Yusheng Xie

Amazon AGI, Aaron Langford, Aayush Shah et al. · amazon-science

We present Amazon Nova, a new generation of state-of-the-art foundation models that deliver frontier intelligence and industry-leading price performance. Amazon Nova Pro is a highly-capable multimodal model with the best combination of accuracy, speed, and cost for a wide range of tasks. Amazon Nova Lite is a low-cost multimodal model that is lightning fast for processing images, video, documents and text. Amazon Nova Micro is a text-only model that delivers our lowest-latency responses at very low cost. Amazon Nova Canvas is an image generation model that creates professional grade images with rich customization controls. Amazon Nova Reel is a video generation model offering high-quality outputs, customization, and motion control. Our models were built responsibly and with a commitment to customer trust, security, and reliability. We report benchmarking results for core capabilities, agentic performance, long context, functional adaptation, runtime performance, and human evaluation.

Taojiannan Yang, Yi Zhu, Yusheng Xie et al.

Recent vision transformer based video models mostly follow the ``image pre-training then finetuning" paradigm and have achieved great success on multiple video benchmarks. However, full finetuning such a video model could be computationally expensive and unnecessary, given the pre-trained image transformer models have demonstrated exceptional transferability. In this work, we propose a novel method to Adapt pre-trained Image Models (AIM) for efficient video understanding. By freezing the pre-trained image model and adding a few lightweight Adapters, we introduce spatial adaptation, temporal adaptation and joint adaptation to gradually equip an image model with spatiotemporal reasoning capability. We show that our proposed AIM can achieve competitive or even better performance than prior arts with substantially fewer tunable parameters on four video action recognition benchmarks. Thanks to its simplicity, our method is also generally applicable to different image pre-trained models, which has the potential to leverage more powerful image foundation models in the future. The project webpage is \url{https://adapt-image-models.github.io/}.

Yash Patel, Yusheng Xie, Yi Zhu et al. · amazon-science

Learning to segment images purely by relying on the image-text alignment from web data can lead to sub-optimal performance due to noise in the data. The noise comes from the samples where the associated text does not correlate with the image's visual content. Instead of purely relying on the alignment from the noisy data, this paper proposes a novel loss function termed SimCon, which accounts for intra-modal similarities to determine the appropriate set of positive samples to align. Further, using multiple views of the image (created synthetically) for training and combining the SimCon loss with it makes the training more robust. This version of the loss is termed MV-SimCon. The empirical results demonstrate that using the proposed loss function leads to consistent improvements on zero-shot, text supervised semantic segmentation and outperforms state-of-the-art by $+3.0\%$, $+3.3\%$ and $+6.9\%$ on PASCAL VOC, PASCAL Context and MSCOCO, respectively. With test time augmentations, we set a new record by improving these results further to $58.7\%$, $26.6\%$, and $33.3\%$ on PASCAL VOC, PASCAL Context, and MSCOCO, respectively. In addition, using the proposed loss function leads to robust training and faster convergence.

Xinghe Cheng, Guiyong Zhuang, Yusheng Xie et al.

Personalized exercise recommendation dynamically aligns pedagogical resources with individual knowledge mastery, which is crucial for satisfying students' dynamic learning needs in modern education. The field is currently driven by two dominant paradigms: Item-Level Exercise Recommendation (ILER) optimizes for immediate single-step state transitions, while Path-Level Exercise Recommendation (PLER) constructs coherent learning paths to maximize cumulative gains. Despite sharing the same ultimate objective, disparate evaluation setups have kept these two lines of research isolated, hindering unified benchmarking and fair comparison. To fill the gap, in this paper, we present a Unified Benchmark for Exercise Recommendation (UniER), a comprehensive evaluation framework that unifies ILER and PLER. Specifically, we introduce Weighted Cognitive Gain (WCG) as a unified metric to measure cross-paradigm algorithmic performance. Our benchmark encompasses 9 datasets spanning four generation methods, facilitating the comparison of 18 representative ILER/PLER methods. Through multi-dimensional analyses covering effectiveness, generalizability, robustness, and efficiency, our results reveal the systematic dominance of PLER and expose the pedagogical failure of ILER's fragmented recommendations under extreme sparsity and noise. Furthermore, we provide an open-source codebase of UniER to foster reproducible research and outline potential directions for future investigations.

