Difficult Task Yes but Simple Task No: Unveiling the Laziness in Multimodal LLMs

Multimodal Large Language Models (MLLMs) demonstrate a strong understanding of the real world and can even handle complex tasks. However, they still fail on some straightforward visual question-answering (VQA) problems. This paper dives deeper into this issue, revealing that models tend to err when answering easy questions (e.g. Yes/No questions) about an image, even though they can correctly describe it. We refer to this model behavior discrepancy between difficult and simple questions as model laziness. To systematically investigate model laziness, we manually construct LazyBench, a benchmark that includes Yes/No, multiple choice, short answer questions, and image description tasks that are related to the same subjects in the images. Based on LazyBench, we observe that laziness widely exists in current advanced MLLMs (e.g. GPT-4o, Gemini-1.5-pro, Claude 3 and LLaVA-v1.5-13B), and it is more pronounced on stronger models. We also analyze the VQA v2 (LLaVA-v1.5-13B) benchmark and find that about half of its failure cases are caused by model laziness, which further highlights the importance of ensuring that the model fully utilizes its capability. To this end, we conduct preliminary exploration on how to mitigate laziness and find that chain of thought (CoT) can effectively address this issue.

TRACE: Temporal Grounding Video LLM via Causal Event Modeling

Video Temporal Grounding (VTG) is a crucial capability for video understanding models and plays a vital role in downstream tasks such as video browsing and editing. To effectively handle various tasks simultaneously and enable zero-shot prediction, there is a growing trend in employing video LLMs for VTG tasks. However, current video LLM-based methods rely exclusively on natural language generation, lacking the ability to model the clear structure inherent in videos, which restricts their effectiveness in tackling VTG tasks. To address this issue, this paper first formally introduces causal event modeling framework, which represents videos as sequences of events, and predict the current event using previous events, video inputs, and textural instructions. Each event consists of three components: timestamps, salient scores, and textual captions. We then propose a novel task-interleaved video LLM called TRACE to effectively implement the causal event modeling framework in practice. The TRACE processes visual frames, timestamps, salient scores, and text as distinct tasks, employing various encoders and decoding heads for each. Task tokens are arranged in an interleaved sequence according to the causal event modeling framework's formulation. Extensive experiments on various VTG tasks and datasets demonstrate the superior performance of TRACE compared to state-of-the-art video LLMs. Our model and code are available at \url{https://github.com/gyxxyg/TRACE}.

FPT+: A Parameter and Memory Efficient Transfer Learning Method for High-resolution Medical Image Classification

The success of large-scale pre-trained models has established fine-tuning as a standard method for achieving significant improvements in downstream tasks. However, fine-tuning the entire parameter set of a pre-trained model is costly. Parameter-efficient transfer learning (PETL) has recently emerged as a cost-effective alternative for adapting pre-trained models to downstream tasks. Despite its advantages, the increasing model size and input resolution present challenges for PETL, as the training memory consumption is not reduced as effectively as the parameter usage. In this paper, we introduce Fine-grained Prompt Tuning plus (FPT+), a PETL method designed for high-resolution medical image classification, which significantly reduces memory consumption compared to other PETL methods. FPT+ performs transfer learning by training a lightweight side network and accessing pre-trained knowledge from a large pre-trained model (LPM) through fine-grained prompts and fusion modules. Specifically, we freeze the LPM and construct a learnable lightweight side network. The frozen LPM processes high-resolution images to extract fine-grained features, while the side network employs the corresponding down-sampled low-resolution images to minimize the memory usage. To enable the side network to leverage pre-trained knowledge, we propose fine-grained prompts and fusion modules, which collaborate to summarize information through the LPM's intermediate activations. We evaluate FPT+ on eight medical image datasets of varying sizes, modalities, and complexities. Experimental results demonstrate that FPT+ outperforms other PETL methods, using only 1.03% of the learnable parameters and 3.18% of the memory required for fine-tuning an entire ViT-B model. Our code is available at https://github.com/YijinHuang/FPT.

