arXiv subjects

Junchen Jiang

Publications and source records attributed to Junchen Jiang.

At least 37 records · Page 2Linked to original sources

arXiv2025

DroidSpeak: KV Cache Sharing for Cross-LLM Communication and Multi-LLM Serving

Compound AI systems, such as agentic systems, are an emerging trend in large-scale enterprise settings, with multiple LLMs specialized for different users, tasks, and/or roles working together. In these scenarios, different models often process inputs that share the same context prefix. Although much work was done in the past to enable the reuse of prefix KV caches across inputs for a single model, how to enable one model to reuse the prefix KV caches of a different model remains an open question. We introduce DroidSpeak, the first distributed LLM inference system that enables KV cache reuse across distributed nodes running inference of different LLMs, so long as the LLMs have the same architecture. We present the first study that aims at understanding the impact of sharing KV caches across different LLMs, and if/when such sharing affects quality. Inspired by the findings, we present DroidSpeak, which selectively recomputes a few layers of the KV cache produced by another LLM and reuses the remaining layers, with negligible quality loss. Moreover, carefully pipelining the layer-wise re-computation and the loading of reused KV cache further improves the inference performance. Experiments on diverse datasets and model pairs demonstrate that DroidSpeak achieves up to 4x throughput improvement and about 3.1x faster prefill (time to first token), with negligible loss of quality in F1 scores, Rouge-L or code similarity score, compared to the baseline which does not allow any sharing across models.

cs.MA↗

arXiv2025

PrefillOnly: An Inference Engine for Prefill-only Workloads in Large Language Model Applications

Besides typical generative applications, like ChatGPT, GitHub Copilot, and Cursor, we observe an emerging trend that LLMs are increasingly used in traditional discriminative tasks, such as recommendation, credit verification, and data labeling. The key characteristic of these emerging use cases is that the LLM generates only a single output token, rather than an arbitrarily long sequence of tokens. We call this prefill-only workload. However, since existing LLM engines assume arbitrary output lengths, they fail to leverage the unique properties of prefill-only workloads. In this paper, we present PrefillOnly, the first LLM inference engine that improves the inference throughput and latency by fully embracing the properties of prefill-only workloads. First, since it generates only one token, PrefillOnly only needs to store the KV cache of only the last computed layer, rather than of all layers. This drastically reduces the GPU memory footprint of LLM inference and allows handling long inputs without using solutions that reduces throughput, such as cross-GPU KV cache parallelization. Second, because the output length is fixed, rather than arbitrary, PrefillOnly can precisely determine the job completion time (JCT) of each prefill-only request before it starts. This enables efficient JCT-aware scheduling policies such as shortest remaining job first. PrefillOnly can process upto 4x larger queries per second without inflating average and P99 latency.

cs.DC↗

arXiv2025

CacheBlend: Fast Large Language Model Serving for RAG with Cached Knowledge Fusion

Large language models (LLMs) often incorporate multiple text chunks in their inputs to provide the necessary contexts. To speed up the prefill of the long LLM inputs, one can pre-compute the KV cache of a text and re-use the KV cache when the context is reused as the prefix of another LLM input. However, the reused text chunks are not always the input prefix, which makes precomputed KV caches not directly usable since they ignore the text's cross-attention with the preceding texts. Thus, the benefits of reusing KV caches remain largely unrealized. This paper tackles just one challenge: when an LLM input contains multiple text chunks, how to quickly combine their precomputed KV caches in order to achieve the same generation quality as the expensive full prefill (i.e., without reusing KV cache)? This challenge naturally arises in retrieval-augmented generation (RAG) where the input is supplemented with multiple retrieved texts as the context. We present CacheBlend, a scheme that reuses the precomputed KV caches, regardless prefix or not, and selectively recomputes the KV values of a small subset of tokens to partially update each reused KV cache. In the meantime, the small extra delay for recomputing some tokens can be pipelined with the retrieval of KV caches within the same job, allowing CacheBlend to store KV caches in slower devices with more storage capacity while retrieving them without increasing the inference delay. By comparing CacheBlend with the state-of-the-art KV cache reusing schemes on three open-source LLMs of various sizes and four popular benchmark datasets of different tasks, we show that CacheBlend reduces time-to-first-token (TTFT) by 2.2-3.3x and increases the inference throughput by 2.8-5x from full KV recompute without compromising generation quality. The code is available at https://github.com/LMCache/LMCache.

