Zangwei Zheng

Zangwei Zheng, Mingyuan Ma, Kai Wang et al. · berkeley

Continual learning (CL) can help pre-trained vision-language models efficiently adapt to new or under-trained data distributions without re-training. Nevertheless, during the continual training of the Contrastive Language-Image Pre-training (CLIP) model, we observe that the model's zero-shot transfer ability significantly degrades due to catastrophic forgetting. Existing CL methods can mitigate forgetting by replaying previous data. However, since the CLIP dataset is private, replay methods cannot access the pre-training dataset. In addition, replaying data of previously learned downstream tasks can enhance their performance but comes at the cost of sacrificing zero-shot performance. To address this challenge, we propose a novel method ZSCL to prevent zero-shot transfer degradation in the continual learning of vision-language models in both feature and parameter space. In the feature space, a reference dataset is introduced for distillation between the current and initial models. The reference dataset should have semantic diversity but no need to be labeled, seen in pre-training, or matched image-text pairs. In parameter space, we prevent a large parameter shift by averaging weights during the training. We propose a more challenging Multi-domain Task Incremental Learning (MTIL) benchmark to evaluate different methods, where tasks are from various domains instead of class-separated in a single dataset. Our method outperforms other methods in the traditional class-incremental learning setting and the MTIL by 9.7% average score. Our code locates at https://github.com/Thunderbeee/ZSCL.

Ziheng Qin, Kai Wang, Zangwei Zheng et al.

Data pruning aims to obtain lossless performances with less overall cost. A common approach is to filter out samples that make less contribution to the training. This could lead to gradient expectation bias compared to the original data. To solve this problem, we propose \textbf{InfoBatch}, a novel framework aiming to achieve lossless training acceleration by unbiased dynamic data pruning. Specifically, InfoBatch randomly prunes a portion of less informative samples based on the loss distribution and rescales the gradients of the remaining samples to approximate the original gradient. As a plug-and-play and architecture-agnostic framework, InfoBatch consistently obtains lossless training results on classification, semantic segmentation, vision pertaining, and instruction fine-tuning tasks. On CIFAR10/100, ImageNet-1K, and ADE20K, InfoBatch losslessly saves 40\% overall cost. For pertaining MAE and diffusion model, InfoBatch can respectively save 24.8\% and 27\% cost. For LLaMA instruction fine-tuning, InfoBatch is also able to save 20\% cost and is compatible with coreset selection methods. The code is publicly available at \href{https://github.com/henryqin1997/InfoBatch}{github.com/NUS-HPC-AI-Lab/InfoBatch}.

Zangwei Zheng, Xiangyu Yue, Kai Wang et al. · berkeley

Though vision transformers (ViTs) have exhibited impressive ability for representation learning, we empirically find that they cannot generalize well to unseen domains with previous domain generalization algorithms. In this paper, we propose a novel approach DoPrompt based on prompt learning to embed the knowledge of source domains in domain prompts for target domain prediction. Specifically, domain prompts are prepended before ViT input tokens from the corresponding source domain. Each domain prompt learns domain-specific knowledge efficiently since it is optimized only for one domain. Meanwhile, we train a prompt adapter to produce a suitable prompt for each input image based on the learned source domain prompts. At test time, the adapted prompt generated by the prompt adapter can exploit the similarity between the feature of the out-of-domain image and source domains to properly integrate the source domain knowledge. Extensive experiments are conducted on four benchmark datasets. Our approach achieves 1.4% improvements in the averaged accuracy, which is 3.5 times the improvement of the state-of-the-art algorithm with a ViT backbone.

Zangwei Zheng, Pengtai Xu, Xuan Zou et al.

The click-through rate (CTR) prediction task is to predict whether a user will click on the recommended item. As mind-boggling amounts of data are produced online daily, accelerating CTR prediction model training is critical to ensuring an up-to-date model and reducing the training cost. One approach to increase the training speed is to apply large batch training. However, as shown in computer vision and natural language processing tasks, training with a large batch easily suffers from the loss of accuracy. Our experiments show that previous scaling rules fail in the training of CTR prediction neural networks. To tackle this problem, we first theoretically show that different frequencies of ids make it challenging to scale hyperparameters when scaling the batch size. To stabilize the training process in a large batch size setting, we develop the adaptive Column-wise Clipping (CowClip). It enables an easy and effective scaling rule for the embeddings, which keeps the learning rate unchanged and scales the L2 loss. We conduct extensive experiments with four CTR prediction networks on two real-world datasets and successfully scaled 128 times the original batch size without accuracy loss. In particular, for CTR prediction model DeepFM training on the Criteo dataset, our optimization framework enlarges the batch size from 1K to 128K with over 0.1% AUC improvement and reduces training time from 12 hours to 10 minutes on a single V100 GPU. Our code locates at https://github.com/bytedance/LargeBatchCTR.

