Security Operations Center (SoC) analysts gather threat reports from openly accessible global threat databases and customize them manually to suit a particular organization's needs. These analysts also depend on internal repositories, which act as private local knowledge database for an organization. Credible cyber intelligence, critical operational details, and relevant organizational information are all stored in these local knowledge databases. Analysts undertake a labor intensive task utilizing these global and local knowledge databases to manually create organization's unique threat response and mitigation strategies. Recently, Large Language Models (LLMs) have shown the capability to efficiently process large diverse knowledge sources. We leverage this ability to process global and local knowledge databases to automate the generation of organization-specific threat intelligence. In this work, we present LOCALINTEL, a novel automated knowledge contextualization system that, upon prompting, retrieves threat reports from the global threat repositories and uses its local knowledge database to contextualize them for a specific organization. LOCALINTEL comprises of three key phases: global threat intelligence retrieval, local knowledge retrieval, and contextualized completion generation. The former retrieves intelligence from global threat repositories, while the second retrieves pertinent knowledge from the local knowledge database. Finally, the fusion of these knowledge sources is orchestrated through a generator to produce a contextualized completion.

Large language models (LLMs) have exhibited great potential in autonomously completing tasks across real-world applications. Despite this, these LLM agents introduce unexpected safety risks when operating in interactive environments. Instead of centering on LLM-generated content safety in most prior studies, this work addresses the imperative need for benchmarking the behavioral safety of LLM agents within diverse environments. We introduce R-Judge, a benchmark crafted to evaluate the proficiency of LLMs in judging safety risks given agent interaction records. R-Judge comprises 162 agent interaction records, encompassing 27 key risk scenarios among 7 application categories and 10 risk types. It incorporates human consensus on safety with annotated safety risk labels and high-quality risk descriptions. Utilizing R-Judge, we conduct a comprehensive evaluation of 8 prominent LLMs commonly employed as the backbone for agents. The best-performing model, GPT-4, achieves 72.29% in contrast to the human score of 89.38%, showing considerable room for enhancing the risk awareness of LLMs. Notably, leveraging risk descriptions as environment feedback significantly improves model performance, revealing the importance of salient safety risk feedback. Furthermore, we design an effective chain of safety analysis technique to help the judgment of safety risks and conduct an in-depth case study to facilitate future research. R-Judge is publicly available at https://github.com/Lordog/R-Judge.

The increasing demand for intelligent systems capable of interpreting and reasoning about visual content requires the development of Large Multi-Modal Models (LMMs) that are not only accurate but also have explicit reasoning capabilities. This paper presents a novel approach to imbue an LMM with the ability to conduct explicit reasoning based on visual content and textual instructions. We introduce a system that can ask a question to acquire necessary knowledge, thereby enhancing the robustness and explicability of the reasoning process. Our method comprises the development of a novel dataset generated by a Large Language Model (LLM), designed to promote chain-of-thought reasoning combined with a question-asking mechanism. We designed an LMM, which has high capabilities on region awareness to address the intricate requirements of image-text alignment. The model undergoes a three-stage training phase, starting with large-scale image-text alignment using a large-scale datasets, followed by instruction tuning, and fine-tuning with a focus on chain-of-thought reasoning. The results demonstrate a stride toward a more robust, accurate, and interpretable LMM, capable of reasoning explicitly and seeking information proactively when confronted with ambiguous visual input.

ChatGPT, as a language model based on large-scale pre-training, has exerted a profound influence on the domain of machine translation. In ChatGPT, a "Prompt" refers to a segment of text or instruction employed to steer the model towards generating a specific category of response. The design of the translation prompt emerges as a key aspect that can wield influence over factors such as the style, precision and accuracy of the translation to a certain extent. However, there is a lack of a common standard and methodology on how to design and select a translation prompt. Accordingly, this paper proposes a generic taxonomy, which defines gradable translation prompts in terms of expression type, translation style, POS information and explicit statement, thus facilitating the construction of prompts endowed with distinct attributes tailored for various translation tasks. Specific experiments and cases are selected to validate and illustrate the effectiveness of the method.

Constrained decoding, a technique for enforcing constraints on language model outputs, offers a way to control text generation without retraining or architectural modifications. Its application is, however, typically restricted to models that give users access to next-token distributions (usually via softmax logits), which poses a limitation with blackbox large language models (LLMs). This paper introduces sketchguided constrained decoding (SGCD), a novel approach to constrained decoding for blackbox LLMs, which operates without access to the logits of the blackbox LLM. SGCD utilizes a locally hosted auxiliary model to refine the output of an unconstrained blackbox LLM, effectively treating this initial output as a "sketch" for further elaboration. This approach is complementary to traditional logit-based techniques and enables the application of constrained decoding in settings where full model transparency Figure 1: Overview of sketch-guided constrained decoding is unavailable. We demonstrate the efficacy of (SGCD). In the initial sketching phase, a blackbox SGCD through experiments in closed information LLM generates a preliminary "sketch" answer without extraction and constituency parsing, showing applying any constraints. Then, in the constrained how it enhances the utility and flexibility of generation phase, an auxiliary model, the constrained blackbox LLMs for complex NLP tasks.

