FTII-Bench: A Comprehensive Multimodal Benchmark for Flow Text with Image Insertion

Benefiting from the revolutionary advances in large language models (LLMs) and foundational vision models, large vision-language models (LVLMs) have also made significant progress. However, current benchmarks focus on tasks that evaluating only a single aspect of LVLM capabilities (e.g., recognition, detection, understanding). These tasks fail to fully demonstrate LVLMs' potential in complex application scenarios. To comprehensively assess the performance of existing LVLMs, we propose a more challenging task called the Flow Text with Image Insertion task (FTII). This task requires LVLMs to simultaneously possess outstanding abilities in image comprehension, instruction understanding, and long-text interpretation. Specifically, given several text paragraphs and a set of candidate images, as the text paragraphs accumulate, the LVLMs are required to select the most suitable image from the candidates to insert after the corresponding paragraph. Constructing a benchmark for such a task is highly challenging, particularly in determining the sequence of flowing text and images. To address this challenge, we turn to professional news reports, which naturally contain a gold standard for image-text sequences. Based on this, we introduce the Flow Text with Image Insertion Benchmark (FTII-Bench), which includes 318 high-quality Chinese image-text news articles and 307 high-quality English image-text news articles, covering 10 different news domains. Using these 625 high-quality articles, we construct problems of two different types with multiple levels of difficulty. Furthermore, we establish two different evaluation pipelines based on the CLIP model and existing LVLMs. We evaluate 9 open-source and 2 closed-source LVLMs as well as 2 CLIP-based models. Results indicate that even the most advanced models (e.g., GPT-4o) face significant challenges when tackling the FTII task.

Understanding Robustness of Parameter-Efficient Tuning for Image Classification

Parameter-efficient tuning (PET) techniques calibrate the model's predictions on downstream tasks by freezing the pre-trained models and introducing a small number of learnable parameters. However, despite the numerous PET methods proposed, their robustness has not been thoroughly investigated. In this paper, we systematically explore the robustness of four classical PET techniques (e.g., VPT, Adapter, AdaptFormer, and LoRA) under both white-box attacks and information perturbations. For white-box attack scenarios, we first analyze the performance of PET techniques using FGSM and PGD attacks. Subsequently, we further explore the transferability of adversarial samples and the impact of learnable parameter quantities on the robustness of PET methods. Under information perturbation attacks, we introduce four distinct perturbation strategies, including Patch-wise Drop, Pixel-wise Drop, Patch Shuffle, and Gaussian Noise, to comprehensively assess the robustness of these PET techniques in the presence of information loss. Via these extensive studies, we enhance the understanding of the robustness of PET methods, providing valuable insights for improving their performance in computer vision applications. The code is available at https://github.com/JCruan519/PETRobustness.

MM-CamObj: A Comprehensive Multimodal Dataset for Camouflaged Object Scenarios

Large visual-language models (LVLMs) have achieved great success in multiple applications. However, they still encounter challenges in complex scenes, especially those involving camouflaged objects. This is primarily due to the lack of samples related to camouflaged scenes in the training dataset. To mitigate this issue, we construct the MM-CamObj dataset for the first time, comprising two subsets: CamObj-Align and CamObj-Instruct. Specifically, CamObj-Align contains 11,363 image-text pairs, and it is designed for VL alignment and injecting rich knowledge of camouflaged scenes into LVLMs. CamObj-Instruct is collected for fine-tuning the LVLMs with improved instruction-following capabilities, and it includes 11,363 images and 68,849 conversations with diverse instructions. Based on the MM-CamObj dataset, we propose the CamObj-Llava, an LVLM specifically designed for addressing tasks in camouflaged scenes. To facilitate our model's effective acquisition of knowledge about camouflaged objects and scenes, we introduce a curriculum learning strategy with six distinct modes. Additionally, we construct the CamObj-Bench to evaluate the existing LVLMs' capabilities of understanding, recognition, localization and count in camouflage scenes. This benchmark includes 600 images and 7 tasks, with a total of 9,449 questions. Extensive experiments are conducted on the CamObj-Bench with CamObj-Llava, 8 existing open-source and 3 closed-source LVLMs. Surprisingly, the results indicate that our model achieves a 25.84% improvement in 4 out of 7 tasks compared to GPT-4o. Code and datasets will be available at https://github.com/JCruan519/MM-CamObj.

