Foundation Cures Personalization: Recovering Facial Personalized Models' Prompt Consistency

Facial personalization represents a crucial downstream task in the domain of text-to-image generation. To preserve identity fidelity while ensuring alignment with user-defined prompts, current mainstream frameworks for facial personalization predominantly employ identity embedding mechanisms to associate identity information with textual embeddings. However, our experiments show that identity embeddings compromise the effectiveness of other tokens within the prompt, thereby hindering high prompt consistency, particularly when prompts involve multiple facial attributes. Moreover, previous works overlook the fact that their corresponding foundation models hold great potential to generate faces aligning to prompts well and can be easily leveraged to cure these ill-aligned attributes in personalized models. Building upon these insights, we propose FreeCure, a training-free framework that harnesses the intrinsic knowledge from the foundation models themselves to improve the prompt consistency of personalization models. First, by extracting cross-attention and semantic maps from the denoising process of foundation models, we identify easily localized attributes (e.g., hair, accessories, etc). Second, we enhance multiple attributes in the outputs of personalization models through a novel noise-blending strategy coupled with an inversion-based process. Our approach offers several advantages: it eliminates the need for training; it effectively facilitates the enhancement for a wide array of facial attributes in a non-intrusive manner; and it can be seamlessly integrated into existing popular personalization models. FreeCure has demonstrated significant improvements in prompt consistency across a diverse set of state-of-the-art facial personalization models while maintaining the integrity of original identity fidelity.

HiRes-LLaVA: Restoring Fragmentation Input in High-Resolution Large Vision-Language Models

High-resolution inputs enable Large Vision-Language Models (LVLMs) to discern finer visual details, enhancing their comprehension capabilities. To reduce the training and computation costs caused by high-resolution input, one promising direction is to use sliding windows to slice the input into uniform patches, each matching the input size of the well-trained vision encoder. Although efficient, this slicing strategy leads to the fragmentation of original input, i.e., the continuity of contextual information and spatial geometry is lost across patches, adversely affecting performance in cross-patch context perception and position-specific tasks. To overcome these shortcomings, we introduce HiRes-LLaVA, a novel framework designed to efficiently process any size of high-resolution input without altering the original contextual and geometric information. HiRes-LLaVA comprises two innovative components: (i) a SliceRestore adapter that reconstructs sliced patches into their original form, efficiently extracting both global and local features via down-up-sampling and convolution layers, and (ii) a Self-Mining Sampler to compresses the vision tokens based on themselves, preserving the original context and positional information while reducing training overhead. To assess the ability of handling context fragmentation, we construct a new benchmark, EntityGrid-QA, consisting of edge-related and position-related tasks. Our comprehensive experiments demonstrate the superiority of HiRes-LLaVA on both existing public benchmarks and on EntityGrid-QA, particularly on document-oriented tasks, establishing new standards for handling high-resolution inputs.

See Through Their Minds: Learning Transferable Neural Representation from Cross-Subject fMRI

Deciphering visual content from functional Magnetic Resonance Imaging (fMRI) helps illuminate the human vision system. However, the scarcity of fMRI data and noise hamper brain decoding model performance. Previous approaches primarily employ subject-specific models, sensitive to training sample size. In this paper, we explore a straightforward but overlooked solution to address data scarcity. We propose shallow subject-specific adapters to map cross-subject fMRI data into unified representations. Subsequently, a shared deeper decoding model decodes cross-subject features into the target feature space. During training, we leverage both visual and textual supervision for multi-modal brain decoding. Our model integrates a high-level perception decoding pipeline and a pixel-wise reconstruction pipeline guided by high-level perceptions, simulating bottom-up and top-down processes in neuroscience. Empirical experiments demonstrate robust neural representation learning across subjects for both pipelines. Moreover, merging high-level and low-level information improves both low-level and high-level reconstruction metrics. Additionally, we successfully transfer learned general knowledge to new subjects by training new adapters with limited training data. Compared to previous state-of-the-art methods, notably pre-training-based methods (Mind-Vis and fMRI-PTE), our approach achieves comparable or superior results across diverse tasks, showing promise as an alternative method for cross-subject fMRI data pre-training. Our code and pre-trained weights will be publicly released at https://github.com/YulongBonjour/See_Through_Their_Minds.

