Existing evaluation frameworks for retrieval-augmented generation (RAG) systems focus on answerable queries, but they overlook the importance of appropriately rejecting unanswerable requests. In this paper, we introduce UAEval4RAG, a framework designed to evaluate whether RAG systems can handle unanswerable queries effectively. We define a taxonomy with six unanswerable categories, and UAEval4RAG automatically synthesizes diverse and challenging queries for any given knowledge base with unanswered ratio and acceptable ratio metrics. We conduct experiments with various RAG components, including retrieval models, rewriting methods, rerankers, language models, and prompting strategies, and reveal hidden trade-offs in performance of RAG systems. Our findings highlight the critical role of component selection and prompt design in optimizing RAG systems to balance the accuracy of answerable queries with high rejection rates of unanswerable ones. UAEval4RAG provides valuable insights and tools for developing more robust and reliable RAG systems.

We present SoundMorpher, an open-world sound morphing method designed to generate perceptually uniform morphing trajectories. Traditional sound morphing techniques typically assume a linear relationship between the morphing factor and sound perception, achieving smooth transitions by linearly interpolating the semantic features of source and target sounds while gradually adjusting the morphing factor. However, these methods oversimplify the complexities of sound perception, resulting in limitations in morphing quality. In contrast, SoundMorpher explores an explicit relationship between the morphing factor and the perception of morphed sounds, leveraging log Mel-spectrogram features. This approach further refines the morphing sequence by ensuring a constant target perceptual difference for each transition and determining the corresponding morphing factors using binary search. To address the lack of a formal quantitative evaluation framework for sound morphing, we propose a set of metrics based on three established objective criteria. These metrics enable comprehensive assessment of morphed results and facilitate direct comparisons between methods, fostering advancements in sound morphing research. Extensive experiments demonstrate the effectiveness and versatility of SoundMorpher in real-world scenarios, showcasing its potential in applications such as creative music composition, film post-production, and interactive audio technologies. Our demonstration and codes are available at~\url{https://xinleiniu.github.io/SoundMorpher-demo/}.

UPenn Wharton School Associate Professor Ethan Mollick weighs in on the Biden White House's new guidelines for artificial intelligence in the workplace on'Fox News Live.' The parents of a Massachusetts high school senior who used artificial intelligence (AI) for a social studies project have filed a lawsuit against his teachers and the school after their son received detention and a "D" grade. "He's been accused of cheating, and it wasn't cheating, there was no rule in the handbook against AI," Jennifer Harris, who along with her husband, Dale, are named as plaintiffs in the lawsuit filed in Massachusetts' Plymouth County District Court last month against the Hingham High School administration and the school district, told Boston 25 News. The lawsuit alleges that their son will "suffer irreparable harm that is imminent" over the grade that his parents say kept him out of the National Honor Society, which they claim is threatening his standing with top tier colleges. "So, our argument to the school was could you fail him with a 59 instead of a 53 so he can have a B minus? He's applying to top tier schools," Harris told the news station.

Large language models (LLMs) undergo safety alignment to ensure safe conversations with humans. However, this paper introduces a training-free attack method capable of reversing safety alignment, converting the outcomes of stronger alignment into greater potential for harm by accessing only LLM output token distributions. Specifically, our method achieves this reversal by contrasting the output token distribution of a safety-aligned language model (e.g., Llama-2-chat) against its pre-trained version (e.g., Llama-2), so that the token predictions are shifted towards the opposite direction of safety alignment. We name this method emulated disalignment (ED) because sampling from this contrastive distribution provably emulates the result of fine-tuning to minimize a safety reward. Our experiments with ED across three evaluation datasets and four model families (Llama-1, Llama-2, Mistral, and Alpaca) show that ED doubles the harmfulness of pre-trained models and outperforms strong baselines, achieving the highest harmful rates in 43 out of 48 evaluation subsets by a large margin. Eventually, given ED's reliance on language model output token distributions, which particularly compromises open-source models, our findings highlight the need to reassess the open accessibility of language models, even if they have been safety-aligned. Code is available at https://github.com/ZHZisZZ/emulated-disalignment.

Large Language Models (LLMs) sometimes suffer from producing hallucinations, especially LLMs may generate untruthful responses despite knowing the correct knowledge. Activating the truthfulness within LLM is the key to fully unlocking LLM's knowledge potential. In this paper, we propose TruthX, an inference-time intervention method to activate the truthfulness of LLM by identifying and editing the features within LLM's internal representations that govern the truthfulness. TruthX employs an auto-encoder to map LLM's representations into semantic and truthful latent spaces respectively, and applies contrastive learning to identify a truthful editing direction within the truthful space. During inference, by editing LLM's internal representations in truthful space, TruthX effectively enhances the truthfulness of LLM. Experiments show that TruthX improves the truthfulness of 13 advanced LLMs by an average of 20% on TruthfulQA benchmark. Further analyses suggest that TruthX can control LLM to produce truthful or hallucinatory responses via editing only one vector in LLM's internal representations.

