Abstract:Low-bit quantization has become widespread for compressing image super-resolution (SR) models for edge deployment, which allows advanced SR models to enjoy compact low-bit parameters and efficient integer/bitwise constructions for storage compression and inference acceleration, respectively. However, it is notorious that low-bit quantization degrades the accuracy of SR models compared to their full-precision (FP) counterparts. Despite several efforts to alleviate the degradation, the transformer-based SR model still suffers severe degradation due to its distinctive activation distribution. In this work, we present a dual-stage low-bit post-training quantization (PTQ) method for image super-resolution, namely 2DQuant, which achieves efficient and accurate SR under low-bit quantization. The proposed method first investigates the weight and activation and finds that the distribution is characterized by coexisting symmetry and asymmetry, long tails. Specifically, we propose Distribution-Oriented Bound Initialization (DOBI), using different searching strategies to search a coarse bound for quantizers. To obtain refined quantizer parameters, we further propose Distillation Quantization Calibration (DQC), which employs a distillation approach to make the quantized model learn from its FP counterpart. Through extensive experiments on different bits and scaling factors, the performance of DOBI can reach the state-of-the-art (SOTA) while after stage two, our method surpasses existing PTQ in both metrics and visual effects. 2DQuant gains an increase in PSNR as high as 4.52dB on Set5 (x2) compared with SOTA when quantized to 2-bit and enjoys a 3.60x compression ratio and 5.08x speedup ratio. The code and models will be available at https://github.com/Kai-Liu001/2DQuant.
Abstract:Advanced diffusion models (DMs) perform impressively in image super-resolution (SR), but the high memory and computational costs hinder their deployment. Binarization, an ultra-compression algorithm, offers the potential for effectively accelerating DMs. Nonetheless, due to the model structure and the multi-step iterative attribute of DMs, existing binarization methods result in significant performance degradation. In this paper, we introduce a novel binarized diffusion model, BI-DiffSR, for image SR. First, for the model structure, we design a UNet architecture optimized for binarization. We propose the consistent-pixel-downsample (CP-Down) and consistent-pixel-upsample (CP-Up) to maintain dimension consistent and facilitate the full-precision information transfer. Meanwhile, we design the channel-shuffle-fusion (CS-Fusion) to enhance feature fusion in skip connection. Second, for the activation difference across timestep, we design the timestep-aware redistribution (TaR) and activation function (TaA). The TaR and TaA dynamically adjust the distribution of activations based on different timesteps, improving the flexibility and representation alability of the binarized module. Comprehensive experiments demonstrate that our BI-DiffSR outperforms existing binarization methods. Code is available at https://github.com/zhengchen1999/BI-DiffSR.
Abstract:In recent years, significant progress has been made in multivariate time series forecasting using Linear-based, Transformer-based, and Convolution-based models. However, these approaches face notable limitations: linear forecasters struggle with representation capacities, attention mechanisms suffer from quadratic complexity, and convolutional models have a restricted receptive field. These constraints impede their effectiveness in modeling complex time series, particularly those with numerous variables. Additionally, many models adopt the Channel-Independent (CI) strategy, treating multivariate time series as uncorrelated univariate series while ignoring their correlations. For models considering inter-channel relationships, whether through the self-attention mechanism, linear combination, or convolution, they all incur high computational costs and focus solely on weighted summation relationships, neglecting potential proportional relationships between channels. In this work, we address these issues by leveraging the newly introduced state space model and propose \textbf{C-Mamba}, a novel approach that captures cross-channel dependencies while maintaining linear complexity without losing the global receptive field. Our model consists of two key components: (i) channel mixup, where two channels are mixed to enhance the training sets; (ii) channel attention enhanced patch-wise Mamba encoder that leverages the ability of the state space models to capture cross-time dependencies and models correlations between channels by mining their weight relationships. Our model achieves state-of-the-art performance on seven real-world time series datasets. Moreover, the proposed mixup and attention strategy exhibits strong generalizability across other frameworks.
