Personal use of this material is permitted. Abstract --As Large Language Models (LLMs) are widely used, understanding them systematically is key to improving their safety and realizing their full potential. Although many models are aligned using techniques such as reinforcement learning from human feedback (RLHF), they are still vulnerable to jailbreaking attacks. Some of the existing adversarial attack methods search for discrete tokens that may jailbreak a target model while others try to optimize the continuous space represented by the tokens of the model's vocabulary. While techniques based on the discrete space may prove to be inefficient, optimization of continuous token embeddings requires projections to produce discrete tokens, which might render them ineffective. T o fully utilize the constraints and the structures of the space, we develop an intrinsic optimization technique using exponentiated gradient descent with the Bregman projection method to ensure that the optimized one-hot encoding always stays within the probability simplex. We prove the convergence of the technique and implement an efficient algorithm that is effective in jailbreaking several widely used LLMs. We demonstrate the efficacy of the proposed technique using five open-source LLMs on four openly available datasets. The results show that the technique achieves a higher success rate with great efficiency compared to three other state-of-the-art jailbreaking techniques.

Recent advances in generative models have made it increasingly difficult to distinguish real data from model-generated synthetic data. Using synthetic data for successive training of future model generations creates "self-consuming loops", which may lead to model collapse or training instability. Furthermore, synthetic data is often subject to human feedback and curated by users based on their preferences. Ferbach et al. (2024) recently showed that when data is curated according to user preferences, the self-consuming retraining loop drives the model to converge toward a distribution that optimizes those preferences. However, in practice, data curation is often noisy or adversarially manipulated. For example, competing platforms may recruit malicious users to adversarially curate data and disrupt rival models. In this paper, we study how generative models evolve under self-consuming retraining loops with noisy and adversarially curated data. We theoretically analyze the impact of such noisy data curation on generative models and identify conditions for the robustness of the retraining process. Building on this analysis, we design attack algorithms for competitive adversarial scenarios, where a platform with a limited budget employs malicious users to misalign a rival's model from actual user preferences. Experiments on both synthetic and real-world datasets demonstrate the effectiveness of the proposed algorithms.

Spiking Large Language Models (LLMs) have emerged as an energy-efficient alternative to conventional LLMs through their event-driven computation. To effectively obtain spiking LLMs, researchers develop different ANN-to-SNN conversion methods by leveraging pre-trained ANN parameters while inheriting the energy efficiency of SNN. However, existing conversion methods struggle with extreme activation outliers and incompatible nonlinear operations of ANN-based LLMs. To address this, we propose a loss-less ANN-SNN conversion for fully spike-driven LLMs, termed LAS. Specifically, LAS introduces two novel neurons to convert the activation outlier and nonlinear operation of ANN-based LLMs. Moreover, LAS tailors the spike-equivalent Transformer components for spiking LLMs, which can ensure full spiking conversion without any loss of performance. Experimental results on six language models and two vision-language models demonstrate that LAS achieves loss-less conversion. Notably, on OPT-66B, LAS even improves the accuracy of 2\% on the WSC task. In addition, the parameter and ablation studies further verify the effectiveness of LAS. The source code is available at https://github.com/lc783/LAS

Large Language Models (LLMs) have gained significant popularity due to their remarkable capabilities in text understanding and generation. However, despite their widespread deployment in inference services such as ChatGPT, concerns about the potential leakage of sensitive user data have arisen. Existing solutions primarily rely on privacy-enhancing technologies to mitigate such risks, facing the trade-off among efficiency, privacy, and utility. To narrow this gap, we propose Cape, a context-aware prompt perturbation mechanism based on differential privacy, to enable efficient inference with an improved privacy-utility trade-off. Concretely, we introduce a hybrid utility function that better captures the token similarity. Additionally, we propose a bucketized sampling mechanism to handle large sampling space, which might lead to long-tail phenomenons. Extensive experiments across multiple datasets, along with ablation studies, demonstrate that Cape achieves a better privacy-utility trade-off compared to prior state-of-the-art works.

In the summer of 2023, Ilya Sutskever, a co-founder and the chief scientist of OpenAI, was meeting with a group of new researchers at the company. By all traditional metrics, Sutskever should have felt invincible: He was the brain behind the large language models that helped build ChatGPT, then the fastest-growing app in history; his company's valuation had skyrocketed; and OpenAI was the unrivaled leader of the industry believed to power the future of Silicon Valley. But the chief scientist seemed to be at war with himself. Sutskever had long believed that artificial general intelligence, or AGI, was inevitable--now, as things accelerated in the generative-AI industry, he believed AGI's arrival was imminent, according to Geoff Hinton, an AI pioneer who was his Ph.D. adviser and mentor, and another person familiar with Sutskever's thinking. To people around him, Sutskever seemed consumed by thoughts of this impending civilizational transformation. What would the world look like when a supreme AGI emerged and surpassed humanity? And what responsibility did OpenAI have to ensure an end state of extraordinary prosperity, not extraordinary suffering?