Peng Tang, Srikar Appalaraju, R. Manmatha et al.

We present Multiple-Question Multiple-Answer (MQMA), a novel approach to do text-VQA in encoder-decoder transformer models. The text-VQA task requires a model to answer a question by understanding multi-modal content: text (typically from OCR) and an associated image. To the best of our knowledge, almost all previous approaches for text-VQA process a single question and its associated content to predict a single answer. In order to answer multiple questions from the same image, each question and content are fed into the model multiple times. In contrast, our proposed MQMA approach takes multiple questions and content as input at the encoder and predicts multiple answers at the decoder in an auto-regressive manner at the same time. We make several novel architectural modifications to standard encoder-decoder transformers to support MQMA. We also propose a novel MQMA denoising pre-training task which is designed to teach the model to align and delineate multiple questions and content with associated answers. MQMA pre-trained model achieves state-of-the-art results on multiple text-VQA datasets, each with strong baselines. Specifically, on OCR-VQA (+2.5%), TextVQA (+1.4%), ST-VQA (+0.6%), DocVQA (+1.1%) absolute improvements over the previous state-of-the-art approaches.

Chao Zeng, Songwei Liu, Yusheng Xie et al.

Large Language Models (LLMs) have revolutionized natural language processing tasks. However, their practical application is constrained by substantial memory and computational demands. Post-training quantization (PTQ) is considered an effective method to accelerate LLM inference. Despite its growing popularity in LLM model compression, PTQ deployment faces two major challenges. First, low-bit quantization leads to performance degradation. Second, restricted by the limited integer computing unit type on GPUs, quantized matrix operations with different precisions cannot be effectively accelerated. To address these issues, we introduce a novel arbitrary-bit quantization algorithm and inference framework, ABQ-LLM. It achieves superior performance across various quantization settings and enables efficient arbitrary-precision quantized inference on the GPU. ABQ-LLM introduces several key innovations: (1) a distribution correction method for transformer blocks to mitigate distribution differences caused by full quantization of weights and activations, improving performance at low bit-widths. (2) the bit balance strategy to counteract performance degradation from asymmetric distribution issues at very low bit-widths (e.g., 2-bit). (3) an innovative quantization acceleration framework that reconstructs the quantization matrix multiplication of arbitrary precision combinations based on BTC (Binary TensorCore) equivalents, gets rid of the limitations of INT4/INT8 computing units. ABQ-LLM can convert each component bit width gain into actual acceleration gain, maximizing performance under mixed precision(e.g., W6A6, W2A8). Based on W2*A8 quantization configuration on LLaMA-7B model, it achieved a WikiText2 perplexity of 7.59 (2.17$\downarrow $ vs 9.76 in AffineQuant). Compared to SmoothQuant, we realized 1.6$\times$ acceleration improvement and 2.7$\times$ memory compression gain.

Liangjian Chen, Shih-Yao Lin, Yusheng Xie et al.

Estimating 3D hand poses from a single RGB image is challenging because depth ambiguity leads the problem ill-posed. Training hand pose estimators with 3D hand mesh annotations and multi-view images often results in significant performance gains. However, existing multi-view datasets are relatively small with hand joints annotated by off-the-shelf trackers or automated through model predictions, both of which may be inaccurate and can introduce biases. Collecting a large-scale multi-view 3D hand pose images with accurate mesh and joint annotations is valuable but strenuous. In this paper, we design a spin match algorithm that enables a rigid mesh model matching with any target mesh ground truth. Based on the match algorithm, we propose an efficient pipeline to generate a large-scale multi-view hand mesh (MVHM) dataset with accurate 3D hand mesh and joint labels. We further present a multi-view hand pose estimation approach to verify that training a hand pose estimator with our generated dataset greatly enhances the performance. Experimental results show that our approach achieves the performance of 0.990 in $\text{AUC}_{\text{20-50}}$ on the MHP dataset compared to the previous state-of-the-art of 0.939 on this dataset. Our datasset is public available. \footnote{\url{https://github.com/Kuzphi/MVHM}} Our datasset is available at~\href{https://github.com/Kuzphi/MVHM}{\color{blue}{https://github.com/Kuzphi/MVHM}}.