Smart Sampling: Helping from Friendly Neighbors for Decentralized Federated Learning

Federated Learning (FL) is gaining widespread interest for its ability to share knowledge while preserving privacy and reducing communication costs. Unlike Centralized FL, Decentralized FL (DFL) employs a network architecture that eliminates the need for a central server, allowing direct communication among clients and leading to significant communication resource savings. However, due to data heterogeneity, not all neighboring nodes contribute to enhancing the local client's model performance. In this work, we introduce \textbf{\emph{AFIND+}}, a simple yet efficient algorithm for sampling and aggregating neighbors in DFL, with the aim of leveraging collaboration to improve clients' model performance. AFIND+ identifies helpful neighbors, adaptively adjusts the number of selected neighbors, and strategically aggregates the sampled neighbors' models based on their contributions. Numerical results on real-world datasets with diverse data partitions demonstrate that AFIND+ outperforms other sampling algorithms in DFL and is compatible with most existing DFL optimization algorithms.

Save It All: Enabling Full Parameter Tuning for Federated Large Language Models via Cycle Black Gradient Descent

The advent of large language models (LLMs) has revolutionized the deep learning paradigm, yielding impressive results across a wide array of tasks. However, the pre-training or fine-tuning of LLMs within a federated learning (FL) framework poses substantial challenges, including considerable computational and memory resource demands, as well as communication bottlenecks between servers and clients. Existing solutions either make the unrealistic assumption that the entire model is exchanged for training, or apply parameter-effective fine-tuning methods from centralized learning to train LLMs in FL which tend to underperform during training or fine-tuning stages due to the limited search subspace of parameter updating. In this paper, we introduce a novel method for the efficient training and fine-tuning of LLMs in FL, with minimal resource consumption. Our approach, termed FedCyBGD, utilizes Cycle Block Gradient Descent to periodically update the model. In particular, we design a compression scheme for FedCyBGD, aiming to further decrease the model download cost. It enables full parameter training in FL with only selected block updates and uploads, thereby reducing communication, computation, and memory costs. Our method achieves state-of-the-art performance for FL LLM training, while significantly reducing associated costs. Codes are provided here.

Saliency-guided and Patch-based Mixup for Long-tailed Skin Cancer Image Classification

Medical image datasets often exhibit long-tailed distributions due to the inherent challenges in medical data collection and annotation. In long-tailed contexts, some common disease categories account for most of the data, while only a few samples are available in the rare disease categories, resulting in poor performance of deep learning methods. To address this issue, previous approaches have employed class re-sampling or re-weighting techniques, which often encounter challenges such as overfitting to tail classes or difficulties in optimization during training. In this work, we propose a novel approach, namely \textbf{S}aliency-guided and \textbf{P}atch-based \textbf{Mix}up (SPMix) for long-tailed skin cancer image classification. Specifically, given a tail-class image and a head-class image, we generate a new tail-class image by mixing them under the guidance of saliency mapping, which allows for preserving and augmenting the discriminative features of the tail classes without any interference of the head-class features. Extensive experiments are conducted on the ISIC2018 dataset, demonstrating the superiority of SPMix over existing state-of-the-art methods.

Client2Vec: Improving Federated Learning by Distribution Shifts Aware Client Indexing

Federated Learning (FL) is a privacy-preserving distributed machine learning paradigm. Nonetheless, the substantial distribution shifts among clients pose a considerable challenge to the performance of current FL algorithms. To mitigate this challenge, various methods have been proposed to enhance the FL training process. This paper endeavors to tackle the issue of data heterogeneity from another perspective -- by improving FL algorithms prior to the actual training stage. Specifically, we introduce the Client2Vec mechanism, which generates a unique client index for each client before the commencement of FL training. Subsequently, we leverage the generated client index to enhance the subsequent FL training process. To demonstrate the effectiveness of the proposed Client2Vec method, we conduct three case studies that assess the impact of the client index on the FL training process. These case studies encompass enhanced client sampling, model aggregation, and local training. Extensive experiments conducted on diverse datasets and model architectures show the efficacy of Client2Vec across all three case studies. Our code is avaliable at \url{https://github.com/LINs-lab/client2vec}.