Ever since ChatGPT was released to the public in November 2022, people have been using it to generate text, from emails to blog posts to bad poetry, much of which they post online. Since that release, the companies that build the large language models (LLMs) on which such chatbots are based--such as OpenAI's GPT 3.5, the technology underlying ChatGPT--have also continued to put out newer versions of their models, training them with new text data, some of which they scraped off the Web. That means, inevitably, that some of the training data used to create LLMs did not come from humans, but from the LLMs themselves. That has led computer scientists to worry about a phenomenon they call model collapse. Basically, model collapse happens when the training data no longer matches real-world data, leading the new LLM to produce gibberish, in a 21st-century version of the classic computer aphorism "garbage in, garbage out."

The news: A bill that allows clinics to sell unproven treatments has been passed in Montana. Under the legislation, doctors can apply for a license to open an experimental treatment clinic and recommend and sell therapies not approved by the Food and Drug Administration (FDA) to their patients. Why it matters: Once it's signed by the governor, the law will be the most expansive in the country in allowing access to drugs that have not been fully tested. The bill allows for any drug produced in the state to be sold in it, providing it has been through phase I clinical trials--but these trials do not determine if the drug is effective. The big picture: The bill was drafted and lobbied for by people interested in extending human lifespans.

Code large language models (LLMs) have shown impressive capabilities on a multitude of software engineering tasks. In particular, they have demonstrated remarkable utility in the task of code repair. However, common benchmarks used to evaluate the performance of code LLMs are often limited to small-scale settings. In this work, we build upon kGym, which shares a benchmark for system-level Linux kernel bugs and a platform to run experiments on the Linux kernel. This paper introduces CrashFixer, the first LLM-based software repair agent that is applicable to Linux kernel bugs. Inspired by the typical workflow of a kernel developer, we identify the key capabilities an expert developer leverages to resolve a kernel crash. Using this as our guide, we revisit the kGym platform and identify key system improvements needed to practically run LLM-based agents at the scale of the Linux kernel (50K files and 20M lines of code). We implement these changes by extending kGym to create an improved platform - called kGymSuite, which will be open-sourced. Finally, the paper presents an evaluation of various repair strategies for such complex kernel bugs and showcases the value of explicitly generating a hypothesis before attempting to fix bugs in complex systems such as the Linux kernel. We also evaluated CrashFixer's capabilities on still open bugs, and found at least two patch suggestions considered plausible to resolve the reported bug.

We propose ELIS, a serving system for Large Language Models (LLMs) featuring an Iterative Shortest Remaining Time First (ISRTF) scheduler designed to efficiently manage inference tasks with the shortest remaining tokens. Current LLM serving systems often employ a first-come-first-served scheduling strategy, which can lead to the "head-of-line blocking" problem. To overcome this limitation, it is necessary to predict LLM inference times and apply a shortest job first scheduling strategy. However, due to the auto-regressive nature of LLMs, predicting the inference latency is challenging. ELIS addresses this challenge by training a response length predictor for LLMs using the BGE model, an encoder-based state-of-the-art model. Additionally, we have devised the ISRTF scheduling strategy, an optimization of shortest remaining time first tailored to existing LLM iteration batching. To evaluate our work in an industrial setting, we simulate streams of requests based on our study of real-world user LLM serving trace records. Furthermore, we implemented ELIS as a cloud-native scheduler system on Kubernetes to evaluate its performance in production environments. Our experimental results demonstrate that ISRTF reduces the average job completion time by up to 19.6%.

Pre-training on image-text colonoscopy records offers substantial potential for improving endoscopic image analysis, but faces challenges including non-informative background images, complex medical terminology, and ambiguous multi-lesion descriptions. We introduce Endo-CLIP, a novel self-supervised framework that enhances Contrastive Language-Image Pre-training (CLIP) for this domain. Endo-CLIP's three-stage framework--cleansing, attunement, and unification--addresses these challenges by: (1) removing background frames, (2) leveraging large language models (LLMs) to extract clinical attributes for fine-grained contrastive learning, and (3) employing patient-level cross-attention to resolve multi-polyp ambiguities. Extensive experiments demonstrate that Endo-CLIP significantly outperforms state-of-the-art pre-training methods in zero-shot and few-shot polyp detection and classification, paving the way for more accurate and clinically relevant endoscopic analysis. Code will be made publicly available on https://github.com/chrlott/