From Summary to Action: Enhancing Large Language Models for Complex Tasks with Open World APIs

The distinction between humans and animals lies in the unique ability of humans to use and create tools. Tools empower humans to overcome physiological limitations, fostering the creation of magnificent civilizations. Similarly, enabling foundational models like Large Language Models (LLMs) with the capacity to learn external tool usage may serve as a pivotal step toward realizing artificial general intelligence. Previous studies in this field have predominantly pursued two distinct approaches to augment the tool invocation capabilities of LLMs. The first approach emphasizes the construction of relevant datasets for model fine-tuning. The second approach, in contrast, aims to fully exploit the inherent reasoning abilities of LLMs through in-context learning strategies. In this work, we introduce a novel tool invocation pipeline designed to control massive real-world APIs. This pipeline mirrors the human task-solving process, addressing complicated real-life user queries. At each step, we guide LLMs to summarize the achieved results and determine the next course of action. We term this pipeline `from Summary to action', Sum2Act for short. Empirical evaluations of our Sum2Act pipeline on the ToolBench benchmark show significant performance improvements, outperforming established methods like ReAct and DFSDT. This highlights Sum2Act's effectiveness in enhancing LLMs for complex real-world tasks.

BrainCLIP: Bridging Brain and Visual-Linguistic Representation via CLIP for Generic Natural Visual Stimulus Decoding from fMRI

Reconstructing perceived natural images or decoding their categories from fMRI signals are challenging tasks with great scientific significance. Due to the lack of paired samples, most existing methods fail to generate semantically recognizable reconstruction and are difficult to generalize to novel classes. In this work, we propose, for the first time, a task-agnostic brain decoding model by unifying the visual stimulus classification and reconstruction tasks in a semantic space. We denote it as BrainCLIP, which leverages CLIP's cross-modal generalization ability to bridge the modality gap between brain activities, images, and texts. Specifically, BrainCLIP is a VAE-based architecture that transforms fMRI patterns into the CLIP embedding space by combining visual and textual supervision. Note that previous works rarely use multi-modal supervision for visual stimulus decoding. Our experiments demonstrate that textual supervision can significantly boost the performance of decoding models compared to the condition where only image supervision exists. BrainCLIP can be applied to multiple scenarios like fMRI-to-image generation, fMRI-image-matching, and fMRI-text-matching. Compared with BraVL, a recently proposed multi-modal method for fMRI-based brain decoding, BrainCLIP achieves significantly better performance on the novel class classification task. BrainCLIP also establishes a new state-of-the-art for fMRI-based natural image reconstruction in terms of high-level image features.

Image Super-Resolution using Efficient Striped Window Transformer

Recently, transformer-based methods have made impressive progress in single-image super-resolu-tion (SR). However, these methods are difficult to apply to lightweight SR (LSR) due to the challenge of balancing model performance and complexity. In this paper, we propose an efficient striped window transformer (ESWT). ESWT consists of efficient transformation layers (ETLs), allowing a clean structure and avoiding redundant operations. Moreover, we designed a striped window mechanism to obtain a more efficient ESWT in modeling long-term dependencies. To further exploit the potential of the transformer, we propose a novel flexible window training strategy. Without any additional cost, this strategy can further improve the performance of ESWT. Extensive experiments show that the proposed method outperforms state-of-the-art transformer-based LSR methods with fewer parameters, faster inference, smaller FLOPs, and less memory consumption, achieving a better trade-off between model performance and complexity.

Phase Transitions in Recovery of Structured Signals from Corrupted Measurements

This paper is concerned with the problem of recovering a structured signal from a relatively small number of corrupted random measurements. Sharp phase transitions have been numerically observed in practice when different convex programming procedures are used to solve this problem. This paper is devoted to presenting theoretical explanations for these phenomenons by employing some basic tools from Gaussian process theory. Specifically, we identify the precise locations of the phase transitions for both constrained and penalized recovery procedures. Our theoretical results show that these phase transitions are determined by some geometric measures of structure, e.g., the spherical Gaussian width of a tangent cone and the Gaussian (squared) distance to a scaled subdifferential. By utilizing the established phase transition theory, we further investigate the relationship between these two kinds of recovery procedures, which also reveals an optimal strategy (in the sense of Lagrange theory) for choosing the tradeoff parameter in the penalized recovery procedure. Numerical experiments are provided to verify our theoretical results.

The Herbarium Challenge 2019 Dataset

Herbarium sheets are invaluable for botanical research, and considerable time and effort is spent by experts to label and identify specimens on them. In view of recent advances in computer vision and deep learning, developing an automated approach to help experts identify specimens could significantly accelerate research in this area. Whereas most existing botanical datasets comprise photos of specimens in the wild, herbarium sheets exhibit dried specimens, which poses new challenges. We present a challenge dataset of herbarium sheet images labeled by experts, with the intent of facilitating the development of automated identification techniques for this challenging scenario.