Fuzhao Xue, Jianghai Chen, Aixin Sun et al.

Transformer-based models have delivered impressive results on many tasks, particularly vision and language tasks. In many model training situations, conventional configurations are typically adopted. For example, we often set the base model with hidden dimensions (i.e. model width) to be 768 and the number of transformer layers (i.e. model depth) to be 12. In this paper, we revisit these conventional configurations. Through theoretical analysis and experimental evaluation, we show that the masked autoencoder is effective in alleviating the over-smoothing issue in deep transformer training. Based on this finding, we propose Bamboo, an idea of using deeper and narrower transformer configurations, for masked autoencoder training. On ImageNet, with such a simple change in configuration, re-designed model achieves 87.1% top-1 accuracy and outperforms SoTA models like MAE and BEiT. On language tasks, re-designed model outperforms BERT with default setting by 1.1 points on average, on GLUE datasets.

Yang Luo, Xiaozhe Ren, Zangwei Zheng et al.

Adaptive gradient methods, such as Adam and LAMB, have demonstrated excellent performance in the training of large language models. Nevertheless, the need for adaptivity requires maintaining second-moment estimates of the per-parameter gradients, which entails a high cost of extra memory overheads. To solve this problem, several memory-efficient optimizers (e.g., Adafactor) have been proposed to obtain a drastic reduction in auxiliary memory usage, but with a performance penalty. In this paper, we first study a confidence-guided strategy to reduce the instability of existing memory efficient optimizers. Based on this strategy, we propose CAME to simultaneously achieve two goals: fast convergence as in traditional adaptive methods, and low memory usage as in memory-efficient methods. Extensive experiments demonstrate the training stability and superior performance of CAME across various NLP tasks such as BERT and GPT-2 training. Notably, for BERT pre-training on the large batch size of 32,768, our proposed optimizer attains faster convergence and higher accuracy compared with the Adam optimizer. The implementation of CAME is publicly available.

Zangwei Zheng, Xiangyu Peng, Tianji Yang et al.

Vision and language are the two foundational senses for humans, and they build up our cognitive ability and intelligence. While significant breakthroughs have been made in AI language ability, artificial visual intelligence, especially the ability to generate and simulate the world we see, is far lagging behind. To facilitate the development and accessibility of artificial visual intelligence, we created Open-Sora, an open-source video generation model designed to produce high-fidelity video content. Open-Sora supports a wide spectrum of visual generation tasks, including text-to-image generation, text-to-video generation, and image-to-video generation. The model leverages advanced deep learning architectures and training/inference techniques to enable flexible video synthesis, which could generate video content of up to 15 seconds, up to 720p resolution, and arbitrary aspect ratios. Specifically, we introduce Spatial-Temporal Diffusion Transformer (STDiT), an efficient diffusion framework for videos that decouples spatial and temporal attention. We also introduce a highly compressive 3D autoencoder to make representations compact and further accelerate training with an ad hoc training strategy. Through this initiative, we aim to foster innovation, creativity, and inclusivity within the community of AI content creation. By embracing the open-source principle, Open-Sora democratizes full access to all the training/inference/data preparation codes as well as model weights. All resources are publicly available at: https://github.com/hpcaitech/Open-Sora.

Fuzhao Xue, Zian Zheng, Yao Fu et al.

To help the open-source community have a better understanding of Mixture-of-Experts (MoE) based large language models (LLMs), we train and release OpenMoE, a series of fully open-sourced and reproducible decoder-only MoE LLMs, ranging from 650M to 34B parameters and trained on up to over 1T tokens. Our investigation confirms that MoE-based LLMs can offer a more favorable cost-effectiveness trade-off than dense LLMs, highlighting the potential effectiveness for future LLM development. One more important contribution of this study is an in-depth analysis of the routing mechanisms within our OpenMoE models, leading to three significant findings: Context-Independent Specialization, Early Routing Learning, and Drop-towards-the-End. We discovered that routing decisions in MoE models are predominantly based on token IDs, with minimal context relevance. The token-to-expert assignments are determined early in the pre-training phase and remain largely unchanged. This imperfect routing can result in performance degradation, particularly in sequential tasks like multi-turn conversations, where tokens appearing later in a sequence are more likely to be dropped. Finally, we rethink our design based on the above-mentioned observations and analysis. To facilitate future MoE LLM development, we propose potential strategies for mitigating the issues we found and further improving off-the-shelf MoE LLM designs.