Leveraging recent advancements in large language models, modern neural code completion models have demonstrated the capability to generate highly accurate code suggestions. However, their massive size poses challenges in terms of computational costs and environmental impact, hindering their widespread adoption in practical scenarios. Dynamic inference emerges as a promising solution, as it allocates minimal computation during inference while maintaining the model's performance. In this research, we explore dynamic inference within the context of code completion. Initially, we conducted an empirical investigation on GPT-2, focusing on the inference capabilities of intermediate layers for code completion. We found that 54.4% of tokens can be accurately generated using just the first layer, signifying significant computational savings potential. Moreover, despite using all layers, the model still fails to predict 14.5% of tokens correctly, and the subsequent completions continued from them are rarely considered helpful, with only a 4.2% Acceptance Rate. These findings motivate our exploration of dynamic inference in code completion and inspire us to enhance it with a decision-making mechanism that stops the generation of incorrect code. We thus propose a novel dynamic inference method specifically tailored for code completion models. This method aims not only to produce correct predictions with largely reduced computation but also to prevent incorrect predictions proactively. Our extensive evaluation shows that it can averagely skip 1.7 layers out of 16 layers in the models, leading to an 11.2% speedup with only a marginal 1.1% reduction in ROUGE-L.

Large Language Models (LLMs) have emerged as powerful tools for natural language processing tasks, revolutionizing the field with their ability to understand and generate human-like text. As the demand for more sophisticated LLMs continues to grow, there is a pressing need to address the computational challenges associated with their scale and complexity. This paper presents a comprehensive survey on hardware accelerators designed to enhance the performance and energy efficiency of Large Language Models. By examining a diverse range of accelerators, including GPUs, FPGAs, and custom-designed architectures, we explore the landscape of hardware solutions tailored to meet the unique computational demands of LLMs. The survey encompasses an in-depth analysis of architecture, performance metrics, and energy efficiency considerations, providing valuable insights for researchers, engineers, and decision-makers aiming to optimize the deployment of LLMs in real-world applications.

The advent of large language models (LLMs) such as Chat-GPT has attracted considerable attention in various domains due to their remarkable performance and versatility. As the use of these models continues to grow, the importance of effective prompt engineering has come to the fore. Prompt optimization emerges as a crucial challenge, as it has a direct impact on model performance and the extraction of relevant information. Recently, evolutionary algorithms (EAs) have shown promise in addressing this issue, paving the way for novel optimization strategies. In this work, we propose a evolutionary multi-objective (EMO) approach specifically tailored for prompt optimization called EMO-Prompts, using sentiment analysis as a case study. We use sentiment analysis capabilities as our experimental targets. Our results demonstrate that EMO-Prompts effectively generates prompts capable of guiding the LLM to produce texts embodying two conflicting emotions simultaneously.

Recent advancements in Multimodal Large Language Models (MLLMs) have significantly enhanced the comprehension of multimedia content, bringing together diverse modalities such as text, images, and videos. However, a critical challenge faced by these models, especially when processing video inputs, is the occurrence of hallucinations - erroneous perceptions or interpretations, particularly at the event level. This study introduces an innovative method to address event-level hallucinations in MLLMs, focusing on specific temporal understanding in video content. Our approach leverages a novel framework that extracts and utilizes event-specific information from both the event query and the provided video to refine MLLMs' response. We propose a unique mechanism that decomposes on-demand event queries into iconic actions. Subsequently, we employ models like CLIP and BLIP2 to predict specific timestamps for event occurrences. Our evaluation, conducted using the Charades-STA dataset, demonstrates a significant reduction in temporal hallucinations and an improvement in the quality of event-related responses. This research not only provides a new perspective in addressing a critical limitation of MLLMs but also contributes a quantitatively measurable method for evaluating MLLMs in the context of temporal-related questions.

Simulation technologies have been widely utilized in many scientific research fields such as weather forecasting, fluid mechanics and biological populations. It is the best tool to handle problems in complex systems, where closed-form expressions are unavailable and the target distribution in the representation space is too complex to be fully represented by a deep learning (DL) model. We believe that the development of simulation technologies is consistent with scientific paradigms. This paper induces the evolution of scientific paradigms from the perspective of data, algorithms, and computational power. Building upon this perspective, we divide simulation technologies into three stages aligning with the emergence of new paradigms, and find that advanced simulation technologies are typical instances of paradigms integration. Moreover, we propose the concept of behavioral simulation (BS), specifically sophisticated behavioral simulation (SBS), representing a higher degree of paradigms integration based on foundation models to simulate complex social systems involving sophisticated human strategies and behaviors. BS and further SBS are designed to tackle challenges concerning the complex human system that surpasses the capacity of traditional agent-based modeling simulation (ABMS), which can be regarded as a possible next paradigm for science. Through this work, we look forward to more powerful BS and SBS applications in scientific research branches within social science.