16 papers
Benchmarking Composable Compression Techniques in Mixture-of-Experts LLMs
Afsara Benazir, Chen Chen, Rongxiao Qu +3
Mixture-of-Experts (MoE) LLMs scale model capacity efficiently through sparse activation, but their large expert parameter footprint, routing imbalance, and long-context KV-cache g…
Argus-Unified: Towards A Compact and Economical Unified Model for Image Understanding and Generation
Weiming Zhuang, Jiabo Huang, Jingtao Li +4
Unifying visual understanding and generation in one model holds immense promise, but remains challenging and expensive due to heavy compute and data demands and conflicts between t…
On the Limits of Token Reduction for Efficient Unified Vision Language Training
Siyi Chen, Weiming Zhuang, Jingtao Li +1
Unified vision-language models (VLMs) integrate visual understanding and visual generation within a single autoregressive backbone, but their joint training is computationally expe…
HyperVision: A Channel-Adaptive Ground-Based Hyperspectral Vision Pre-trained Backbone
Guanyiman Fu, Jingtao Li, Zihang Cheng +8
While hyperspectral imaging provides rich spatial-spectral information across hundreds of narrow wavelength bands for precise material identification, ground-based hyperspectral pr…
VibeToken: Scaling 1D Image Tokenizers and Autoregressive Models for Dynamic Resolution Generations
Maitreya Patel, Jingtao Li, Weiming Zhuang +2
We introduce an efficient, resolution-agnostic autoregressive (AR) image synthesis approach that generalizes to arbitrary resolutions and aspect ratios, narrowing the gap to diffus…
UniCompress: Token Compression for Unified Vision-Language Understanding and Generation
Ziyao Wang, Chen Chen, Jingtao Li +4
Unified models aim to support both understanding and generation by encoding images into discrete tokens and processing them alongside text within a single autoregressive framework.…
Unlocking Cognitive Capabilities and Analyzing the Perception-Logic Trade-off
Longyin Zhang, Shuo Sun, Yingxu He +15
Recent advancements in Multimodal Large Language Models (MLLMs) pursue omni-perception capabilities, yet integrating robust sensory grounding with complex reasoning remains a chall…
StelLA: Subspace Learning in Low-rank Adaptation using Stiefel Manifold
Zhizhong Li, Sina Sajadmanesh, Jingtao Li +1
Low-rank adaptation (LoRA) has been widely adopted as a parameter-efficient technique for fine-tuning large-scale pre-trained models. However, it still lags behind full fine-tuning…
FEDEXCHANGE: Bridging the Domain Gap in Federated Object Detection for Free
Haolin Yuan, Jingtao Li, Weiming Zhuang +2
Federated Object Detection (FOD) enables clients to collaboratively train a global object detection model without accessing their local data from diverse domains. However, signific…
UNIFORM: Unifying Knowledge from Large-scale and Diverse Pre-trained Models
Yimu Wang, Weiming Zhuang, Chen Chen +3
In the era of deep learning, the increasing number of pre-trained models available online presents a wealth of knowledge. These models, developed with diverse architectures and tra…
Closer to Reality: Practical Semi-Supervised Federated Learning for Foundation Model Adaptation
Guangyu Sun, Jingtao Li, Weiming Zhuang +2
Foundation models (FMs) exhibit remarkable generalization but require adaptation to downstream tasks, particularly in privacy-sensitive applications. Due to data privacy regulation…
Training-Free Layout-to-Image Generation with Marginal Attention Constraints
Huancheng Chen, Jingtao Li, Weiming Zhuang +2
Recently, many text-to-image diffusion models have excelled at generating high-resolution images from text but struggle with precise control over spatial composition and object cou…
Replay-Free Continual Low-Rank Adaptation with Dynamic Memory
Huancheng Chen, Jingtao Li, Weiming Zhuang +2
We revisit continual learning~(CL), which enables pre-trained vision transformers (ViTs) to sequentially fine-tune on new downstream tasks over time. However, as the scale of these…
COALA: A Practical and Vision-Centric Federated Learning Platform
Weiming Zhuang, Jian Xu, Chen Chen +2
We present COALA, a vision-centric Federated Learning (FL) platform, and a suite of benchmarks for practical FL scenarios, which we categorize into three levels: task, data, and mo…
Stretching Each Dollar: Diffusion Training from Scratch on a Micro-Budget
Vikash Sehwag, Xianghao Kong, Jingtao Li +2
As scaling laws in generative AI push performance, they also simultaneously concentrate the development of these models among actors with large computational resources. With a focu…
Is Synthetic Image Useful for Transfer Learning? An Investigation into Data Generation, Volume, and Utilization
Yuhang Li, Xin Dong, Chen Chen +4
Synthetic image data generation represents a promising avenue for training deep learning models, particularly in the realm of transfer learning, where obtaining real images within…