LLaMA-MoE: Building Mixture-of-Experts from LLaMA with Continual Pre-training

Mixture-of-Experts (MoE) has gained increasing popularity as a promising framework for scaling up large language models (LLMs). However, training MoE from scratch in a large-scale setting still suffers from data-hungry and instability problems. Motivated by this limit, we investigate building MoE models from existing dense large language models. Specifically, based on the well-known LLaMA-2 7B model, we obtain an MoE model by: (1) Expert Construction, which partitions the parameters of original Feed-Forward Networks (FFNs) into multiple experts; (2) Continual Pre-training, which further trains the transformed MoE model and additional gate networks. In this paper, we comprehensively explore different methods for expert construction and various data sampling strategies for continual pre-training. After these stages, our LLaMA-MoE models could maintain language abilities and route the input tokens to specific experts with part of the parameters activated. Empirically, by training 200B tokens, LLaMA-MoE-3.5B models significantly outperform dense models that contain similar activation parameters. The source codes and models are available at https://github.com/pjlab-sys4nlp/llama-moe .

Dynamic Data Mixing Maximizes Instruction Tuning for Mixture-of-Experts

Mixture-of-Experts (MoE) models have shown remarkable capability in instruction tuning, especially when the number of tasks scales. However, previous methods simply merge all training tasks (e.g. creative writing, coding, and mathematics) and apply fixed sampling weights, without considering the importance of different tasks as the model training state changes. In this way, the most helpful data cannot be effectively distinguished, leading to suboptimal model performance. To reduce the potential redundancies of datasets, we make the first attempt and propose a novel dynamic data mixture for MoE instruction tuning. Specifically, inspired by MoE's token routing preference, we build dataset-level representations and then capture the subtle differences among datasets. Finally, we propose to dynamically adjust the sampling weight of datasets by their inter-redundancies, thus maximizing global performance under a limited training budget. The experimental results on two MoE models demonstrate the effectiveness of our approach on both downstream knowledge \& reasoning tasks and open-ended queries. Code and models are available at https://github.com/Spico197/MoE-SFT .

iDAT: inverse Distillation Adapter-Tuning

Adapter-Tuning (AT) method involves freezing a pre-trained model and introducing trainable adapter modules to acquire downstream knowledge, thereby calibrating the model for better adaptation to downstream tasks. This paper proposes a distillation framework for the AT method instead of crafting a carefully designed adapter module, which aims to improve fine-tuning performance. For the first time, we explore the possibility of combining the AT method with knowledge distillation. Via statistical analysis, we observe significant differences in the knowledge acquisition between adapter modules of different models. Leveraging these differences, we propose a simple yet effective framework called inverse Distillation Adapter-Tuning (iDAT). Specifically, we designate the smaller model as the teacher and the larger model as the student. The two are jointly trained, and online knowledge distillation is applied to inject knowledge of different perspective to student model, and significantly enhance the fine-tuning performance on downstream tasks. Extensive experiments on the VTAB-1K benchmark with 19 image classification tasks demonstrate the effectiveness of iDAT. The results show that using existing AT method within our iDAT framework can further yield a 2.66% performance gain, with only an additional 0.07M trainable parameters. Our approach compares favorably with state-of-the-arts without bells and whistles. Our code is available at https://github.com/JCruan519/iDAT.

