The trend of modeless ML inference is increasingly growing in popularity as it hides the complexity of model inference from users and caters to diverse user and application accuracy requirements. Previous work mostly focuses on modeless inference in data centers. To provide low-latency inference, in this paper, we promote modeless inference at the edge. The edge environment introduces additional challenges related to low power consumption, limited device memory, and volatile network environments. To address these challenges, we propose HawkVision, which provides low-latency modeless serving of vision DNNs. HawkVision leverages a two-layer edge-DC architecture that employs confidence scaling to reduce the number of model options while meeting diverse accuracy requirements. It also supports lossy inference under volatile network environments. Our experimental results show that HawkVision outperforms current serving systems by up to 1.6X in P99 latency for providing modeless service. Our FPGA prototype demonstrates similar performance at certain accuracy levels with up to a 3.34X reduction in power consumption.
相關內容
Predictive models are a crucial component of many robotic systems. Yet, constructing accurate predictive models for a variety of deformable objects, especially those with unknown physical properties, remains a significant challenge. This paper introduces AdaptiGraph, a learning-based dynamics modeling approach that enables robots to predict, adapt to, and control a wide array of challenging deformable materials with unknown physical properties. AdaptiGraph leverages the highly flexible graph-based neural dynamics (GBND) framework, which represents material bits as particles and employs a graph neural network (GNN) to predict particle motion. Its key innovation is a unified physical property-conditioned GBND model capable of predicting the motions of diverse materials with varying physical properties without retraining. Upon encountering new materials during online deployment, AdaptiGraph utilizes a physical property optimization process for a few-shot adaptation of the model, enhancing its fit to the observed interaction data. The adapted models can precisely simulate the dynamics and predict the motion of various deformable materials, such as ropes, granular media, rigid boxes, and cloth, while adapting to different physical properties, including stiffness, granular size, and center of pressure. On prediction and manipulation tasks involving a diverse set of real-world deformable objects, our method exhibits superior prediction accuracy and task proficiency over non-material-conditioned and non-adaptive models. The project page is available at //robopil.github.io/adaptigraph/ .
Serving large language models (LLMs) in production can incur substantial costs, which has prompted recent advances in inference system optimizations. Today, these systems are evaluated against conventional latency and throughput metrics (eg. TTFT, TBT, Normalised Latency and TPOT). However, these metrics fail to fully capture the nuances of LLM inference, leading to an incomplete assessment of user-facing performance crucial for real-time applications such as chat and translation. In this paper, we first identify the pitfalls of current performance metrics in evaluating LLM inference systems. We then propose Metron, a comprehensive performance evaluation framework that includes fluidity-index -- a novel metric designed to reflect the intricacies of the LLM inference process and its impact on real-time user experience. Finally, we evaluate various existing open-source platforms and model-as-a-service offerings using Metron, discussing their strengths and weaknesses. Metron is available at //github.com/project-metron/metron.
Large language models (LLMs) have greatly impacted the natural language processing (NLP) field, particularly for the English language. These models have demonstrated capabilities in understanding and generating human-like text. The success of language models largely depends on the availability of high-quality instruction datasets, which consist of detailed task descriptions and corresponding responses that are essential for training the models to address a variety of prompts accurately. However, the availability and quality of these resources vary by language. While models perform well in English, they often need help with languages like Arabic, due to the lack of datasets for fine-tuning Arabic-specific tasks. To address this issue, we introduce InstAr-500k, a new Arabic instruction dataset created by generating and collecting content that covers several domains and instruction types. We assess this dataset by fine-tuning an open-source Gemma-7B model on several downstream tasks to improve its functionality. Based on multiple evaluations, our fine-tuned model achieves excellent performance on several Arabic NLP benchmarks. These outcomes emphasize the effectiveness of our dataset in elevating the capabilities of language models for Arabic. Our instruction dataset bridges the performance gap between English and Arabic language models by providing resources that amplify Arabic NLP development. Building on this foundation, we developed a model, GemmAr-7B-V1, specifically tuned to excel at a wide range of Arabic NLP tasks.
Facial Action Units (AU) is a vital concept in the realm of affective computing, and AU detection has always been a hot research topic. Existing methods suffer from overfitting issues due to the utilization of a large number of learnable parameters on scarce AU-annotated datasets or heavy reliance on substantial additional relevant data. Parameter-Efficient Transfer Learning (PETL) provides a promising paradigm to address these challenges, whereas its existing methods lack design for AU characteristics. Therefore, we innovatively investigate PETL paradigm to AU detection, introducing AUFormer and proposing a novel Mixture-of-Knowledge Expert (MoKE) collaboration mechanism. An individual MoKE specific to a certain AU with minimal learnable parameters first integrates personalized multi-scale and correlation knowledge. Then the MoKE collaborates with other MoKEs in the expert group to obtain aggregated information and inject it into the frozen Vision Transformer (ViT) to achieve parameter-efficient AU detection. Additionally, we design a Margin-truncated Difficulty-aware Weighted Asymmetric Loss (MDWA-Loss), which can encourage the model to focus more on activated AUs, differentiate the difficulty of unactivated AUs, and discard potential mislabeled samples. Extensive experiments from various perspectives, including within-domain, cross-domain, data efficiency, and micro-expression domain, demonstrate AUFormer's state-of-the-art performance and robust generalization abilities without relying on additional relevant data. The code for AUFormer is available at //github.com/yuankaishen2001/AUFormer.