Abstract:Dataset condensation is a newborn technique that generates a small dataset that can be used in training deep neural networks to lower training costs. The objective of dataset condensation is to ensure that the model trained with the synthetic dataset can perform comparably to the model trained with full datasets. However, existing methods predominantly concentrate on classification tasks, posing challenges in their adaptation to time series forecasting (TS-forecasting). This challenge arises from disparities in the evaluation of synthetic data. In classification, the synthetic data is considered well-distilled if the model trained with the full dataset and the model trained with the synthetic dataset yield identical labels for the same input, regardless of variations in output logits distribution. Conversely, in TS-forecasting, the effectiveness of synthetic data distillation is determined by the distance between predictions of the two models. The synthetic data is deemed well-distilled only when all data points within the predictions are similar. Consequently, TS-forecasting has a more rigorous evaluation methodology compared to classification. To mitigate this gap, we theoretically analyze the optimization objective of dataset condensation for TS-forecasting and propose a new one-line plugin of dataset condensation designated as Dataset Condensation for Time Series Forecasting (CondTSF) based on our analysis. Plugging CondTSF into previous dataset condensation methods facilitates a reduction in the distance between the predictions of the model trained with the full dataset and the model trained with the synthetic dataset, thereby enhancing performance. We conduct extensive experiments on eight commonly used time series datasets. CondTSF consistently improves the performance of all previous dataset condensation methods across all datasets, particularly at low condensing ratios.
Abstract:Recent literature has found that an effective method to customize or further improve large language models (LLMs) is to add dynamic adapters, such as low-rank adapters (LoRA) with Mixture-of-Experts (MoE) structures. Though such dynamic adapters incur modest computational complexity, they surprisingly lead to huge inference latency overhead, slowing down the decoding speed by 2.5+ times. In this paper, we analyze the fine-grained costs of the dynamic adapters and find that the fragmented CUDA kernel calls are the root cause. Therefore, we propose LoRA-Switch, a system-algorithm co-designed architecture for efficient dynamic adapters. Unlike most existing dynamic structures that adopt layer-wise or block-wise dynamic routing, LoRA-Switch introduces a token-wise routing mechanism. It switches the LoRA adapters and weights for each token and merges them into the backbone for inference. For efficiency, this switching is implemented with an optimized CUDA kernel, which fuses the merging operations for all LoRA adapters at once. Based on experiments with popular open-source LLMs on common benchmarks, our approach has demonstrated similar accuracy improvement as existing dynamic adapters, while reducing the decoding latency by more than 2.4 times.
Abstract:Image super-resolution (SR) methods typically model degradation to improve reconstruction accuracy in complex and unknown degradation scenarios. However, extracting degradation information from low-resolution images is challenging, which limits the model performance. To boost image SR performance, one feasible approach is to introduce additional priors. Inspired by advancements in multi-modal methods and text prompt image processing, we introduce text prompts to image SR to provide degradation priors. Specifically, we first design a text-image generation pipeline to integrate text into SR dataset through the text degradation representation and degradation model. The text representation applies a discretization manner based on the binning method to describe the degradation abstractly. This representation method can also maintain the flexibility of language. Meanwhile, we propose the PromptSR to realize the text prompt SR. The PromptSR employs the diffusion model and the pre-trained language model (e.g., T5 and CLIP). We train the model on the generated text-image dataset. Extensive experiments indicate that introducing text prompts into image SR, yields excellent results on both synthetic and real-world images. Code: https://github.com/zhengchen1999/PromptSR.
