Large language models are finetuned to refuse questions about hazardous knowledge, but these protections can often be bypassed. Unlearning methods aim at completely removing hazardous capabilities from models and make them inaccessible to adversaries. This work challenges the fundamental differences between unlearning and traditional safety post-training from an adversarial perspective. We demonstrate that existing jailbreak methods, previously reported as ineffective against unlearning, can be successful when applied carefully. Furthermore, we develop a variety of adaptive methods that recover most supposedly unlearned capabilities. For instance, we show that finetuning on 10 unrelated examples or removing specific directions in the activation space can recover most hazardous capabilities for models edited with RMU, a state-of-the-art unlearning method. Our findings challenge the robustness of current unlearning approaches and question their advantages over safety training.

Generating accurate and consistent visual aids is a critical challenge in mathematics education, where visual representations like geometric shapes and functions play a pivotal role in enhancing student comprehension. This paper introduces a novel multi-agent framework that leverages Large Language Models (LLMs) to automate the creation of complex mathematical visualizations alongside coherent problem text. Our approach not only simplifies the generation of precise visual aids but also aligns these aids with the problem's core mathematical concepts, improving both problem creation and assessment. By integrating multiple agents, each responsible for distinct tasks such as numeric calculation, geometry validation, and visualization, our system delivers mathematically accurate and contextually relevant problems with visual aids. Evaluation across Geometry and Function problem types shows that our method significantly outperforms basic LLMs in terms of text coherence, consistency, relevance and similarity, while maintaining the essential geometrical and functional integrity of the original problems. Although some challenges remain in ensuring consistent visual outputs, our framework demonstrates the immense potential of LLMs in transforming the way educators generate and utilize visual aids in math education.

Large Language Models (LLMs) have demonstrated impressive capabilities in reasoning using Chain-of-Thought (CoT) prompting. However, CoT can be biased by users' instruction. In this work, we study the reasoning robustness of LLMs to typographical errors, which can naturally occur in users' queries. We design an Adversarial Typo Attack ($\texttt{ATA}$) algorithm that iteratively samples typos for words that are important to the query and selects the edit that is most likely to succeed in attacking. It shows that LLMs are sensitive to minimal adversarial typographical changes. Notably, with 1 character edit, Mistral-7B-Instruct's accuracy drops from 43.7% to 38.6% on GSM8K, while with 8 character edits the performance further drops to 19.2%. To extend our evaluation to larger and closed-source LLMs, we develop the $\texttt{R$^2$ATA}$ benchmark, which assesses models' $\underline{R}$easoning $\underline{R}$obustness to $\underline{\texttt{ATA}}$. It includes adversarial typographical questions derived from three widely used reasoning datasets-GSM8K, BBH, and MMLU-by applying $\texttt{ATA}$ to open-source LLMs. $\texttt{R$^2$ATA}$ demonstrates remarkable transferability and causes notable performance drops across multiple super large and closed-source LLMs.

We explore the capability of four open-sourcelarge language models (LLMs) in argumentation mining (AM). We conduct experiments on three different corpora; persuasive essays(PE), argumentative microtexts (AMT) Part 1 and Part 2, based on two argumentation mining sub-tasks: (i) argumentative discourse units classifications (ADUC), and (ii) argumentative relation classification (ARC). This work aims to assess the argumentation capability of open-source LLMs, including Mistral 7B, Mixtral8x7B, LlamA2 7B and LlamA3 8B in both, zero-shot and few-shot scenarios. Our analysis contributes to further assessing computational argumentation with open-source LLMs in future research efforts.

We present an innovative approach for solving mathematical problems in Bengali, developed for the DL Sprint 3.0 BUET CSE Fest 2024 Competition. Our method uses advanced deep learning models, notably the Qwen 2.5 series, with improvements made through prompt engineering, model quantization, and Tool Integrated Reasoning (TIR) to handle complex calculations. Initially, we explored various model architectures, including fine-tuned Mistral and quantized Qwen models, refining them with translation techniques, Retrieval-Augmented Generation (RAG), and custom dataset curation. Manual hyperparameter tuning optimized parameters like temperature and top-p to enhance model adaptability and accuracy. Removal of RAG and parameter adjustments further improved robustness. Our approach highlights the potential of advanced NLP techniques in solving Bengali mathematical problems.

