GermanPartiesQA: Benchmarking Commercial Large Language Models for Political Bias and Sycophancy Jan Batzner 1, 3 *, V olker Stocker 1, 2, Stefan Schmid 2, 1, Gjergji Kasneci 3 1 Weizenbaum Institute Berlin 2 Technical University Berlin 3 Technical University Munich Abstract LLMs are changing the way humans create and interact with content, potentially affecting citizens' political opinions and voting decisions. As LLMs increasingly shape our digital information ecosystems, auditing to evaluate biases, sycophancy, or steerability has emerged as an active field of research. In this paper, we evaluate and compare the alignment of six LLMs by OpenAI, Anthropic, and Cohere with German party positions and evaluate sycophancy based on a prompt experiment. We contribute to evaluating political bias and sycophancy in multi-party systems across major commercial LLMs. First, we develop the benchmark dataset GermanPar-tiesQA based on the V oting Advice Application W ahl-o-Mat covering 10 state and 1 national elections between 2021 and 2023. In our study, we find a left-green tendency across all examined LLMs. We then conduct our prompt experiment for which we use the benchmark and sociodemographic data of leading German parliamentarians to evaluate changes in LLMs responses. To differentiate between sycophancy and steerabilty, we use "I am [politician X], ... " and "Y ou are [politician X], ... " prompts. Against our expectations, we do not observe notable differences between prompting "I am" and "Y ou are". While our findings underscore that LLM responses can be ideologically steered with political personas, they suggest that observed changes in LLM outputs could be better described as personalization to the given context rather than sycophancy. 1 INTRODUCTION Large language models (LLMs) are changing the way humans create and consume content. The unprecedented pace with which end-users have adopted ChatGPT [14] has not only brought LLMs and generative AI to public attention but has emphasized their increasing potential to influence societal, economic, and political outcomes. Generative AI applications can impact citizens in various ways directly or indirectly as they may be consumer-facing (e.g., LLM-based chat interfaces like ChatGPT) or not (e.g., users may interact with content created by or with the support of LLMs, with the role of LLMs being less transparent to citizens).

Low-rank adaptation, also known as LoRA, has emerged as a prominent method for parameter-efficient fine-tuning of foundation models. Despite its computational efficiency, LoRA still yields inferior performance compared to full fine-tuning. In this paper, we first uncover a fundamental connection between the optimization processes of LoRA and full fine-tuning: using LoRA for optimization is mathematically equivalent to full fine-tuning using a low-rank gradient for parameter updates. And this low-rank gradient can be expressed in terms of the gradients of the two low-rank matrices in LoRA. Leveraging this insight, we introduce LoRA-Pro, a method that enhances LoRA's performance by strategically adjusting the gradients of these low-rank matrices. This adjustment allows the low-rank gradient to more accurately approximate the full fine-tuning gradient, thereby narrowing the performance gap between LoRA and full fine-tuning. Furthermore, we theoretically derive the optimal solutions for adjusting the gradients of the low-rank matrices, applying them during fine-tuning in LoRA-Pro. We conduct extensive experiments across natural language understanding, dialogue generation, mathematical reasoning, code generation, and image classification tasks, demonstrating that LoRA-Pro substantially improves LoRA's performance, effectively narrowing the gap with full fine-tuning. Foundational models (Radford et al., 2021; Brown et al., 2020; Achiam et al., 2023; Kirillov et al., 2023; Rombach et al., 2022; Touvron et al., 2023) have become the cornerstone of modern deep learning. Remarkably, some foundation models even demonstrate emergent properties (Hoffmann et al., 2022; Kaplan et al., 2020). Due to these advantages, foundational models have been widely applied to various downstream applications. Nevertheless, it still requires additional fine-tuning when applied to downstream tasks, where the huge parameter size of foundation models result in high cost in this stage.