cs.LG↗

arXiv2025

HyperRAG: Enhancing Quality-Efficiency Tradeoffs in Retrieval-Augmented Generation with Reranker KV-Cache Reuse

Retrieval-Augmented Generation (RAG) has emerged as a powerful paradigm for enhancing the performance of large language models (LLMs) by integrating external knowledge into the generation process. A key component of RAG pipelines is the reranker, which selects the most relevant documents from a pool of retrieved candidates and significantly improves the quality of the generated responses. While rerankers refine the selection of retrieved documents in RAG pipelines, they introduce computational challenges that hinder high throughput and low latency. To address this problem, we propose HyperRAG, a system that optimizes the trade-off between quality and efficiency in RAG pipelines by leveraging KV-cache reuse for efficient reranker inference. By reusing document-side KV-cache, HyperRAG achieves both high-quality generation and system-level efficiency. To fully realize the benefits of KV-cache reuse, HyperRAG incorporates a range of system-level optimizations designed to enhance efficiency and scalability. Experiments show that HyperRAG achieves a 2 - 3 throughput improvement with decoder-only rerankers while also delivering higher downstream performance compared with traditional RAG service.

cs.CL↗

arXiv2025

Towards More Economical Context-Augmented LLM Generation by Reusing Stored KV Cache

Across large language model (LLM) applications, we observe an emerging trend for reusing KV caches to save the prefill delays of processing repeated input texts in different LLM inputs. This has led to a broad design space, including colocating stored KV caches with (or close to) GPUs to various KV cache compression. However, a key question remains unanswered: can these delay reductions also be economically favorable? Specifically, we ask whether a developer can use public cloud services to store precomputed KV caches and reuse them to save delay without incurring more costs in terms of compute, storage, and network. To answer this question, we propose an validated analytical model for the cloud cost (in compute, storage, and network) of storing and reusing KV caches based on various workload parameters, such as reuse frequency, generated text lengths, model sizes, etc. Preliminary results show that KV cache reusing is able to save both delay and cloud cost across a range of workloads with long context. And we call more efforts on building more economical context augmented LLM by KV cache reusing.

cs.NI↗

arXiv2025

Loss-tolerant neural video codec aware congestion control for real time video communication

Because of reinforcement learning's (RL) ability to automatically create more adaptive controlling logics beyond the hand-crafted heuristics, numerous effort has been made to apply RL to congestion control (CC) design for real time video communication (RTC) applications and has successfully shown promising benefits over the rule-based RTC CCs. Online reinforcement learning is often adopted to train the RL models so the models can directly adapt to real network environments. However, its trail-and-error manner can also cause catastrophic degradation of the quality of experience (QoE) of RTC application at run time. Thus, safeguard strategies such as falling back to hand-crafted heuristics can be used to run along with RL models to guarantee the actions explored in the training sensible, despite that these safeguard strategies interrupt the learning process and make it more challenging to discover optimal RL policies. The recent emergence of loss-tolerant neural video codecs (NVC) naturally provides a layer of protection for the online learning of RL-based congestion control because of its resilience to packet losses, but such packet loss resilience have not been fully exploited in prior works yet. In this paper, we present a reinforcement learning (RL) based congestion control which can be aware of and takes advantage of packet loss tolerance characteristic of NVCs via reward in online RL learning. Through extensive evaluation on various videos and network traces in a simulated environment, we demonstrate that our NVC-aware CC running with the loss-tolerant NVC reduces the training time by 41\% compared to other prior RL-based CCs. It also boosts the mean video quality by 0.3 to 1.6dB, lower the tail frame delay by 3 to 200ms, and reduces the video stalls by 20\% to 77\% in comparison with other baseline RTC CCs.