Xiangyu Peng, Zangwei Zheng, Chenhui Shen et al.

Video generation models have achieved remarkable progress in the past year. The quality of AI video continues to improve, but at the cost of larger model size, increased data quantity, and greater demand for training compute. In this report, we present Open-Sora 2.0, a commercial-level video generation model trained for only $200k. With this model, we demonstrate that the cost of training a top-performing video generation model is highly controllable. We detail all techniques that contribute to this efficiency breakthrough, including data curation, model architecture, training strategy, and system optimization. According to human evaluation results and VBench scores, Open-Sora 2.0 is comparable to global leading video generation models including the open-source HunyuanVideo and the closed-source Runway Gen-3 Alpha. By making Open-Sora 2.0 fully open-source, we aim to democratize access to advanced video generation technology, fostering broader innovation and creativity in content creation. All resources are publicly available at: https://github.com/hpcaitech/Open-Sora.

Yang Luo, Zangwei Zheng, Ziheng Qin et al.

Large-batch training has become a cornerstone in accelerating the training of deep neural networks, yet it poses challenges in optimization and generalization. Existing optimizers like AdamW present performance degradation during language models' large-batch training, due to the information bottleneck in attention layers caused by the sharp increase of max attention logit. While the LAMB optimizer partially addresses this issue, some attention layers still face this issue. The reason is that $l_2$-norm-based trust ratios in LAMB are less effective in directly influencing the max value of query/key weights. Furthermore, the weight-wise trust ratio in LAMB is error-prone as it overlooks relationships of weight values within rows or columns. Building on these observations, we propose a novel optimizer, MERIT, which leverages the max-norm to calculate the trust ratio to constrain the max attention logit more effectively. Moreover, we further construct element-wise trust ratios to provide more robust update scaling by focusing on local weight structures. Extensive experiments of large-batch training across various sizes of GPT-2 models demonstrate the superior performance of MERIT. Notably, during the training of GPT-2 Medium, MERIT enables a 6k batch size without any performance degradation compared to the standard batch size (480) with 48B training tokens. This work highlights the importance of considering the max attention logit and finer-granularity trust ratio in large-batch training. It successfully improves the training stability and paves the way for larger batch usage, enabling faster development and iteration of large language models. Code is available at https://github.com/NUS-HPC-AI-Lab/MERIT.

Zirui Zhu, Yong Liu, Zangwei Zheng et al.

Click-Through Rate (CTR) prediction holds paramount significance in online advertising and recommendation scenarios. Despite the proliferation of recent CTR prediction models, the improvements in performance have remained limited, as evidenced by open-source benchmark assessments. Current researchers tend to focus on developing new models for various datasets and settings, often neglecting a crucial question: What is the key challenge that truly makes CTR prediction so demanding? In this paper, we approach the problem of CTR prediction from an optimization perspective. We explore the typical data characteristics and optimization statistics of CTR prediction, revealing a strong positive correlation between the top hessian eigenvalue and feature frequency. This correlation implies that frequently occurring features tend to converge towards sharp local minima, ultimately leading to suboptimal performance. Motivated by the recent advancements in sharpness-aware minimization (SAM), which considers the geometric aspects of the loss landscape during optimization, we present a dedicated optimizer crafted for CTR prediction, named Helen. Helen incorporates frequency-wise Hessian eigenvalue regularization, achieved through adaptive perturbations based on normalized feature frequencies. Empirical results under the open-source benchmark framework underscore Helen's effectiveness. It successfully constrains the top eigenvalue of the Hessian matrix and demonstrates a clear advantage over widely used optimization algorithms when applied to seven popular models across three public benchmark datasets on BARS. Our code locates at github.com/NUS-HPC-AI-Lab/Helen.

Xuanlei Zhao, Shenggan Cheng, Chang Chen et al.

