Retrieval-Augmented Generation (RAG) has shown significant improvements in various natural language processing tasks by integrating the strengths of large language models (LLMs) and external knowledge databases. However, RAG introduces long sequence generation and leads to high computation and memory costs. We propose Thoth, a novel multilevel dynamic caching system tailored for RAG. Our analysis benchmarks current RAG systems, pinpointing the performance bottleneck (i.e., long sequence due to knowledge injection) and optimization opportunities (i.e., caching knowledge's intermediate states). Based on these insights, we design Thoth, which organizes the intermediate states of retrieved knowledge in a knowledge tree and caches them in the GPU and host memory hierarchy. Thoth proposes a replacement policy that is aware of LLM inference characteristics and RAG retrieval patterns. It also dynamically overlaps the retrieval and inference steps to minimize the end-to-end latency. We implement Thoth and evaluate it on vLLM, a state-of-the-art LLM inference system and Faiss, a state-of-the-art vector database. The experimental results show that Thoth reduces the time to first token (TTFT) by up to 4x and improves the throughput by up to 2.1x compared to vLLM integrated with Faiss.
The context window of large language models (LLMs) is rapidly increasing, leading to a huge variance in resource usage between different requests as well as between different phases of the same request. Restricted by static parallelism strategies, existing LLM serving systems cannot efficiently utilize the underlying resources to serve variable-length requests in different phases. To address this problem, we propose a new parallelism paradigm, elastic sequence parallelism (ESP), to elastically adapt to the variance between different requests and phases. Based on ESP, we design and build LoongServe, an LLM serving system that (1) improves computation efficiency by elastically adjusting the degree of parallelism in real-time, (2) improves communication efficiency by reducing key-value cache migration overhead and overlapping partial decoding communication with computation, and (3) improves GPU memory efficiency by reducing key-value cache fragmentation across instances. Our evaluation under diverse real-world datasets shows that LoongServe improves the maximum throughput by up to 3.85$\times$ compared to the chunked prefill and 5.81$\times$ compared to the prefill-decoding disaggregation.
Language-image pre-training largely relies on how precisely and thoroughly a text describes its paired image. In practice, however, the contents of an image can be so rich that well describing them requires lengthy captions (e.g., with 10 sentences), which are usually missing in existing datasets. Consequently, there are currently no clear evidences on whether and how language-image pre-training could benefit from long captions. To figure this out, we first re-caption 30M images with detailed descriptions using a pre-trained Multi-modality Large Language Model (MLLM), and then study the usage of the resulting captions under a contrastive learning framework. We observe that, each sentence within a long caption is very likely to describe the image partially (e.g., an object). Motivated by this, we propose to dynamically sample sub-captions from the text label to construct multiple positive pairs, and introduce a grouping loss to match the embeddings of each sub-caption with its corresponding local image patches in a self-supervised manner. Experimental results on a wide rage of downstream tasks demonstrate the consistent superiority of our method, termed DreamLIP, over previous alternatives, highlighting its fine-grained representational capacity. It is noteworthy that, on the tasks of image-text retrieval and semantic segmentation, our model trained with 30M image-text pairs achieves on par or even better performance than CLIP trained with 400M pairs. Project page is available at https://zyf0619sjtu.github.io/dream-lip.
Humans constantly interact with their surrounding environments. Current human-centric generative models mainly focus on synthesizing humans plausibly interacting with static scenes and objects, while the dynamic human action-reaction synthesis for ubiquitous causal human-human interactions is less explored. Human-human interactions can be regarded as asymmetric with actors and reactors in atomic interaction periods. In this paper, we comprehensively analyze the asymmetric, dynamic, synchronous, and detailed nature of human-human interactions and propose the first multi-setting human action-reaction synthesis benchmark to generate human reactions conditioned on given human actions. To begin with, we propose to annotate the actor-reactor order of the interaction sequences for the NTU120, InterHuman, and Chi3D datasets. Based on them, a diffusion-based generative model with a Transformer decoder architecture called ReGenNet together with an explicit distance-based interaction loss is proposed to predict human reactions in an online manner, where the future states of actors are unavailable to reactors. Quantitative and qualitative results show that our method can generate instant and plausible human reactions compared to the baselines, and can generalize to unseen actor motions and viewpoint changes.
This article describes the 2023 IEEE Low-Power Computer Vision Challenge (LPCVC). Since 2015, LPCVC has been an international competition devoted to tackling the challenge of computer vision (CV) on edge devices. Most CV researchers focus on improving accuracy, at the expense of ever-growing sizes of machine models. LPCVC balances accuracy with resource requirements. Winners must achieve high accuracy with short execution time when their CV solutions run on an embedded device, such as Raspberry PI or Nvidia Jetson Nano. The vision problem for 2023 LPCVC is segmentation of images acquired by Unmanned Aerial Vehicles (UAVs, also called drones) after disasters. The 2023 LPCVC attracted 60 international teams that submitted 676 solutions during the submission window of one month. This article explains the setup of the competition and highlights the winners' methods that improve accuracy and shorten execution time.