This paper presents Hardware Description Language Generative Pre-trained Transformers (HDL-GPT), a novel approach that leverages the vast repository of open-source High Definition Language (HDL) codes to train superior quality large code models. The core premise of this paper is the hypothesis that high-quality HDL is all you need to create models with exceptional performance and broad zero-shot generalization abilities. The paper elucidates the methods employed for the curation and augmentation of large corpora from open-source HDL code, transforming highly variable quality data into high-quality data through careful prompting and context maintenance. We demonstrate that the careful selection, filtering, and augmentation of data across HDLs can yield powerful models that surpass current state-of-the-art models. We also explore the impact of different fine-tuning methods on the quality of results. We describe experimental results across a range of fine-tuned SOTA LLMs, substantiating our claims. We demonstrate improvements of 50% to 200% over SOTA HDL models on current benchmarks in tasks ranging from HDL circuit explanations, code generation, formal and simulation testbench creation, triaging bugs, and fixing them. HDL-GPT opens new avenues for the development of advanced model training techniques for circuit design tasks.

Improving Domain-Specific ASR with LLM-Generated Contextual Descriptions Jiwon Suh, Injae Na, W oohwan Jung Department of Applied Artificial Intelligence, Hanyang University, Republic of Korea {jwsuh0205, suhoij47, whjung}@hanyang.ac.kr Abstract End-to-end automatic speech recognition (E2E ASR) systems have significantly improved speech recognition through training on extensive datasets. Despite these advancements, they still struggle to accurately recognize domain specific words, such as proper nouns and technical terminologies. To address this problem, we propose a method to utilize the state-of-the-art Whisper without modifying its architecture, preserving its generalization performance while enabling it to leverage descriptions effectively. Moreover, we propose two additional training techniques to improve the domain specific ASR: decoder fine-tuning, and context perturbation. We also propose a method to use a Large Language Model (LLM) to generate descriptions with simple metadata, when descriptions are unavailable. Our experiments demonstrate that proposed methods notably enhance domain-specific ASR accuracy on real-life datasets, with LLMgenerated descriptions outperforming human-crafted ones in effectiveness. Introduction Recent advancements in end-to-end (E2E) automatic speech recognition (ASR) systems, such as Wav2V ec 2.0 [1] and Whisper [2], have significantly improved the capabilities of speech recognition through extensive training on large datasets. However, these systems often encounter difficulties in accurately identifying domain specific terms, such as proper nouns and technical jargon.

Previous research on testing the vulnerabilities in Large Language Models (LLMs) using adversarial attacks has primarily focused on nonsensical prompt injections, which are easily detected upon manual or automated review (e.g., via byte entropy). However, the exploration of innocuous human-understandable malicious prompts augmented with adversarial injections remains limited. In this research, we explore converting a nonsensical suffix attack into a sensible prompt via a situation-driven contextual re-writing. This allows us to show suffix conversion without any gradients, using only LLMs to perform the attacks, and thus better understand the scope of possible risks. We combine an independent, meaningful adversarial insertion and situations derived from movies to check if this can trick an LLM. The situations are extracted from the IMDB dataset, and prompts are defined following a few-shot chain-of-thought prompting. Our approach demonstrates that a successful situation-driven attack can be executed on both open-source and proprietary LLMs. We find that across many LLMs, as few as 1 attempt produces an attack and that these attacks transfer between LLMs.

Diffusion models revolutionize image generation by leveraging natural language to guide the creation of multimedia content. Despite significant advancements in such generative models, challenges persist in depicting detailed human-object interactions, especially regarding pose and object placement accuracy. We introduce a training-free method named Reasoning and Correcting Diffusion (ReCorD) to address these challenges. Our model couples Latent Diffusion Models with Visual Language Models to refine the generation process, ensuring precise depictions of HOIs. We propose an interaction-aware reasoning module to improve the interpretation of the interaction, along with an interaction correcting module to refine the output image for more precise HOI generation delicately. Through a meticulous process of pose selection and object positioning, ReCorD achieves superior fidelity in generated images while efficiently reducing computational requirements. We conduct comprehensive experiments on three benchmarks to demonstrate the significant progress in solving text-to-image generation tasks, showcasing ReCorD's ability to render complex interactions accurately by outperforming existing methods in HOI classification score, as well as FID and Verb CLIP-Score. Project website is available at https://alberthkyhky.github.io/ReCorD/ .