Scaling multi-dimensional transformers to long sequences is indispensable across various domains. However, the challenges of large memory requirements and slow speeds of such sequences necessitate sequence parallelism. All existing approaches fall under the category of embedded sequence parallelism, which are limited to shard along a single sequence dimension, thereby introducing significant communication overhead. However, the nature of multi-dimensional transformers involves independent calculations across multiple sequence dimensions. To this end, we propose Dynamic Sequence Parallelism (DSP) as a novel abstraction of sequence parallelism. DSP dynamically switches the parallel dimension among all sequences according to the computation stage with efficient resharding strategy. DSP offers significant reductions in communication costs, adaptability across modules, and ease of implementation with minimal constraints. Experimental evaluations demonstrate DSP's superiority over state-of-the-art embedded sequence parallelism methods by remarkable throughput improvements ranging from 32.2% to 10x, with less than 25% communication volume.

Yang Luo, Zangwei Zheng, Zirui Zhu et al.

The increase in parameter size of multimodal large language models (MLLMs) introduces significant capabilities, particularly in-context learning, where MLLMs enhance task performance without updating pre-trained parameters. This effectiveness, however, hinges on the appropriate selection of in-context examples, a process that is currently biased towards visual data, overlooking textual information. Furthermore, the area of supervised retrievers for MLLMs, crucial for optimal in-context example selection, continues to be uninvestigated. Our study offers an in-depth evaluation of the impact of textual information on the unsupervised selection of in-context examples in multimodal contexts, uncovering a notable sensitivity of retriever performance to the employed modalities. Responding to this, we introduce a novel supervised MLLM-retriever MSIER that employs a neural network to select examples that enhance multimodal in-context learning efficiency. This approach is validated through extensive testing across three distinct tasks, demonstrating the method's effectiveness. Additionally, we investigate the influence of modalities on our supervised retrieval method's training and pinpoint factors contributing to our model's success. This exploration paves the way for future advancements, highlighting the potential for refined in-context learning in MLLMs through the strategic use of multimodal data.

Fuzhao Xue, Yao Fu, Wangchunshu Zhou et al.

Recent research has highlighted the importance of dataset size in scaling language models. However, large language models (LLMs) are notoriously token-hungry during pre-training, and high-quality text data on the web is approaching its scaling limit for LLMs. To further enhance LLMs, a straightforward approach is to repeat the pre-training data for additional epochs. In this study, we empirically investigate three key aspects under this approach. First, we explore the consequences of repeating pre-training data, revealing that the model is susceptible to overfitting, leading to multi-epoch degradation. Second, we examine the key factors contributing to multi-epoch degradation, finding that significant factors include dataset size, model parameters, and training objectives, while less influential factors consist of dataset quality and model FLOPs. Finally, we explore whether widely used regularization can alleviate multi-epoch degradation. Most regularization techniques do not yield significant improvements, except for dropout, which demonstrates remarkable effectiveness but requires careful tuning when scaling up the model size. Additionally, we discover that leveraging mixture-of-experts (MoE) enables cost-effective and efficient hyper-parameter tuning for computationally intensive dense LLMs with comparable trainable parameters, potentially impacting efficient LLM development on a broader scale.

Zangwei Zheng, Xiaozhe Ren, Fuzhao Xue et al.

Large language models (LLMs) have revolutionized the field of AI, demonstrating unprecedented capacity across various tasks. However, the inference process for LLMs comes with significant computational costs. In this paper, we propose an efficient LLM inference pipeline that harnesses the power of LLMs. Our approach begins by tapping into the potential of LLMs to accurately perceive and predict the response length with minimal overhead. By leveraging this information, we introduce an efficient sequence scheduling technique that groups queries with similar response lengths into micro-batches. We evaluate our approach on real-world instruction datasets using the LLaMA-based model, and our results demonstrate an impressive 86% improvement in inference throughput without compromising effectiveness. Notably, our method is orthogonal to other inference acceleration techniques, making it a valuable addition to many existing toolkits (e.g., FlashAttention, Quantization) for LLM inference.

Xiangyu Yue, Zangwei Zheng, Colorado Reed et al.