cs.NI↗

arXiv2024

OneAdapt: Fast Configuration Adaptation for Video Analytics Applications via Backpropagation

Deep learning inference on streaming media data, such as object detection in video or LiDAR feeds and text extraction from audio waves, is now ubiquitous. To achieve high inference accuracy, these applications typically require significant network bandwidth to gather high-fidelity data and extensive GPU resources to run deep neural networks (DNNs). While the high demand for network bandwidth and GPU resources could be substantially reduced by optimally adapting the configuration knobs, such as video resolution and frame rate, current adaptation techniques fail to meet three requirements simultaneously: adapt configurations (i) with minimum extra GPU or bandwidth overhead; (ii) to reach near-optimal decisions based on how the data affects the final DNN's accuracy, and (iii) do so for a range of configuration knobs. This paper presents OneAdapt, which meets these requirements by leveraging a gradient-ascent strategy to adapt configuration knobs. The key idea is to embrace DNNs' differentiability to quickly estimate the accuracy's gradient to each configuration knob, called AccGrad. Specifically, OneAdapt estimates AccGrad by multiplying two gradients: InputGrad (i.e. how each configuration knob affects the input to the DNN) and DNNGrad (i.e. how the DNN input affects the DNN inference output). We evaluate OneAdapt across five types of configurations, four analytic tasks, and five types of input data. Compared to state-of-the-art adaptation schemes, OneAdapt cuts bandwidth usage and GPU usage by 15-59% while maintaining comparable accuracy or improves accuracy by 1-5% while using equal or fewer resources.

cs.LG↗

arXiv2024

LLMSteer: Improving Long-Context LLM Inference by Steering Attention on Reused Contexts

As large language models (LLMs) show impressive performance on complex tasks, they still struggle with longer contextual understanding and high computational costs. To balance efficiency and quality, we introduce LLMSteer, a fine-tuning-free framework that enhances LLMs through query-independent attention steering. Tested on popular LLMs and datasets, LLMSteer narrows the performance gap with baselines by 65.9% and reduces the runtime delay by up to 4.8x compared to recent attention steering methods.

cs.LG↗

arXiv2024

Do Large Language Models Need a Content Delivery Network?

As the use of large language models (LLMs) expands rapidly, so does the range of knowledge needed to supplement various LLM queries. Thus, enabling flexible and efficient injection of new knowledge in LLM inference is critical. Three high-level options exist: (i) embedding the knowledge in LLM's weights (i.e., fine-tuning), (ii) including the knowledge as a part of LLM's text input (i.e., in-context learning), or (iii) injecting the KV caches of the new knowledge to LLM during prefill. This paper argues that, although fine-tuning and in-context learning are popular, using KV caches as the medium of knowledge could simultaneously enable more modular management of knowledge injection and more efficient LLM serving with low cost and fast response. To realize these benefits, we envision a Knowledge Delivery Network (KDN), a new system component in LLM services that dynamically optimizes the storage, transfer, and composition of KV cache across LLM engines and other compute and storage resources. We believe that, just like content delivery networks (CDNs), such as Akamai, enabled the success of the Internet ecosystem through their efficient data delivery, KDNs will be critical to the success of LLM applications through their efficient knowledge delivery. We have open-sourced a KDN prototype at https://github.com/LMCache/LMCache.