Sentiment or mood can express themselves on various levels in music. In automatic analysis, the actual audio data is usually analyzed, but the lyrics can also play a crucial role in the perception of moods. We first evaluate various models for sentiment analysis based on lyrics and audio separately. The corresponding approaches already show satisfactory results, but they also exhibit weaknesses, the causes of which we examine in more detail. Furthermore, different approaches to combining the audio and lyrics results are proposed and evaluated. Considering both modalities generally leads to improved performance. We investigate misclassifications and (also intentional) contradictions between audio and lyrics sentiment more closely, and identify possible causes. Finally, we address fundamental problems in this research area, such as high subjectivity, lack of data, and inconsistency in emotion taxonomies.

Despite the great success of large language models (LLMs) in various tasks, they suffer from generating hallucinations. We introduce Truth Forest, a method that enhances truthfulness in LLMs by uncovering hidden truth representations using multi-dimensional orthogonal probes. Specifically, it creates multiple orthogonal bases for modeling truth by incorporating orthogonal constraints into the probes. Moreover, we introduce Random Peek, a systematic technique considering an extended range of positions within the sequence, reducing the gap between discerning and generating truth features in LLMs. By employing this approach, we improved the truthfulness of Llama-2-7B from 40.8\% to 74.5\% on TruthfulQA. Likewise, significant improvements are observed in fine-tuned models. We conducted a thorough analysis of truth features using probes. Our visualization results show that orthogonal probes capture complementary truth-related features, forming well-defined clusters that reveal the inherent structure of the dataset.

This paper introduces the Terrain Recognition And Contact Force Estimation Paw, a compact and sensorized shoe designed for legged robots. The paw end-effector is made of silicon that deforms upon the application of contact forces, while an embedded micro camera is utilized to capture images of the deformed inner surface inside the shoe, and a microphone picks up audio signals. Processed through machine learning techniques, the images are mapped to compute an accurate estimate of the cumulative 3D force vector, while the audio signals are analyzed to identify the terrain class (e.g., gravel, snow). By leveraging its on-edge computation ability, the paw enhances the capabilities of legged robots by providing key information in real-time that can be used to adapt locomotion control strategies. To assess the performance of this novel sensorized paw, we conducted experiments on the data collected through a specially-designed testbed for force estimation, as well as data from recordings of the audio signatures of different terrains interacting with the paw. The results demonstrate the accuracy and effectiveness of the system, highlighting its potential for improving legged robot performance.

We study how to perform unlearning, i.e. forgetting undesirable (mis)behaviors, on large language models (LLMs). We show at least three scenarios of aligning LLMs with human preferences can benefit from unlearning: (1) removing harmful responses, (2) erasing copyright-protected content as requested, and (3) eliminating hallucinations. Unlearning, as an alignment technique, has three advantages. (1) It only requires negative (e.g. harmful) examples, which are much easier and cheaper to collect (e.g. via red teaming or user reporting) than positive (e.g. helpful and often human-written) examples required in RLHF (RL from human feedback). (2) It is computationally efficient. (3) It is especially effective when we know which training samples cause the misbehavior. To the best of our knowledge, our work is among the first to explore LLM unlearning. We are also among the first to formulate the settings, goals, and evaluations in LLM unlearning. We show that if practitioners only have limited resources, and therefore the priority is to stop generating undesirable outputs rather than to try to generate desirable outputs, unlearning is particularly appealing. Despite only having negative samples, our ablation study shows that unlearning can still achieve better alignment performance than RLHF with just 2% of its computational time.

With the advent of modern AI architectures, a shift has happened towards end-to-end architectures. This pivot has led to neural architectures being trained without domain-specific biases/knowledge, optimized according to the task. We in this paper, learn audio embeddings via diverse feature representations, in this case, domain-specific. For the case of audio classification over hundreds of categories of sound, we learn robust separate embeddings for diverse audio properties such as pitch, timbre, and neural representation, along with also learning it via an end-to-end architecture. We observe handcrafted embeddings, e.g., pitch and timbre-based, although on their own, are not able to beat a fully end-to-end representation, yet adding these together with end-to-end embedding helps us, significantly improve performance. This work would pave the way to bring some domain expertise with end-to-end models to learn robust, diverse representations, surpassing the performance of just training end-to-end models.