Recent advances in large language models (LLMs) have led to significant progress in robotics, enabling embodied agents to better understand and execute open-ended tasks. However, existing approaches using LLMs face limitations in grounding their outputs within the physical environment and aligning with the capabilities of the robot. This challenge becomes even more pronounced with smaller language models, which are more computationally efficient but less robust in task planning and execution. In this paper, we present a novel modular architecture designed to enhance the robustness of LLM-driven robotics by addressing these grounding and alignment issues. We formalize the task planning problem within a goal-conditioned POMDP framework, identify key failure modes in LLM-driven planning, and propose targeted design principles to mitigate these issues. Our architecture introduces an ``expected outcomes'' module to prevent mischaracterization of subgoals and a feedback mechanism to enable real-time error recovery. Experimental results, both in simulation and on physical robots, demonstrate that our approach significantly improves task success rates for pick-and-place and manipulation tasks compared to both larger LLMs and standard baselines. Through hardware experiments, we also demonstrate how our architecture can be run efficiently and locally. This work highlights the potential of smaller, locally-executable LLMs in robotics and provides a scalable, efficient solution for robust task execution.

This paper introduces a new environment LLM-PySC2 (the Large Language Model StarCraft II Learning Environment), a platform derived from DeepMind's StarCraft II Learning Environment that serves to develop Large Language Models (LLMs) based decision-making methodologies. This environment is the first to offer the complete StarCraft II action space, multi-modal observation interfaces, and a structured game knowledge database, which are seamlessly connected with various LLMs to facilitate the research of LLMs-based decision-making. To further support multi-agent research, we developed an LLM collaborative framework that supports multi-agent concurrent queries and multi-agent communication. In our experiments, the LLM-PySC2 environment is adapted to be compatible with the StarCraft Multi-Agent Challenge (SMAC) task group and provided eight new scenarios focused on macro-decision abilities. We evaluated nine mainstream LLMs in the experiments, and results show that sufficient parameters are necessary for LLMs to make decisions, but improving reasoning ability does not directly lead to better decision-making outcomes. Our findings further indicate the importance of enabling large models to learn autonomously in the deployment environment through parameter training or train-free learning techniques. Ultimately, we expect that the LLM-PySC2 environment can promote research on learning methods for LLMs, helping LLM-based methods better adapt to task scenarios.

The growth of digital documents presents significant challenges in efficient management and knowledge extraction. Traditional methods often struggle with complex documents, leading to issues such as hallucinations and high latency in responses from Large Language Models (LLMs). ZeroG, an innovative approach, significantly mitigates these challenges by leveraging knowledge distillation and prompt tuning to enhance model performance. ZeroG utilizes a smaller model that replicates the behavior of a larger teacher model, ensuring contextually relevant and grounded responses, by employing a black-box distillation approach, it creates a distilled dataset without relying on intermediate features, optimizing computational efficiency. This method significantly enhances accuracy and reduces response times, providing a balanced solution for modern document management. Incorporating advanced techniques for document ingestion and metadata utilization, ZeroG improves the accuracy of question-and-answer systems. The integration of graph databases and robust metadata management further streamlines information retrieval, allowing for precise and context-aware responses. By transforming how organizations interact with complex data, ZeroG enhances productivity and user experience, offering a scalable solution for the growing demands of digital document management.

Upon release, ChatGPT3.5 shocked the software engineering community by its ability to generate answers to specialized questions about coding. Immediately, many educators wondered if it was possible to use the chatbot as a support tool that helps students answer their programming questions. This article evaluates this possibility at three levels: fundamental Computer Science knowledge (basic algorithms and data structures), core competency (design patterns), and advanced knowledge (quantum computing). In each case, we ask normalized questions several times to ChatGPT3.5, then look at the correctness of answers, and finally check if this creates issues. The main result is that the performances of ChatGPT3.5 degrades drastically as the specialization of the domain increases: for basic algorithms it returns answers that are almost always correct, for design patterns the generated code contains many code smells and is generally of low quality, but it is still sometimes able to fix it (if asked), and for quantum computing it is often blatantly wrong.

Vision Language Models (VLMs) are central to Visual Question Answering (VQA) systems and are typically deployed in the cloud due to their high computational demands. However, this cloud-only approach underutilizes edge computational resources and requires significant bandwidth for transmitting raw images. In this paper, we introduce an edge-cloud collaborative VQA system, called LLaVA-AlignedVQ, which features a novel Aligned Vector Quantization algorithm (AlignedVQ) that efficiently compress intermediate features without compromising accuracy to support partitioned execution. Our experiments demonstrate that LLaVA-AlignedVQ achieves approximately 1365x compression rate of intermediate features, reducing data transmission overhead by 96.8% compared to transmitting JPEG90-compressed images to the cloud. LLaVA-AlignedVQ achieves an inference speedup of 2-15x while maintaining high accuracy, remaining within -2.23% to +1.6% of the original model's accuracy performance across eight VQA datasets, compared to the cloud-only solution.