cs.CL↗

arXiv2024

NetLLM: Adapting Large Language Models for Networking

Many networking tasks now employ deep learning (DL) to solve complex prediction and optimization problems. However, current design philosophy of DL-based algorithms entails intensive engineering overhead due to the manual design of deep neural networks (DNNs) for different networking tasks. Besides, DNNs tend to achieve poor generalization performance on unseen data distributions/environments. Motivated by the recent success of large language models (LLMs), this work studies the LLM adaptation for networking to explore a more sustainable design philosophy. With the powerful pre-trained knowledge, the LLM is promising to serve as the foundation model to achieve "one model for all tasks" with even better performance and stronger generalization. In pursuit of this vision, we present NetLLM, the first framework that provides a coherent design to harness the powerful capabilities of LLMs with low efforts to solve networking problems. Specifically, NetLLM empowers the LLM to effectively process multimodal data in networking and efficiently generate task-specific answers. Besides, NetLLM drastically reduces the costs of fine-tuning the LLM to acquire domain knowledge for networking. Across three networking-related use cases - viewport prediction, adaptive bitrate streaming and cluster job scheduling, we showcase that the NetLLM-adapted LLM significantly outperforms state-of-the-art algorithms.

cs.NI↗

arXiv2024

CacheGen: KV Cache Compression and Streaming for Fast Large Language Model Serving

As large language models (LLMs) take on complex tasks, their inputs are supplemented with longer contexts that incorporate domain knowledge. Yet using long contexts is challenging, as nothing can be generated until the whole context is processed by the LLM. While the context-processing delay can be reduced by reusing the KV cache of a context across different inputs, fetching the KV cache, which contains large tensors, over the network can cause high extra network delays. CacheGen is a fast context-loading module for LLM systems. First, CacheGen uses a custom tensor encoder, leveraging KV cache's distributional properties to encode a KV cache into more compact bitstream representations with negligible decoding overhead, to save bandwidth usage. Second, CacheGen adapts the compression level of different parts of a KV cache to cope with changes in available bandwidth, in order to maintain low context-loading delay and high generation quality. % When available bandwidth drops, CacheGen may raise the compression level for a part of the context or recompute its KV cache on the fly. We test CacheGen on popular LLMs and datasets. Compared to the recent systems that reuse the KV cache, CacheGen reduces the KV cache size by 3.5-4.3x and the total delay in fetching and processing contexts by 3.2-3.7x with negligible impact on the LLM response quality. Our code is at: https://github.com/UChi-JCL/CacheGen.

cs.NI↗

arXiv2024

Eloquent: A More Robust Transmission Scheme for LLM Token Streaming

To render each generated token in real-time for users, the Large Language Model (LLM) server generates tokens one by one and streams each token (or group of a few tokens) through the network to the user right after generation, which we refer to as LLM token streaming. However, under unstable network conditions, the LLM token streaming experience could suffer greatly from stalls since one packet loss could block the rendering of later tokens even if the packets containing them arrive on time. With a measurement study, we show that current applications suffer from increased stalls under unstable networks. For this emerging token streaming problem in LLM Chatbots that differs from previous multimedia and text applications, we propose a novel transmission scheme, called Eloquent, which puts newly generated tokens as well as currently unacknowledged tokens in the next outgoing packet. This ensures that each packet contains some new tokens and, in the meantime, is independently rendered when received, avoiding the aforementioned stalls caused by missing packets. Through simulation under various networks, we show Eloquent reduces stall ratio (proportion of token rendering wait time) by 71.0% compared to the retransmission method commonly used by real chatbot applications and by 31.6% compared to the baseline packet duplication scheme. By tailoring Eloquent to fit the token-by-token generation of LLM, we enable the Chatbots to respond like an eloquent speaker for users to better enjoy pervasive AI.

cs.NI↗

arXiv2024

Earth+: on-board satellite imagery compression leveraging historical earth observations