Zhenyu Wu, Duc Hoang, Shih-Yao Lin et al.

Estimating the 3D hand pose from a monocular RGB image is important but challenging. A solution is training on large-scale RGB hand images with accurate 3D hand keypoint annotations. However, it is too expensive in practice. Instead, we have developed a learning-based approach to synthesize realistic, diverse, and 3D pose-preserving hand images under the guidance of 3D pose information. We propose a 3D-aware multi-modal guided hand generative network (MM-Hand), together with a novel geometry-based curriculum learning strategy. Our extensive experimental results demonstrate that the 3D-annotated images generated by MM-Hand qualitatively and quantitatively outperform existing options. Moreover, the augmented data can consistently improve the quantitative performance of the state-of-the-art 3D hand pose estimators on two benchmark datasets. The code will be available at https://github.com/ScottHoang/mm-hand.

Robik Shrestha, Yang Zou, Qiuyu Chen et al. · amazon-science

Existing text-to-image generative models reflect or even amplify societal biases ingrained in their training data. This is especially concerning for human image generation where models are biased against certain demographic groups. Existing attempts to rectify this issue are hindered by the inherent limitations of the pre-trained models and fail to substantially improve demographic diversity. In this work, we introduce Fair Retrieval Augmented Generation (FairRAG), a novel framework that conditions pre-trained generative models on reference images retrieved from an external image database to improve fairness in human generation. FairRAG enables conditioning through a lightweight linear module that projects reference images into the textual space. To enhance fairness, FairRAG applies simple-yet-effective debiasing strategies, providing images from diverse demographic groups during the generative process. Extensive experiments demonstrate that FairRAG outperforms existing methods in terms of demographic diversity, image-text alignment, and image fidelity while incurring minimal computational overhead during inference.

Hao Li, Yang Zou, Ying Wang et al. · amazon-science

Scaling up model and data size has been quite successful for the evolution of LLMs. However, the scaling law for the diffusion based text-to-image (T2I) models is not fully explored. It is also unclear how to efficiently scale the model for better performance at reduced cost. The different training settings and expensive training cost make a fair model comparison extremely difficult. In this work, we empirically study the scaling properties of diffusion based T2I models by performing extensive and rigours ablations on scaling both denoising backbones and training set, including training scaled UNet and Transformer variants ranging from 0.4B to 4B parameters on datasets upto 600M images. For model scaling, we find the location and amount of cross attention distinguishes the performance of existing UNet designs. And increasing the transformer blocks is more parameter-efficient for improving text-image alignment than increasing channel numbers. We then identify an efficient UNet variant, which is 45% smaller and 28% faster than SDXL's UNet. On the data scaling side, we show the quality and diversity of the training set matters more than simply dataset size. Increasing caption density and diversity improves text-image alignment performance and the learning efficiency. Finally, we provide scaling functions to predict the text-image alignment performance as functions of the scale of model size, compute and dataset size.

Hossein Aboutalebi, Hwanjun Song, Yusheng Xie et al.

Development of multimodal interactive systems is hindered by the lack of rich, multimodal (text, images) conversational data, which is needed in large quantities for LLMs. Previous approaches augment textual dialogues with retrieved images, posing privacy, diversity, and quality constraints. In this work, we introduce Multimodal Augmented Generative Images Dialogues (MAGID), a framework to augment text-only dialogues with diverse and high-quality images. Subsequently, a diffusion model is applied to craft corresponding images, ensuring alignment with the identified text. Finally, MAGID incorporates an innovative feedback loop between an image description generation module (textual LLM) and image quality modules (addressing aesthetics, image-text matching, and safety), that work in tandem to generate high-quality and multi-modal dialogues. We compare MAGID to other SOTA baselines on three dialogue datasets, using automated and human evaluation. Our results show that MAGID is comparable to or better than baselines, with significant improvements in human evaluation, especially against retrieval baselines where the image database is small.