Among the widely used parameter-efficient fine-tuning (PEFT) methods, LoRA and its variants have gained considerable popularity because of avoiding additional inference costs. However, there still often exists an accuracy gap between these methods and full fine-tuning (FT). In this work, we first introduce a novel weight decomposition analysis to investigate the inherent differences between FT and LoRA. Aiming to resemble the learning capacity of FT from the findings, we propose Weight-Decomposed Low-Rank Adaptation (DoRA). DoRA decomposes the pre-trained weight into two components, magnitude and direction, for fine-tuning, specifically employing LoRA for directional updates to efficiently minimize the number of trainable parameters. By employing \ours, we enhance both the learning capacity and training stability of LoRA while avoiding any additional inference overhead. \ours~consistently outperforms LoRA on fine-tuning LLaMA, LLaVA, and VL-BART on various downstream tasks, such as commonsense reasoning, visual instruction tuning, and image/video-text understanding. Code is available at //github.com/NVlabs/DoRA.
Ads relevance models are crucial in determining the relevance between user search queries and ad offers, often framed as a classification problem. The complexity of modeling increases significantly with multiple ad types and varying scenarios that exhibit both similarities and differences. In this work, we introduce a novel multi-faceted attention model that performs task aware feature combination and cross task interaction modeling. Our technique formulates the feature combination problem as "language" modeling with auto-regressive attentions across both feature and task dimensions. Specifically, we introduce a new dimension of task ID encoding for task representations, thereby enabling precise relevance modeling across diverse ad scenarios with substantial improvement in generality capability for unseen tasks. We demonstrate that our model not only effectively handles the increased computational and maintenance demands as scenarios proliferate, but also outperforms generalized DNN models and even task-specific models across a spectrum of ad applications using a single unified model.
Generative commonsense reasoning which aims to empower machines to generate sentences with the capacity of reasoning over a set of concepts is a critical bottleneck for text generation. Even the state-of-the-art pre-trained language generation models struggle at this task and often produce implausible and anomalous sentences. One reason is that they rarely consider incorporating the knowledge graph which can provide rich relational information among the commonsense concepts. To promote the ability of commonsense reasoning for text generation, we propose a novel knowledge graph augmented pre-trained language generation model KG-BART, which encompasses the complex relations of concepts through the knowledge graph and produces more logical and natural sentences as output. Moreover, KG-BART can leverage the graph attention to aggregate the rich concept semantics that enhances the model generalization on unseen concept sets. Experiments on benchmark CommonGen dataset verify the effectiveness of our proposed approach by comparing with several strong pre-trained language generation models, particularly KG-BART outperforms BART by 5.80, 4.60, in terms of BLEU-3, 4. Moreover, we also show that the generated context by our model can work as background scenarios to benefit downstream commonsense QA tasks.
The design of deep graph models still remains to be investigated and the crucial part is how to explore and exploit the knowledge from different hops of neighbors in an efficient way. In this paper, we propose a novel RNN-like deep graph neural network architecture by incorporating AdaBoost into the computation of network; and the proposed graph convolutional network called AdaGCN~(AdaBoosting Graph Convolutional Network) has the ability to efficiently extract knowledge from high-order neighbors and integrate knowledge from different hops of neighbors into the network in an AdaBoost way. We also present the architectural difference between AdaGCN and existing graph convolutional methods to show the benefits of our proposal. Finally, extensive experiments demonstrate the state-of-the-art prediction performance and the computational advantage of our approach AdaGCN.
With the capability of modeling bidirectional contexts, denoising autoencoding based pretraining like BERT achieves better performance than pretraining approaches based on autoregressive language modeling. However, relying on corrupting the input with masks, BERT neglects dependency between the masked positions and suffers from a pretrain-finetune discrepancy. In light of these pros and cons, we propose XLNet, a generalized autoregressive pretraining method that (1) enables learning bidirectional contexts by maximizing the expected likelihood over all permutations of the factorization order and (2) overcomes the limitations of BERT thanks to its autoregressive formulation. Furthermore, XLNet integrates ideas from Transformer-XL, the state-of-the-art autoregressive model, into pretraining. Empirically, XLNet outperforms BERT on 20 tasks, often by a large margin, and achieves state-of-the-art results on 18 tasks including question answering, natural language inference, sentiment analysis, and document ranking.
To provide more accurate, diverse, and explainable recommendation, it is compulsory to go beyond modeling user-item interactions and take side information into account. Traditional methods like factorization machine (FM) cast it as a supervised learning problem, which assumes each interaction as an independent instance with side information encoded. Due to the overlook of the relations among instances or items (e.g., the director of a movie is also an actor of another movie), these methods are insufficient to distill the collaborative signal from the collective behaviors of users. In this work, we investigate the utility of knowledge graph (KG), which breaks down the independent interaction assumption by linking items with their attributes. We argue that in such a hybrid structure of KG and user-item graph, high-order relations --- which connect two items with one or multiple linked attributes --- are an essential factor for successful recommendation. We propose a new method named Knowledge Graph Attention Network (KGAT) which explicitly models the high-order connectivities in KG in an end-to-end fashion. It recursively propagates the embeddings from a node's neighbors (which can be users, items, or attributes) to refine the node's embedding, and employs an attention mechanism to discriminate the importance of the neighbors. Our KGAT is conceptually advantageous to existing KG-based recommendation methods, which either exploit high-order relations by extracting paths or implicitly modeling them with regularization. Empirical results on three public benchmarks show that KGAT significantly outperforms state-of-the-art methods like Neural FM and RippleNet. Further studies verify the efficacy of embedding propagation for high-order relation modeling and the interpretability benefits brought by the attention mechanism.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191