VM-UNet: Vision Mamba UNet for Medical Image Segmentation

In the realm of medical image segmentation, both CNN-based and Transformer-based models have been extensively explored. However, CNNs exhibit limitations in long-range modeling capabilities, whereas Transformers are hampered by their quadratic computational complexity. Recently, State Space Models (SSMs), exemplified by Mamba, have emerged as a promising approach. They not only excel in modeling long-range interactions but also maintain a linear computational complexity. In this paper, leveraging state space models, we propose a U-shape architecture model for medical image segmentation, named Vision Mamba UNet (VM-UNet). Specifically, the Visual State Space (VSS) block is introduced as the foundation block to capture extensive contextual information, and an asymmetrical encoder-decoder structure is constructed. We conduct comprehensive experiments on the ISIC17, ISIC18, and Synapse datasets, and the results indicate that VM-UNet performs competitively in medical image segmentation tasks. To our best knowledge, this is the first medical image segmentation model constructed based on the pure SSM-based model. We aim to establish a baseline and provide valuable insights for the future development of more efficient and effective SSM-based segmentation systems. Our code is available at https://github.com/JCruan519/VM-UNet.

CLAPP: Contrastive Language-Audio Pre-training in Passive Underwater Vessel Classification

Existing research on audio classification faces challenges in recognizing attributes of passive underwater vessel scenarios and lacks well-annotated datasets due to data privacy concerns. In this study, we introduce CLAPP (Contrastive Language-Audio Pre-training in Passive Underwater Vessel Classification), a novel model. Our aim is to train a neural network using a wide range of vessel audio and vessel state text pairs obtained from an oceanship dataset. CLAPP is capable of directly learning from raw vessel audio data and, when available, from carefully curated labels, enabling improved recognition of vessel attributes in passive underwater vessel scenarios. Model's zero-shot capability allows predicting the most relevant vessel state description for a given vessel audio, without directly optimizing for the task. Our approach aims to solve 2 challenges: vessel audio-text classification and passive underwater vessel audio attribute recognition. The proposed method achieves new state-of-the-art results on both Deepship and Shipsear public datasets, with a notable margin of about 7%-13% for accuracy compared to prior methods on zero-shot task.

Learning Multi-axis Representation in Frequency Domain for Medical Image Segmentation

Recently, Visual Transformer (ViT) has been extensively used in medical image segmentation (MIS) due to applying self-attention mechanism in the spatial domain to modeling global knowledge. However, many studies have focused on improving models in the spatial domain while neglecting the importance of frequency domain information. Therefore, we propose Multi-axis External Weights UNet (MEW-UNet) based on the U-shape architecture by replacing self-attention in ViT with our Multi-axis External Weights block. Specifically, our block performs a Fourier transform on the three axes of the input features and assigns the external weight in the frequency domain, which is generated by our External Weights Generator. Then, an inverse Fourier transform is performed to change the features back to the spatial domain. We evaluate our model on four datasets, including Synapse, ACDC, ISIC17 and ISIC18 datasets, and our approach demonstrates competitive performance, owing to its effective utilization of frequency domain information.

LAMM: Label Alignment for Multi-Modal Prompt Learning

With the success of pre-trained visual-language (VL) models such as CLIP in visual representation tasks, transferring pre-trained models to downstream tasks has become a crucial paradigm. Recently, the prompt tuning paradigm, which draws inspiration from natural language processing (NLP), has made significant progress in VL field. However, preceding methods mainly focus on constructing prompt templates for text and visual inputs, neglecting the gap in class label representations between the VL models and downstream tasks. To address this challenge, we introduce an innovative label alignment method named \textbf{LAMM}, which can dynamically adjust the category embeddings of downstream datasets through end-to-end training. Moreover, to achieve a more appropriate label distribution, we propose a hierarchical loss, encompassing the alignment of the parameter space, feature space, and logits space. We conduct experiments on 11 downstream vision datasets and demonstrate that our method significantly improves the performance of existing multi-modal prompt learning models in few-shot scenarios, exhibiting an average accuracy improvement of 2.31(\%) compared to the state-of-the-art methods on 16 shots. Moreover, our methodology exhibits the preeminence in continual learning compared to other prompt tuning methods. Importantly, our method is synergistic with existing prompt tuning methods and can boost the performance on top of them. Our code and dataset will be publicly available at https://github.com/gaojingsheng/LAMM.