Hao Li, Shamit Lal, Zhiheng Li et al. · amazon-science

We empirically study the scaling properties of various Diffusion Transformers (DiTs) for text-to-image generation by performing extensive and rigorous ablations, including training scaled DiTs ranging from 0.3B upto 8B parameters on datasets up to 600M images. We find that U-ViT, a pure self-attention based DiT model provides a simpler design and scales more effectively in comparison with cross-attention based DiT variants, which allows straightforward expansion for extra conditions and other modalities. We identify a 2.3B U-ViT model can get better performance than SDXL UNet and other DiT variants in controlled setting. On the data scaling side, we investigate how increasing dataset size and enhanced long caption improve the text-image alignment performance and the learning efficiency.

Benjamin Biggs, Arjun Seshadri, Yang Zou et al.

We present Diffusion Soup, a compartmentalization method for Text-to-Image Generation that averages the weights of diffusion models trained on sharded data. By construction, our approach enables training-free continual learning and unlearning with no additional memory or inference costs, since models corresponding to data shards can be added or removed by re-averaging. We show that Diffusion Soup samples from a point in weight space that approximates the geometric mean of the distributions of constituent datasets, which offers anti-memorization guarantees and enables zero-shot style mixing. Empirically, Diffusion Soup outperforms a paragon model trained on the union of all data shards and achieves a 30% improvement in Image Reward (.34 $\to$ .44) on domain sharded data, and a 59% improvement in IR (.37 $\to$ .59) on aesthetic data. In both cases, souping also prevails in TIFA score (respectively, 85.5 $\to$ 86.5 and 85.6 $\to$ 86.8). We demonstrate robust unlearning -- removing any individual domain shard only lowers performance by 1% in IR (.45 $\to$ .44) -- and validate our theoretical insights on anti-memorization using real data. Finally, we showcase Diffusion Soup's ability to blend the distinct styles of models finetuned on different shards, resulting in the zero-shot generation of hybrid styles.

Simone Bombari, Alessandro Achille, Zijian Wang et al.

Memorization of the relation between entities in a dataset can lead to privacy issues when using a trained model for question answering. We introduce Relational Memorization (RM) to understand, quantify and control this phenomenon. While bounding general memorization can have detrimental effects on the performance of a trained model, bounding RM does not prevent effective learning. The difference is most pronounced when the data distribution is long-tailed, with many queries having only few training examples: Impeding general memorization prevents effective learning, while impeding only relational memorization still allows learning general properties of the underlying concepts. We formalize the notion of Relational Privacy (RP) and, inspired by Differential Privacy (DP), we provide a possible definition of Differential Relational Privacy (DrP). These notions can be used to describe and compute bounds on the amount of RM in a trained model. We illustrate Relational Privacy concepts in experiments with large-scale models for Question Answering.

Ali Furkan Biten, Ron Litman, Yusheng Xie et al.

We propose a novel multimodal architecture for Scene Text Visual Question Answering (STVQA), named Layout-Aware Transformer (LaTr). The task of STVQA requires models to reason over different modalities. Thus, we first investigate the impact of each modality, and reveal the importance of the language module, especially when enriched with layout information. Accounting for this, we propose a single objective pre-training scheme that requires only text and spatial cues. We show that applying this pre-training scheme on scanned documents has certain advantages over using natural images, despite the domain gap. Scanned documents are easy to procure, text-dense and have a variety of layouts, helping the model learn various spatial cues (e.g. left-of, below etc.) by tying together language and layout information. Compared to existing approaches, our method performs vocabulary-free decoding and, as shown, generalizes well beyond the training vocabulary. We further demonstrate that LaTr improves robustness towards OCR errors, a common reason for failure cases in STVQA. In addition, by leveraging a vision transformer, we eliminate the need for an external object detector. LaTr outperforms state-of-the-art STVQA methods on multiple datasets. In particular, +7.6% on TextVQA, +10.8% on ST-VQA and +4.0% on OCR-VQA (all absolute accuracy numbers).