With the increasing deployment of earth observation satellite constellations, the downlink (satellite-to-ground) capacity often limits the freshness, quality, and coverage of the imagery data available to applications on the ground. To overcome the downlink limitation, we present Earth+, a new satellite imagery compression system that, instead of compressing each image individually, pinpoints and downloads only recent imagery changes with respect to the history reference images. To minimize the amount of changes, it is critical to make reference images as fresh as possible. Earth+ enables each satellite to choose fresh reference images from not only its own history images but also past images of other satellites from an entire satellite constellation. To share reference images across satellites, Earth+ utilizes the limited capacity of the existing uplink (ground-to-satellite) by judiciously selecting and compressing reference images while still allowing accurate change detection. In short, Earth+ is the first to make reference-based compression efficient, by enabling constellation-wide sharing of fresh reference images across satellites. Our evaluation shows that Earth+ can reduce the downlink usage by a factor of 3.3 compared to state-of-the-art on-board image compression techniques while not sacrificing image quality, or using more on-board computing or storage resources, or more uplink bandwidth than currently available.

cs.NI↗

arXiv2024

GRACE: Loss-Resilient Real-Time Video through Neural Codecs

In real-time video communication, retransmitting lost packets over high-latency networks is not viable due to strict latency requirements. To counter packet losses without retransmission, two primary strategies are employed -- encoder-based forward error correction (FEC) and decoder-based error concealment. The former encodes data with redundancy before transmission, yet determining the optimal redundancy level in advance proves challenging. The latter reconstructs video from partially received frames, but dividing a frame into independently coded partitions inherently compromises compression efficiency, and the lost information cannot be effectively recovered by the decoder without adapting the encoder. We present a loss-resilient real-time video system called GRACE, which preserves the user's quality of experience (QoE) across a wide range of packet losses through a new neural video codec. Central to GRACE's enhanced loss resilience is its joint training of the neural encoder and decoder under a spectrum of simulated packet losses. In lossless scenarios, GRACE achieves video quality on par with conventional codecs (e.g., H.265). As the loss rate escalates, GRACE exhibits a more graceful, less pronounced decline in quality, consistently outperforming other loss-resilient schemes. Through extensive evaluation on various videos and real network traces, we demonstrate that GRACE reduces undecodable frames by 95% and stall duration by 90% compared with FEC, while markedly boosting video quality over error concealment methods. In a user study with 240 crowdsourced participants and 960 subjective ratings, GRACE registers a 38% higher mean opinion score (MOS) than other baselines.

cs.MM↗

arXiv2023

VidPlat: A Tool for Fast Crowdsourcing of Quality-of-Experience Measurements

For video or web services, it is crucial to measure user-perceived quality of experience (QoE) at scale under various video quality or page loading delays. However, fast QoE measurements remain challenging as they must elicit subjective assessment from human users. Previous work either (1) automates QoE measurements by letting crowdsourcing raters watch and rate QoE test videos or (2) dynamically prunes redundant QoE tests based on previously collected QoE measurements. Unfortunately, it is hard to combine both ideas because traditional crowdsourcing requires QoE test videos to be pre-determined before a crowdsourcing campaign begins. Thus, if researchers want to dynamically prune redundant test videos based on other test videos' QoE, they are forced to launch multiple crowdsourcing campaigns, causing extra overheads to re-calibrate or train raters every time. This paper presents VidPlat, the first open-source tool for fast and automated QoE measurements, by allowing dynamic pruning of QoE test videos within a single crowdsourcing task. VidPlat creates an indirect shim layer between researchers and the crowdsourcing platforms. It allows researchers to define a logic that dynamically determines which new test videos need more QoE ratings based on the latest QoE measurements, and it then redirects crowdsourcing raters to watch QoE test videos dynamically selected by this logic. Other than having fewer crowdsourcing campaigns, VidPlat also reduces the total number of QoE ratings by dynamically deciding when enough ratings are gathered for each test video. It is an open-source platform that future researchers can reuse and customize. We have used VidPlat in three projects (web loading, on-demand video, and online gaming). We show that VidPlat can reduce crowdsourcing cost by 31.8% - 46.0% and latency by 50.9% - 68.8%.