We present the design, implementation and engineering experience in building and deploying MegaScale, a production system for training large language models (LLMs) at the scale of more than 10,000 GPUs. Training LLMs at this scale brings unprecedented challenges to training efficiency and stability. We take a full-stack approach that co-designs the algorithmic and system components across model block and optimizer design, computation and communication overlapping, operator optimization, data pipeline, and network performance tuning. Maintaining high efficiency throughout the training process (i.e., stability) is an important consideration in production given the long extent of LLM training jobs. Many hard stability issues only emerge at large scale, and in-depth observability is the key to address them. We develop a set of diagnosis tools to monitor system components and events deep in the stack, identify root causes, and derive effective techniques to achieve fault tolerance and mitigate stragglers. MegaScale achieves 55.2% Model FLOPs Utilization (MFU) when training a 175B LLM model on 12,288 GPUs, improving the MFU by 1.34x compared to Megatron-LM. We share our operational experience in identifying and fixing failures and stragglers. We hope by articulating the problems and sharing our experience from a systems perspective, this work can inspire future LLM systems research.
This systematic literature review comprehensively examines the application of Large Language Models (LLMs) in forecasting and anomaly detection, highlighting the current state of research, inherent challenges, and prospective future directions. LLMs have demonstrated significant potential in parsing and analyzing extensive datasets to identify patterns, predict future events, and detect anomalous behavior across various domains. However, this review identifies several critical challenges that impede their broader adoption and effectiveness, including the reliance on vast historical datasets, issues with generalizability across different contexts, the phenomenon of model hallucinations, limitations within the models' knowledge boundaries, and the substantial computational resources required. Through detailed analysis, this review discusses potential solutions and strategies to overcome these obstacles, such as integrating multimodal data, advancements in learning methodologies, and emphasizing model explainability and computational efficiency. Moreover, this review outlines critical trends that are likely to shape the evolution of LLMs in these fields, including the push toward real-time processing, the importance of sustainable modeling practices, and the value of interdisciplinary collaboration. Conclusively, this review underscores the transformative impact LLMs could have on forecasting and anomaly detection while emphasizing the need for continuous innovation, ethical considerations, and practical solutions to realize their full potential.
Computational aesthetic evaluation has made remarkable contribution to visual art works, but its application to music is still rare. Currently, subjective evaluation is still the most effective form of evaluating artistic works. However, subjective evaluation of artistic works will consume a lot of human and material resources. The popular AI generated content (AIGC) tasks nowadays have flooded all industries, and music is no exception. While compared to music produced by humans, AI generated music still sounds mechanical, monotonous, and lacks aesthetic appeal. Due to the lack of music datasets with rating annotations, we have to choose traditional aesthetic equations to objectively measure the beauty of music. In order to improve the quality of AI music generation and further guide computer music production, synthesis, recommendation and other tasks, we use Birkhoff's aesthetic measure to design a aesthetic model, objectively measuring the aesthetic beauty of music, and form a recommendation list according to the aesthetic feeling of music. Experiments show that our objective aesthetic model and recommendation method are effective.
Representation disentanglement may help AI fundamentally understand the real world and thus benefit both discrimination and generation tasks. It currently has at least three unresolved core issues: (i) heavy reliance on label annotation and synthetic data -- causing poor generalization on natural scenarios; (ii) heuristic/hand-craft disentangling constraints make it hard to adaptively achieve an optimal training trade-off; (iii) lacking reasonable evaluation metric, especially for the real label-free data. To address these challenges, we propose a \textbf{C}losed-\textbf{L}oop unsupervised representation \textbf{Dis}entanglement approach dubbed \textbf{CL-Dis}. Specifically, we use diffusion-based autoencoder (Diff-AE) as a backbone while resorting to $\beta$-VAE as a co-pilot to extract semantically disentangled representations. The strong generation ability of diffusion model and the good disentanglement ability of VAE model are complementary. To strengthen disentangling, VAE-latent distillation and diffusion-wise feedback are interconnected in a closed-loop system for a further mutual promotion. Then, a self-supervised \textbf{Navigation} strategy is introduced to identify interpretable semantic directions in the disentangled latent space. Finally, a new metric based on content tracking is designed to evaluate the disentanglement effect. Experiments demonstrate the superiority of CL-Dis on applications like real image manipulation and visual analysis.
Large foundation models, including large language models (LLMs), vision transformers (ViTs), diffusion, and LLM-based multimodal models, are revolutionizing the entire machine learning lifecycle, from training to deployment. However, the substantial advancements in versatility and performance these models offer come at a significant cost in terms of hardware resources. To support the growth of these large models in a scalable and environmentally sustainable way, there has been a considerable focus on developing resource-efficient strategies. This survey delves into the critical importance of such research, examining both algorithmic and systemic aspects. It offers a comprehensive analysis and valuable insights gleaned from existing literature, encompassing a broad array of topics from cutting-edge model architectures and training/serving algorithms to practical system designs and implementations. The goal of this survey is to provide an overarching understanding of how current approaches are tackling the resource challenges posed by large foundation models and to potentially inspire future breakthroughs in this field.