Haoyu Ma, Liangjian Chen, Deying Kong et al.

Estimating the 2D human poses in each view is typically the first step in calibrated multi-view 3D pose estimation. But the performance of 2D pose detectors suffers from challenging situations such as occlusions and oblique viewing angles. To address these challenges, previous works derive point-to-point correspondences between different views from epipolar geometry and utilize the correspondences to merge prediction heatmaps or feature representations. Instead of post-prediction merge/calibration, here we introduce a transformer framework for multi-view 3D pose estimation, aiming at directly improving individual 2D predictors by integrating information from different views. Inspired by previous multi-modal transformers, we design a unified transformer architecture, named TransFusion, to fuse cues from both current views and neighboring views. Moreover, we propose the concept of epipolar field to encode 3D positional information into the transformer model. The 3D position encoding guided by the epipolar field provides an efficient way of encoding correspondences between pixels of different views. Experiments on Human 3.6M and Ski-Pose show that our method is more efficient and has consistent improvements compared to other fusion methods. Specifically, we achieve 25.8 mm MPJPE on Human 3.6M with only 5M parameters on 256 x 256 resolution.

Srikar Appalaraju, Bhavan Jasani, Bhargava Urala Kota et al.

We present DocFormer -- a multi-modal transformer based architecture for the task of Visual Document Understanding (VDU). VDU is a challenging problem which aims to understand documents in their varied formats (forms, receipts etc.) and layouts. In addition, DocFormer is pre-trained in an unsupervised fashion using carefully designed tasks which encourage multi-modal interaction. DocFormer uses text, vision and spatial features and combines them using a novel multi-modal self-attention layer. DocFormer also shares learned spatial embeddings across modalities which makes it easy for the model to correlate text to visual tokens and vice versa. DocFormer is evaluated on 4 different datasets each with strong baselines. DocFormer achieves state-of-the-art results on all of them, sometimes beating models 4x its size (in no. of parameters).

Liangjian Chen, Shih-Yao Lin, Yusheng Xie et al.

Estimating 3D hand pose directly from RGB imagesis challenging but has gained steady progress recently bytraining deep models with annotated 3D poses. Howeverannotating 3D poses is difficult and as such only a few 3Dhand pose datasets are available, all with limited samplesizes. In this study, we propose a new framework of training3D pose estimation models from RGB images without usingexplicit 3D annotations, i.e., trained with only 2D informa-tion. Our framework is motivated by two observations: 1)Videos provide richer information for estimating 3D posesas opposed to static images; 2) Estimated 3D poses oughtto be consistent whether the videos are viewed in the for-ward order or reverse order. We leverage these two obser-vations to develop a self-supervised learning model calledtemporal-aware self-supervised network (TASSN). By en-forcing temporal consistency constraints, TASSN learns 3Dhand poses and meshes from videos with only 2D keypointposition annotations. Experiments show that our modelachieves surprisingly good results, with 3D estimation ac-curacy on par with the state-of-the-art models trained with3D annotations, highlighting the benefit of the temporalconsistency in constraining 3D prediction models.

Liangjian Chen, Shih-Yao Lin, Yusheng Xie et al.

