fails to scale to larger models and often even degrades their performance, exposing a critical mismatch between model size and current retrieval effectiveness. M4-RAG provides a foundation for advancing next-generation RAG systems capable of reasoning seamlessly across languages, modalities, and cultural contexts.

Abstract--Large Language Models (LLMs) hold significant promise for mathematics education, yet they often struggle with complex mathematical reasoning. While Retrieval-Augmented Generation (RAG) mitigates these issues by grounding LLMs in external knowledge, its effectiveness remains unstable, heavily dependent on the choice of a single embedding model. Moving beyond static RAG workflows, we draw on agentic workflow patterns, a paradigm that introduces structured task decomposition and collaboration to enhance system performance. We propose and examine two novel approaches that combine the benefits of multiple embedding models. While our Mixture-Embedding RAG approach (fusing retrieved documents) shows limited gains, our Confident RAG method (generating multiple answers and selecting the one with the highest confidence score) demonstrates significant improvement. Experimental results show that Confident RAG achieved average accuracy improvements of approximately 10% over vanilla LLMs and 5% over vanilla RAG. The consistent results across different LLMs and embedding models indicate that Confident RAG is an efficient plug-and-play solution for trustworthy mathematical AI assistants. Finally, we discuss how this work lays the groundwork for deploying Agentic RAG systems in educational settings, where autonomous planning and iterative refinement can be built upon our robust retrieval foundation. ARGE language models (LLMs) have demonstrated remarkable capabilities across various domains [1]-[3], showing particular promise for educational applications. However, their tendency to hallucinate [4] remains a significant barrier to reliable use in learning environments, especially in mathematics education where accuracy is crucial [5].

Most evaluations of large language models focus on standard tasks such as factual question answering or short summarization. This research expands that scope in two directions: first, by comparing two retrieval strategies, Graph RAG, structured knowledge-graph based, and Advanced RAG, hybrid keyword-semantic search, for QA; and second, by evaluating whether LLMs can generate high quality non-traditional academic outputs, specifically slide decks and podcast scripts. We implemented a prototype combining Meta's LLaMA 3 70B open weight and OpenAI's GPT 4o mini API based. QA performance was evaluated using both human ratings across eleven quality dimensions and large language model judges for scalable cross validation. GPT 4o mini with Advanced RAG produced the most accurate responses. Graph RAG offered limited improvements and led to more hallucinations, partly due to its structural complexity and manual setup. Slide and podcast generation was tested with document grounded retrieval. GPT 4o mini again performed best, though LLaMA 3 showed promise in narrative coherence. Human reviewers were crucial for detecting layout and stylistic flaws, highlighting the need for combined human LLM evaluation in assessing emerging academic outputs.

Bangladesh's low-income population faces major barriers to affordable legal advice due to complex legal language, procedural opacity, and high costs. Existing AI legal assistants lack Bengali-language support and jurisdiction-specific adaptation, limiting their effectiveness. To address this, we developed Mina, a multilingual LLM-based legal assistant tailored for the Bangladeshi context. It employs multilingual embeddings and a RAG-based chain-of-tools framework for retrieval, reasoning, translation, and document generation, delivering context-aware legal drafts, citations, and plain-language explanations via an interactive chat interface. Evaluated by law faculty from leading Bangladeshi universities across all stages of the 2022 and 2023 Bangladesh Bar Council Exams, Mina scored 75-80% in Preliminary MCQs, Written, and simulated Viva Voce exams, matching or surpassing average human performance and demonstrating clarity, contextual understanding, and sound legal reasoning. Even under a conservative upper bound, Mina operates at just 0.12-0.61% of typical legal consultation costs in Bangladesh, yielding a 99.4-99.9\% cost reduction relative to human-provided services. These results confirm its potential as a low-cost, multilingual AI assistant that automates key legal tasks and scales access to justice, offering a real-world case study on building domain-specific, low-resource systems and addressing challenges of multilingual adaptation, efficiency, and sustainable public-service AI deployment.

Large language model agents are increasingly used to automate web tasks such as product search, offer comparison, and checkout. Current research explores different interfaces through which these agents interact with websites, including traditional HTML browsing, retrieval-augmented generation (RAG) over pre-crawled content, communication via Web APIs using the Model Context Protocol (MCP), and natural-language querying through the NLWeb interface. However, no prior work has compared these four architectures within a single controlled environment using identical tasks. To address this gap, we introduce a testbed consisting of four simulated e-shops, each offering its products via HTML, MCP, and NLWeb interfaces. For each interface (HTML, RAG, MCP, and NLWeb) we develop specialized agents that perform the same sets of tasks, ranging from simple product searches and price comparisons to complex queries for complementary or substitute products and checkout processes. We evaluate the agents using GPT 4.1, GPT 5, GPT 5 mini, and Claude Sonnet 4 as underlying LLM. Our evaluation shows that the RAG, MCP and NLWeb agents outperform HTML on both effectiveness and efficiency. Averaged over all tasks, F1 rises from 0.67 for HTML to between 0.75 and 0.77 for the other agents. Token usage falls from about 241k for HTML to between 47k and 140k per task. The runtime per task drops from 291 seconds to between 50 and 62 seconds. The best overall configuration is RAG with GPT 5 achieving an F1 score of 0.87 and a completion rate of 0.79. Also taking cost into consideration, RAG with GPT 5 mini offers a good compromise between API usage fees and performance. Our experiments show the choice of the interaction interface has a substantial impact on both the effectiveness and efficiency of LLM-based web agents.