Multi-source Domain Adaptation (MDA) aims to transfer predictive models from multiple, fully-labeled source domains to an unlabeled target domain. However, in many applications, relevant labeled source datasets may not be available, and collecting source labels can be as expensive as labeling the target data itself. In this paper, we investigate Multi-source Few-shot Domain Adaptation (MFDA): a new domain adaptation scenario with limited multi-source labels and unlabeled target data. As we show, existing methods often fail to learn discriminative features for both source and target domains in the MFDA setting. Therefore, we propose a novel framework, termed Multi-Source Few-shot Adaptation Network (MSFAN), which can be trained end-to-end in a non-adversarial manner. MSFAN operates by first using a type of prototypical, multi-domain, self-supervised learning to learn features that are not only domain-invariant but also class-discriminative. Second, MSFAN uses a small, labeled support set to enforce feature consistency and domain invariance across domains. Finally, prototypes from multiple sources are leveraged to learn better classifiers. Compared with state-of-the-art MDA methods, MSFAN improves the mean classification accuracy over different domain pairs on MFDA by 20.2%, 9.4%, and 16.2% on Office, Office-Home, and DomainNet, respectively.

Yuxuan Lou, Fuzhao Xue, Zangwei Zheng et al.

Mixture-of-Experts (MoE), a conditional computation architecture, achieved promising performance by scaling local module (i.e. feed-forward network) of transformer. However, scaling the cross-token module (i.e. self-attention) is challenging due to the unstable training. This work proposes Sparse-MLP, an all-MLP model which applies sparsely-activated MLPs to cross-token modeling. Specifically, in each Sparse block of our all-MLP model, we apply two stages of MoE layers: one with MLP experts mixing information within channels along image patch dimension, the other with MLP experts mixing information within patches along the channel dimension. In addition, by proposing importance-score routing strategy for MoE and redesigning the image representation shape, we further improve our model's computational efficiency. Experimentally, we are more computation-efficient than Vision Transformers with comparable accuracy. Also, our models can outperform MLP-Mixer by 2.5\% on ImageNet Top-1 accuracy with fewer parameters and computational cost. On downstream tasks, i.e. Cifar10 and Cifar100, our models can still achieve better performance than baselines.

Zangwei Zheng, Xiangyu Yue, Kurt Keutzer et al.

Object detection is essential to safe autonomous or assisted driving. Previous works usually utilize RGB images or LiDAR point clouds to identify and localize multiple objects in self-driving. However, cameras tend to fail in bad driving conditions, e.g. bad weather or weak lighting, while LiDAR scanners are too expensive to get widely deployed in commercial applications. Radar has been drawing more and more attention due to its robustness and low cost. In this paper, we propose a scene-aware radar learning framework for accurate and robust object detection. First, the learning framework contains branches conditioning on the scene category of the radar sequence; with each branch optimized for a specific type of scene. Second, three different 3D autoencoder-based architectures are proposed for radar object detection and ensemble learning is performed over the different architectures to further boost the final performance. Third, we propose novel scene-aware sequence mix augmentation (SceneMix) and scene-specific post-processing to generate more robust detection results. In the ROD2021 Challenge, we achieved a final result of average precision of 75.0% and an average recall of 81.0%. Moreover, in the parking lot scene, our framework ranks first with an average precision of 97.8% and an average recall of 98.6%, which demonstrates the effectiveness of our framework.

Xiangyu Yue, Zangwei Zheng, Shanghang Zhang et al.

Unsupervised Domain Adaptation (UDA) transfers predictive models from a fully-labeled source domain to an unlabeled target domain. In some applications, however, it is expensive even to collect labels in the source domain, making most previous works impractical. To cope with this problem, recent work performed instance-wise cross-domain self-supervised learning, followed by an additional fine-tuning stage. However, the instance-wise self-supervised learning only learns and aligns low-level discriminative features. In this paper, we propose an end-to-end Prototypical Cross-domain Self-Supervised Learning (PCS) framework for Few-shot Unsupervised Domain Adaptation (FUDA). PCS not only performs cross-domain low-level feature alignment, but it also encodes and aligns semantic structures in the shared embedding space across domains. Our framework captures category-wise semantic structures of the data by in-domain prototypical contrastive learning; and performs feature alignment through cross-domain prototypical self-supervision. Compared with state-of-the-art methods, PCS improves the mean classification accuracy over different domain pairs on FUDA by 10.5%, 3.5%, 9.0%, and 13.2% on Office, Office-Home, VisDA-2017, and DomainNet, respectively. Our project page is at http://xyue.io/pcs-fuda/index.html