Abstract:3D whole-body human mesh recovery aims to reconstruct the 3D human body, face, and hands from a single image. Although powerful deep learning models have achieved accurate estimation in this task, they require enormous memory and computational resources. Consequently, these methods can hardly be deployed on resource-limited edge devices. In this work, we propose a Binarized Dual Residual Network (BiDRN), a novel quantization method to estimate the 3D human body, face, and hands parameters efficiently. Specifically, we design a basic unit Binarized Dual Residual Block (BiDRB) composed of Local Convolution Residual (LCR) and Block Residual (BR), which can preserve full-precision information as much as possible. For LCR, we generalize it to four kinds of convolutional modules so that full-precision information can be propagated even between mismatched dimensions. We also binarize the face and hands box-prediction network as Binaried BoxNet, which can further reduce the model redundancy. Comprehensive quantitative and qualitative experiments demonstrate the effectiveness of BiDRN, which has a significant improvement over state-of-the-art binarization algorithms. Moreover, our proposed BiDRN achieves comparable performance with full-precision method Hand4Whole while using just 22.1% parameters and 14.8% operations. We will release all the code and pretrained models.
Abstract:Large language models (LLMs) have become phenomenally surging, since 2018--two decades after introducing context-awareness into computing systems. Through taking into account the situations of ubiquitous devices, users and the societies, context-aware computing has enabled a wide spectrum of innovative applications, such as assisted living, location-based social network services and so on. To recognize contexts and make decisions for actions accordingly, various artificial intelligence technologies, such as Ontology and OWL, have been adopted as representations for context modeling and reasoning. Recently, with the rise of LLMs and their improved natural language understanding and reasoning capabilities, it has become feasible to model contexts using natural language and perform context reasoning by interacting with LLMs such as ChatGPT and GPT-4. In this tutorial, we demonstrate the use of texts, prompts, and autonomous agents (AutoAgents) that enable LLMs to perform context modeling and reasoning without requiring fine-tuning of the model. We organize and introduce works in the related field, and name this computing paradigm as the LLM-driven Context-aware Computing (LCaC). In the LCaC paradigm, users' requests, sensors reading data, and the command to actuators are supposed to be represented as texts. Given the text of users' request and sensor data, the AutoAgent models the context by prompting and sends to the LLM for context reasoning. LLM generates a plan of actions and responds to the AutoAgent, which later follows the action plan to foster context-awareness. To prove the concepts, we use two showcases--(1) operating a mobile z-arm in an apartment for assisted living, and (2) planning a trip and scheduling the itinerary in a context-aware and personalized manner.
Abstract:Reproducing research results in the networking community is important for both academia and industry. The current best practice typically resorts to three approaches: (1) looking for publicly available prototypes; (2) contacting the authors to get a private prototype; and (3) manually implementing a prototype following the description of the publication. However, most published network research does not have public prototypes and private prototypes are hard to get. As such, most reproducing efforts are spent on manual implementation based on the publications, which is both time and labor consuming and error-prone. In this paper, we boldly propose reproducing network research results using the emerging large language models (LLMs). In particular, we first prove its feasibility with a small-scale experiment, in which four students with essential networking knowledge each reproduces a different networking system published in prominent conferences and journals by prompt engineering ChatGPT. We report the experiment's observations and lessons and discuss future open research questions of this proposal. This work raises no ethical issue.
Abstract:Mixture of experts (MoE) is a popular technique in deep learning that improves model capacity with conditionally-activated parallel neural network modules (experts). However, serving MoE models in resource-constrained latency-critical edge scenarios is challenging due to the significantly increased model size and complexity. In this paper, we first analyze the behavior pattern of MoE models in continuous inference scenarios, which leads to three key observations about the expert activations, including temporal locality, exchangeability, and skippable computation. Based on these observations, we introduce PC-MoE, an inference framework for resource-constrained continuous MoE model serving. The core of PC-MoE is a new data structure, Parameter Committee, that intelligently maintains a subset of important experts in use to reduce resource consumption. The optimal configuration of Parameter Committee is found offline by a profiling-guided committee planner, and expert swapping and request handling at runtime are managed by an adaptive committee scheduler. To evaluate the effectiveness of PC-MoE, we conduct experiments using state-of-the-art MoE models on common computer vision and natural language processing tasks. The results demonstrate optimal trade-offs between resource consumption and model accuracy achieved by PC-MoE. For instance, on object detection tasks with the Swin-MoE model, our approach can reduce memory usage and latency by 42.34% and 18.63% with only 0.10% accuracy degradation.