cs.HC↗

arXiv2023

Estimating WebRTC Video QoE Metrics Without Using Application Headers

The increased use of video conferencing applications (VCAs) has made it critical to understand and support end-user quality of experience (QoE) by all stakeholders in the VCA ecosystem, especially network operators, who typically do not have direct access to client software. Existing VCA QoE estimation methods use passive measurements of application-level Real-time Transport Protocol (RTP) headers. However, a network operator does not always have access to RTP headers in all cases, particularly when VCAs use custom RTP protocols (e.g., Zoom) or due to system constraints (e.g., legacy measurement systems). Given this challenge, this paper considers the use of more standard features in the network traffic, namely, IP and UDP headers, to provide per-second estimates of key VCA QoE metrics such as frames rate and video resolution. We develop a method that uses machine learning with a combination of flow statistics (e.g., throughput) and features derived based on the mechanisms used by the VCAs to fragment video frames into packets. We evaluate our method for three prevalent VCAs running over WebRTC: Google Meet, Microsoft Teams, and Cisco Webex. Our evaluation consists of 54,696 seconds of VCA data collected from both (1), controlled in-lab network conditions, and (2) real-world networks from 15 households. We show that the ML-based approach yields similar accuracy compared to the RTP-based methods, despite using only IP/UDP data. For instance, we can estimate FPS within 2 FPS for up to 83.05% of one-second intervals in the real-world data, which is only 1.76% lower than using the application-level RTP headers.

cs.NI↗

arXiv2023

Automatic and Efficient Customization of Neural Networks for ML Applications

ML APIs have greatly relieved application developers of the burden to design and train their own neural network models -- classifying objects in an image can now be as simple as one line of Python code to call an API. However, these APIs offer the same pre-trained models regardless of how their output is used by different applications. This can be suboptimal as not all ML inference errors can cause application failures, and the distinction between inference errors that can or cannot cause failures varies greatly across applications. To tackle this problem, we first study 77 real-world applications, which collectively use six ML APIs from two providers, to reveal common patterns of how ML API output affects applications' decision processes. Inspired by the findings, we propose ChameleonAPI, an optimization framework for ML APIs, which takes effect without changing the application source code. ChameleonAPI provides application developers with a parser that automatically analyzes the application to produce an abstract of its decision process, which is then used to devise an application-specific loss function that only penalizes API output errors critical to the application. ChameleonAPI uses the loss function to efficiently train a neural network model customized for each application and deploys it to serve API invocations from the respective application via existing interface. Compared to a baseline that selects the best-of-all commercial ML API, we show that ChameleonAPI reduces incorrect application decisions by 43%.

cs.SE↗

arXiv2022

Enabling Personalized Video Quality Optimization with VidHoc

The emerging video applications greatly increase the demand in network bandwidth that is not easy to scale. To provide higher quality of experience (QoE) under limited bandwidth, a recent trend is to leverage the heterogeneity of quality preferences across individual users. Although these efforts have suggested the great potential benefits, service providers still have not deployed them to realize the promised QoE improvement. The missing piece is an automation of online per-user QoE modeling and optimization scheme for new users. Previous efforts either optimize QoE by known per-user QoE models or learn a user's QoE model by offline approaches, such as analysis of video viewing history and in-lab user study. Relying on such offline modeling is problematic, because QoE optimization will start late for collecting enough data to train an unbiased QoE model. In this paper, we propose VidHoc, the first automatic system that jointly personalizes QoE model and optimizes QoE in an online manner for each new user. VidHoc can build per-user QoE models within a small number of video sessions as well as maintain good QoE. We evaluate VidHoc in a pilot deployment to fifteen users for four months with the care of statistical validity. Compared with other baselines, the results show that VidHoc can save 17.3% bandwidth while maintaining the same QoE or improve QoE by 13.9% with the same bandwidth.

cs.NI↗