Dynamic Mixture of Experts: An Auto-Tuning Approach for Efficient Transformer Models

The Sparse Mixture of Experts (SMoE) has been widely employed to enhance the efficiency of training and inference for Transformer-based foundational models, yielding promising results. However, the performance of SMoE heavily depends on the choice of hyper-parameters, such as the number of experts and the number of experts to be activated (referred to as top-k), resulting in significant computational overhead due to the extensive model training by searching over various hyper-parameter configurations. As a remedy, we introduce the Dynamic Mixture of Experts (DynMoE) technique. DynMoE incorporates (1) a novel gating method that enables each token to automatically determine the number of experts to activate. (2) An adaptive process automatically adjusts the number of experts during training. Extensive numerical results across Vision, Language, and Vision-Language tasks demonstrate the effectiveness of our approach to achieve competitive performance compared to GMoE for vision and language tasks, and MoE-LLaVA for vision-language tasks, while maintaining efficiency by activating fewer parameters. Our code is available at https://github.com/LINs-lab/DynMoE.

VTG-LLM: Integrating Timestamp Knowledge into Video LLMs for Enhanced Video Temporal Grounding

Video Temporal Grounding (VTG) focuses on accurately identifying event timestamps within a particular video based on a linguistic query, playing a vital role in downstream tasks such as video browsing and editing. While Video Large Language Models (video LLMs) have made significant progress in understanding video content, they often face challenges in accurately pinpointing timestamps within videos, which limits their performance on VTG tasks. Therefore, to improve video LLMs' ability to effectively locate timestamps, we argue that two critical aspects need to be enhanced. First, it is essential to have high-quality instructional tuning datasets that encompass mainstream VTG tasks. Second, directly incorporating timestamp knowledge into video LLMs is crucial, as it enables models to efficiently comprehend timestamp information. To address these needs, we first introduce VTG-IT-120K, a high-quality and comprehensive instruction tuning dataset that covers VTG tasks such as moment retrieval, dense video captioning, video summarization, and video highlight detection. Furthermore, we propose a specially designed video LLM model for VTG tasks, VTG-LLM, which (1) effectively integrates timestamp knowledge into visual tokens; (2) incorporates absolute-time tokens that specifically handle timestamp knowledge, thereby avoiding concept shifts; and (3) introduces a lightweight, high-performance slot-based token compression method to facilitate the sampling of more video frames. Comprehensive experiments showcase the superior performance of VTG-LLM in comparison to other video LLM methods across various VTG tasks. Our code and datasets are available at \url{https://github.com/gyxxyg/VTG-LLM}.

FPT: Fine-grained Prompt Tuning for Parameter and Memory Efficient Fine Tuning in High-resolution Medical Image Classification

Parameter-efficient fine-tuning (PEFT) is proposed as a cost-effective way to transfer pre-trained models to downstream tasks, avoiding the high cost of updating entire large-scale pre-trained models (LPMs). In this work, we present Fine-grained Prompt Tuning (FPT), a novel PEFT method for medical image classification. FPT significantly reduces memory consumption compared to other PEFT methods, especially in high-resolution contexts. To achieve this, we first freeze the weights of the LPM and construct a learnable lightweight side network. The frozen LPM takes high-resolution images as input to extract fine-grained features, while the side network is fed low-resolution images to reduce memory usage. To allow the side network to access pre-trained knowledge, we introduce fine-grained prompts that summarize information from the LPM through a fusion module. Important tokens selection and preloading techniques are employed to further reduce training cost and memory requirements. We evaluate FPT on four medical datasets with varying sizes, modalities, and complexities. Experimental results demonstrate that FPT achieves comparable performance to fine-tuning the entire LPM while using only 1.8% of the learnable parameters and 13% of the memory costs of an encoder ViT-B model with a 512 x 512 input resolution.