Inthispaper,weintroduce aPrompt-Agnostic Adversarial Perturbation (PAP) method for customized diffusion models.

Moreover, the model tends to generalize well to unseen tasks that did not appear in the training data.

Diffusion models have revolutionized customized text-to-image generation, allowing for efficient synthesis of photos from personal data with textual descriptions.

Moreover, the model tends to generalize well to unseen tasks that did not appear in the training data.

We show that language models' activations linearly encode when information was learned during training. Our setup involves creating a model with a known training order by sequentially fine-tuning Llama-3.2-1B on six disjoint but otherwise similar datasets about named entities. We find that the average activations of test samples corresponding to the six training datasets encode the training order: when projected into a 2D subspace, these centroids are arranged exactly in the order of training and lie on a straight line. Further, we show that linear probes can accurately (~90%) distinguish "early" vs. "late" entities, generalizing to entities unseen during the probes' own training. The model can also be fine-tuned to explicitly report an unseen entity's training stage (~80% accuracy). Interestingly, the training-order encoding does not seem attributable to simple differences in activation magnitudes, losses, or model confidence. Our paper demonstrates that models are capable of differentiating information by its acquisition time, and carries significant implications for how they might manage conflicting data and respond to knowledge modifications.

Unified multimodal large language models (U-MLLMs) have demonstrated impressive performance in visual understanding and generation in an end-to-end pipeline. Compared with generation-only models (e.g., Stable Diffusion), U-MLLMs may raise new questions about bias in their outputs, which can be affected by their unified capabilities. This gap is particularly concerning given the under-explored risk of propagating harmful stereotypes. In this paper, we benchmark the latest U-MLLMs and find that most exhibit significant demographic biases, such as gender and race bias. To better understand and mitigate this issue, we propose a locate-then-fix strategy, where we audit and show how the individual model component is affected by bias. Our analysis shows that bias originates primarily from the language model. More interestingly, we observe a "partial alignment" phenomenon in U-MLLMs, where understanding bias appears minimal, but generation bias remains substantial. Thus, we propose a novel balanced preference model to balance the demographic distribution with synthetic data. Experiments demonstrate that our approach reduces demographic bias while preserving semantic fidelity. We hope our findings underscore the need for more holistic interpretation and debiasing strategies of U-MLLMs in the future.

Adapter-based vision language models were an early attempt to augment large language models (LLMs) with image inputs (Liu et al., 2024). They use a pretrained image embedding model, like CLIP (Radford et al., 2021), and train adapters to map image embeddings directly into the embedding space of a pretrained LLM. However, separate input spaces can limit multimodal understanding and do not support native generation of images. In contrast, early-fusion multimodal models have been introduced as a more general approach that supports unlimited modalities as both input and output (Chameleon Team, 2024; Gemini Team, 2023; OpenAI, 2024). These models project all modalities into a shared tokenized space and are pretrained from scratch on multimodal inputs. In this work, we will refer to early-fusion multimodal models as multimodal fusion models. Just like LLMs, most vision language models are trained to behave safely and reject harmful requests (Bai et al., 2022). Carlini et al. (2024) demonstrated that bypassing safeguards in adapter-based vision language models is easy because input images can be continuously optimized to maximize harmful outputs. This is in contrast to text input optimization, which requires less efficient discrete optimization methods (Zou et al., 2023).

Large Language Models (LLMs) have great success in natural language processing tasks such as response generation. However, their use in tabular data has been limited due to their inferior performance compared to traditional machine learning models (TMLs) such as XGBoost. We find that the pre-trained knowledge of LLMs enables them to interpret new variables that appear in a test without additional training, a capability central to the concept of Out-of-Variable (OOV). From the findings, we propose a Language-Based-Classifier (LBC), a classifier that maximizes the benefits of LLMs to outperform TMLs on OOV tasks. LBC employs three key methodological strategies: 1) Categorical changes to adjust data to better fit the model's understanding, 2) Advanced order and indicator to enhance data representation to the model, and 3) Using verbalizer to map logit scores to classes during inference to generate model predictions. These strategies, combined with the pre-trained knowledge of LBC, emphasize the model's ability to effectively handle OOV tasks. We empirically and theoretically validate the superiority of LBC. LBC is the first study to apply an LLM-based model to OOV tasks. The source code is at https://github.com/sksmssh/LBCforOOVGen

Large Language Models (LLMs) become the start-of-the-art solutions for a variety of natural language tasks and are integrated into real-world applications. However, LLMs can be potentially harmful in manifesting undesirable safety issues like social biases and toxic content. It is imperative to assess its safety issues before deployment. However, the quality and diversity of test prompts generated by existing methods are still far from satisfactory. Not only are these methods labor-intensive and require large budget costs, but the controllability of test prompt generation is lacking for the specific testing domain of LLM applications. With the idea of LLM for LLM testing, we propose the first LLM, called TroubleLLM, to generate controllable test prompts on LLM safety issues. Extensive experiments and human evaluation illustrate the superiority of TroubleLLM on generation quality and generation controllability.

The past year has seen rapid acceleration in the development of large language models (LLMs). However, without proper steering and safeguards, LLMs will readily follow malicious instructions, provide unsafe advice, and generate toxic content. We introduce SimpleSafetyTests (SST) as a new test suite for rapidly and systematically identifying such critical safety risks. The test suite comprises 100 test prompts across five harm areas that LLMs, for the vast majority of applications, should refuse to comply with. We test 11 open-access and open-source LLMs and four closed-source LLMs, and find critical safety weaknesses. While some of the models do not give a single unsafe response, most give unsafe responses to more than 20% of the prompts, with over 50% unsafe responses in the extreme. Prepending a safety-emphasising system prompt substantially reduces the occurrence of unsafe responses, but does not completely stop them from happening. Trained annotators labelled every model response to SST (n = 3,000). We use these annotations to evaluate five AI safety filters (which assess whether a models' response is unsafe given a prompt) as a way of automatically evaluating models' performance on SST. The filters' performance varies considerably. There are also differences across the five harm areas, and on the unsafe versus safe responses. The widely-used Perspective API has 72% accuracy and a newly-created zero-shot prompt to OpenAI's GPT-4 performs best with 89% accuracy. Content Warning: This paper contains prompts and responses that relate to child abuse, suicide, self-harm and eating disorders, scams and fraud, illegal items, and physical harm.