Retrieval-augmented Generation (RAG) has markedly enhanced the capabilities of Large Language Models (LLMs) in tackling knowledge-intensive tasks. The increasing demands of application scenarios have driven the evolution of RAG, leading to the integration of advanced retrievers, LLMs and other complementary technologies, which in turn has amplified the intricacy of RAG systems. However, the rapid advancements are outpacing the foundational RAG paradigm, with many methods struggling to be unified under the process of "retrieve-then-generate". In this context, this paper examines the limitations of the existing RAG paradigm and introduces the modular RAG framework. By decomposing complex RAG systems into independent modules and specialized operators, it facilitates a highly reconfigurable framework. Modular RAG transcends the traditional linear architecture, embracing a more advanced design that integrates routing, scheduling, and fusion mechanisms. Drawing on extensive research, this paper further identifies prevalent RAG patterns-linear, conditional, branching, and looping-and offers a comprehensive analysis of their respective implementation nuances. Modular RAG presents innovative opportunities for the conceptualization and deployment of RAG systems. Finally, the paper explores the potential emergence of new operators and paradigms, establishing a solid theoretical foundation and a practical roadmap for the continued evolution and practical deployment of RAG technologies.

The integration of large language models (LLMs) with vision-language (VL) tasks has been a transformative development in the realm of artificial intelligence, highlighting the potential of LLMs as a versatile general-purpose chatbot. However, the current trend in this evolution focuses on the integration of vision and language to create models that can operate in more diverse and real-world contexts. We present a novel approach, termed Bottleneck Adapter, specifically crafted for enhancing the multimodal functionalities of these complex models, enabling joint optimization of the entire multimodal LLM framework through a process known as Multimodal Model Tuning (MMT). Our approach utilizes lightweight adapters to connect the image encoder and LLM without the need for large, complex neural networks. Unlike the conventional modular training schemes, our approach adopts an end-to-end optimization regime, which, when combined with the adapters, facilitates the joint optimization using a significantly smaller parameter set. Our method exhibits robust performance with 90.12\% accuracy, outperforming both human-level performance (88.4\%) and LaVIN-7B (89.41\%).

Large Language Models (LLMs) frequently memorize long sequences verbatim, often with serious legal and privacy implications. Much prior work has studied such verbatim memorization using observational data. To complement such work, we develop a framework to study verbatim memorization in a controlled setting by continuing pre-training from Pythia checkpoints with injected sequences. We find that (1) non-trivial amounts of repetition are necessary for verbatim memorization to happen; (2) later (and presumably better) checkpoints are more likely to verbatim memorize sequences, even for out-of-distribution sequences; (3) the generation of memorized sequences is triggered by distributed model states that encode high-level features and makes important use of general language modeling capabilities. Guided by these insights, we develop stress tests to evaluate unlearning methods and find they often fail to remove the verbatim memorized information, while also degrading the LM. Overall, these findings challenge the hypothesis that verbatim memorization stems from specific model weights or mechanisms. Rather, verbatim memorization is intertwined with the LM's general capabilities and thus will be very difficult to isolate and suppress without degrading model quality.

This paper investigates visual analogical reasoning in large multimodal models (LMMs) compared to human adults and children. A "visual analogy" is an abstract rule inferred from one image and applied to another. While benchmarks exist for testing visual reasoning in LMMs, they require advanced skills and omit basic visual analogies that even young children can make. Inspired by developmental psychology, we propose a new benchmark of 1,400 visual transformations of everyday objects to test LMMs on visual analogical reasoning and compare them to children and adults. We structure the evaluation into three stages: identifying what changed (e.g., color, number, etc.), how it changed (e.g., added one object), and applying the rule to new scenarios. Our findings show that while models like GPT-4V, LLaVA-1.5, and MANTIS identify the "what" effectively, they struggle with quantifying the "how" and extrapolating this rule to new objects. In contrast, children and adults exhibit much stronger analogical reasoning at all three stages. Additionally, the strongest tested model, GPT-4V, performs better in tasks involving simple visual attributes like color and size, correlating with quicker human adult response times. Conversely, more complex tasks such as number, rotation, and reflection, which necessitate extensive cognitive processing and understanding of the 3D physical world, present more significant challenges. Altogether, these findings highlight the limitations of training models on data that primarily consists of 2D images and text.