HCHong Chen

Hong Chen

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Hong Chen has a total of 20 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Top Cited Papers
    Open Access

    An empirical study of LLaMA3 quantization: from LLMs to MLLMs

    by Wei Huang, Xingyu Zheng, Xudong Ma, Haotong Qin, Chengtao Lv, Hong Chen, Jie Luo, Xiaojuan Qi, Xianglong Liu, Michele Magno

    The LLaMA family, a collection of foundation language models ranging from 7B to 65B parameters, has become one of the most powerful open-source large language models (LLMs) and the popular LLM backbone of multi-modal large language models (MLLMs), widely used in computer vision and natural language understanding tasks. In particular, LLaMA3 models have recently been released and have achieved impressive performance in various domains with super-large scale pre-training on over 15T tokens of data. Given the wide application of low-bit quantization for LLMs in resource-constrained scenarios, we explore LLaMA3’s capabilities when quantized to low bit-width. This exploration can potentially provide new insights and challenges for the low-bit quantization of LLaMA3 and other future LLMs, especially in addressing performance degradation issues that suffer in LLM compression. Specifically, we comprehensively evaluate the 10 existing post-training quantization and LoRA fine-tuning (LoRA-FT) methods of LLaMA3 on 1-8 bits and various datasets to reveal the low-bit quantization performance of LLaMA3. To uncover the capabilities of low-bit quantized MLLM, we assessed the performance of the LLaMA3-based LLaVA-Next-8B model under 2-4 ultra-low bits with post-training quantization methods. Our experimental results indicate that LLaMA3 still suffers from non-negligible degradation in linguistic and visual contexts, particularly under ultra-low bit widths. This highlights the significant performance gap at low bit-width that needs to be addressed in future developments. We expect that this empirical study will prove valuable in advancing future models, driving LLMs and MLLMs to achieve higher accuracy at lower bit to enhance practicality.


  • Open Access

    QVD: Post-training Quantization for Video Diffusion Models

    by Shilong Tian, Hong Chen, Chengtao Lv, Yu Liu, Jinyang Guo, Xianglong Liu, Shengxi Li, Hao Yang, Tao Xie

    Recently, video diffusion models (VDMs) have garnered significant attention due to their notable advancements in generating coherent and realistic video content. However, processing multiple frame features concurrently, coupled with the considerable model size, results in high latency and extensive memory consumption, hindering their broader application. Post-training quantization (PTQ) is an effective technique to reduce memory footprint and improve computational efficiency. Unlike image diffusion, we observe that the temporal features, which are integrated into all frame features, exhibit pronounced skewness. Furthermore, we investigate significant inter-channel disparities and asymmetries in the activation of video diffusion models, resulting in low coverage of quantization levels by individual channels and increasing the challenge of quantization. To address these issues, we introduce the first PTQ strategy tailored for video diffusion models, dubbed QVD. Specifically, we propose the High Temporal Discriminability Quantization (HTDQ) method, designed for temporal features, which retains the high discriminability of quantized features, providing precise temporal guidance for all video frames. In addition, we present the Scattered Channel Range Integration (SCRI) method which aims to improve the coverage of quantization levels across individual channels. Experimental validations across various models, datasets, and bit-width settings demonstrate the effectiveness of our QVD in terms of diverse metrics. In particular, we achieve near-lossless performance degradation on W8A8, outperforming the current methods by 205.12 in FVD.


  • Open Access

    Machine Learning Assessment of Damage Grade for Post-Earthquake Buildings: A Three-Stage Approach Directly Handling Categorical Features

    by Yutao Li, Chuanguo Jia, Hong Chen, Hongchen Su, Jiahao Chen, Duoduo Wang

    The rapid assessment of post-earthquake building damage for rescue and reconstruction is a crucial strategy to reduce the enormous number of human casualties and economic losses caused by earthquakes. Conventional machine learning (ML) approaches for this problem usually employ one-hot encoding to cope with categorical features, and their overall procedure is neither sufficient nor comprehensive. Therefore, this study proposed a three-stage approach, which can directly handle categorical features and enhance the entire methodology of ML applications. In stage I, an integrated data preprocessing framework involving subjective–objective feature selection was proposed and performed on a dataset of buildings after the 2015 Gorkha earthquake. In stage II, four machine learning models, KNN, XGBoost, CatBoost, and LightGBM, were trained and tested on the dataset. The best model was judged by comprehensive metrics, including the proposed risk coefficient. In stage III, the feature importance, the relationships between the features and the model’s output, and the feature interaction effects were investigated by Shapley additive explanations. The results indicate that the LightGBM model has the best overall performance with the highest accuracy of 0.897, the lowest risk coefficient of 0.042, and the shortest training time of 12.68 s due to its relevant algorithms for directly tackling categorical features. As for its interpretability, the most important features are determined, and information on these features’ impacts and interactions is obtained to improve the reliability of and promote practical engineering applications for the ML models. The proposed three-stage approach can provide a reference for the overall ML implementation process on raw datasets for similar problems.