Legilimens: Practical and Unified Content Moderation for Large Language Model Services

Given the societal impact of unsafe content generated by large language models (LLMs), ensuring that LLM services comply with safety standards is a crucial concern for LLM service providers. Common content moderation methods are limited by an effectiveness-and-efficiency dilemma, where simple models are fragile while sophisticated models consume excessive computational resources. In this paper, we reveal for the first time that effective and efficient content moderation can be achieved by extracting conceptual features from chat-oriented LLMs, despite their initial fine-tuning for conversation rather than content moderation. We propose a practical and unified content moderation framework for LLM services, named Legilimens, which features both effectiveness and efficiency. Our red-team model-based data augmentation enhances the robustness of Legilimens against state-of-the-art jailbreaking. Additionally, we develop a framework to theoretically analyze the cost-effectiveness of Legilimens compared to other methods. We have conducted extensive experiments on five host LLMs, seventeen datasets, and nine jailbreaking methods to verify the effectiveness, efficiency, and robustness of Legilimens against normal and adaptive adversaries. A comparison of Legilimens with both commercial and academic baselines demonstrates the superior performance of Legilimens. Furthermore, we confirm that Legilimens can be applied to few-shot scenarios and extended to multi-label classification tasks.

MEDIC: Zero-shot Music Editing with Disentangled Inversion Control

Text-guided diffusion models catalyze a paradigm shift in audio generation, facilitating the adaptability of source audio to conform to specific textual prompts. Recent advancements introduce inversion techniques, like DDIM inversion, to zero-shot editing, exploiting pre-trained diffusion models for audio modification. Nonetheless, our investigation exposes that DDIM inversion suffers from an accumulation of errors across each diffusion step, undermining its efficacy. And the lack of attention control hinders the fine-grained manipulations of music. To counteract these limitations, we introduce the \textit{Disentangled Inversion} technique, which is designed to disentangle the diffusion process into triple branches, thereby magnifying their individual capabilities for both precise editing and preservation. Furthermore, we propose the \textit{Harmonized Attention Control} framework, which unifies the mutual self-attention and cross-attention with an additional Harmonic Branch to achieve the desired composition and structural information in the target music. Collectively, these innovations comprise the \textit{Disentangled Inversion Control (DIC)} framework, enabling accurate music editing whilst safeguarding structural integrity. To benchmark audio editing efficacy, we introduce \textit{ZoME-Bench}, a comprehensive music editing benchmark hosting 1,100 samples spread across 10 distinct editing categories, which facilitates both zero-shot and instruction-based music editing tasks. Our method demonstrates unparalleled performance in edit fidelity and essential content preservation, outperforming contemporary state-of-the-art inversion techniques.

Conditionally Parameterized, Discretization-Aware Neural Networks for Mesh-Based Modeling of Physical Systems

The numerical simulations of physical systems are heavily dependent on mesh-based models. While neural networks have been extensively explored to assist such tasks, they often ignore the interactions or hierarchical relations between input features, and process them as concatenated mixtures. In this work, we generalize the idea of conditional parametrization -- using trainable functions of input parameters to generate the weights of a neural network, and extend them in a flexible way to encode information critical to the numerical simulations. Inspired by discretized numerical methods, choices of the parameters include physical quantities and mesh topology features. The functional relation between the modeled features and the parameters are built into the network architecture. The method is implemented on different networks, which are applied to several frontier scientific machine learning tasks, including the discovery of unmodeled physics, super-resolution of coarse fields, and the simulation of unsteady flows with chemical reactions. The results show that the conditionally parameterized networks provide superior performance compared to their traditional counterparts. A network architecture named CP-GNet is also proposed as the first deep learning model capable of standalone prediction of reacting flows on irregular meshes.

Multi-level Convolutional Autoencoder Networks for Parametric Prediction of Spatio-temporal Dynamics

A data-driven framework is proposed for the predictive modeling of complex spatio-temporal dynamics, leveraging nested non-linear manifolds. Three levels of neural networks are used, with the goal of predicting the future state of a system of interest in a parametric setting. A convolutional autoencoder is used as the top level to encode the high dimensional input data along spatial dimensions into a sequence of latent variables. A temporal convolutional autoencoder serves as the second level, which further encodes the output sequence from the first level along the temporal dimension, and outputs a set of latent variables that encapsulate the spatio-temporal evolution of the dynamics. A fully connected network is used as the third level to learn the mapping between these latent variables and the global parameters from training data, and predict them for new parameters. For future state predictions, the second level uses a temporal convolutional network to predict subsequent steps of the output sequence from the top level. Latent variables at the bottom-most level are decoded to obtain the dynamics in physical space at new global parameters and/or at a future time. The framework is evaluated on a range of problems involving discontinuities, wave propagation, strong transients, and coherent structures. The sensitivity of the results to different modeling choices is assessed. The results suggest that given adequate data and careful training, effective data-driven predictive models can be constructed. Perspectives are provided on the present approach and its place in the landscape of model reduction.