Estimating3D hand poses from RGB images is essentialto a wide range of potential applications, but is challengingowing to substantial ambiguity in the inference of depth in-formation from RGB images. State-of-the-art estimators ad-dress this problem by regularizing3D hand pose estimationmodels during training to enforce the consistency betweenthe predicted3D poses and the ground-truth depth maps.However, these estimators rely on both RGB images and thepaired depth maps during training. In this study, we proposea conditional generative adversarial network (GAN) model,called Depth-image Guided GAN (DGGAN), to generate re-alistic depth maps conditioned on the input RGB image, anduse the synthesized depth maps to regularize the3D handpose estimation model, therefore eliminating the need forground-truth depth maps. Experimental results on multiplebenchmark datasets show that the synthesized depth mapsproduced by DGGAN are quite effective in regularizing thepose estimation model, yielding new state-of-the-art resultsin estimation accuracy, notably reducing the mean3D end-point errors (EPE) by4.7%,16.5%, and6.8%on the RHD,STB and MHP datasets, respectively.

Srikar Appalaraju, Yi Zhu, Yusheng Xie et al.

Self-supervised representation learning has seen remarkable progress in the last few years. More recently, contrastive instance learning has shown impressive results compared to its supervised learning counterparts. However, even with the ever increased interest in contrastive instance learning, it is still largely unclear why these methods work so well. In this paper, we aim to unravel some of the mysteries behind their success, which are the good practices. Through an extensive empirical analysis, we hope to not only provide insights but also lay out a set of best practices that led to the success of recent work in self-supervised representation learning.

Liangjian Chen, Shih-Yao Lin, Yusheng Xie et al.

Hand pose estimation from a monocular RGB image is an important but challenging task. The main factor affecting its performance is the lack of a sufficiently large training dataset with accurate hand-keypoint annotations. In this work, we circumvent this problem by proposing an effective method for generating realistic hand poses and show that state-of-the-art algorithms for hand pose estimation can be greatly improved by utilizing the generated hand poses as training data. Specifically, we first adopt an augmented reality (AR) simulator to synthesize hand poses with accurate hand-keypoint labels. Although the synthetic hand poses come with precise joint labels, eliminating the need of manual annotations, they look unnatural and are not the ideal training data. To produce more realistic hand poses, we propose to blend a synthetic hand pose with a real background, such as arms and sleeves. To this end, we develop tonality-alignment generative adversarial networks (TAGANs), which align the tonality and color distributions between synthetic hand poses and real backgrounds, and can generate high quality hand poses. We evaluate TAGAN on three benchmarks, including the RHP, STB, and CMU-PS hand pose datasets. With the aid of the synthesized poses, our method performs favorably against the state-of-the-arts in both 2D and 3D hand pose estimations.

Sheng Shen, Yaliang Li, Nan Du et al.

Question answering (QA) has achieved promising progress recently. However, answering a question in real-world scenarios like the medical domain is still challenging, due to the requirement of external knowledge and the insufficient quantity of high-quality training data. In the light of these challenges, we study the task of generating medical QA pairs in this paper. With the insight that each medical question can be considered as a sample from the latent distribution of questions given answers, we propose an automated medical QA pair generation framework, consisting of an unsupervised key phrase detector that explores unstructured material for validity, and a generator that involves a multi-pass decoder to integrate structural knowledge for diversity. A series of experiments have been conducted on a real-world dataset collected from the National Medical Licensing Examination of China. Both automatic evaluation and human annotation demonstrate the effectiveness of the proposed method. Further investigation shows that, by incorporating the generated QA pairs for training, significant improvement in terms of accuracy can be achieved for the examination QA system.

Yusheng Xie, Nan Du, Wei Fan et al.

Variational inference provides approximations to the computationally intractable posterior distribution in Bayesian networks. A prominent medical application of noisy-or Bayesian network is to infer potential diseases given observed symptoms. Previous studies focus on approximating a handful of complicated pathological cases using variational transformation. Our goal is to use variational transformation as part of a novel hybridized inference for serving reliable and real time diagnosis at web scale. We propose a hybridized inference that allows variational parameters to be estimated without disease posteriors or priors, making the inference faster and much of its computation recyclable. In addition, we propose a transformation ranking algorithm that is very stable to large variances in network prior probabilities, a common issue that arises in medical applications of Bayesian networks. In experiments, we perform comparative study on a large real life medical network and scalability study on a much larger (36,000x) synthesized network.