Related papers: Slice-Level Scheduling for High Throughput and Loa…

LAPS identifies and disaggregates requests with different prompt lengths in LLM serving to reduce TTFT latency. While recent systems have decoupled the prefill and decode stages to improve throughput, they still rely on unified scheduling…

Speculative decoding (SD) accelerates large language model (LLM) inference by using a smaller draft model to propose multiple tokens that are verified by a larger target model in parallel. While prior work demonstrates substantial speedups…

User experience is a critical factor Large Language Model (LLM) serving systems must consider, where service level objectives (SLOs) considering the experience of individual requests and system level metrics (SLMs) considering the overall…

Conversational recommender systems (CRSs) aim to recommend high-quality items to users through a dialogue interface. It usually contains multiple sub-tasks, such as user preference elicitation, recommendation, explanation, and item…

Dynamically scheduled high-level synthesis (HLS) enables the use of load-store queues (LSQs) which can disambiguate data hazards at circuit runtime, increasing throughput in codes with unpredictable memory accesses. However, the increased…

While model serving has unlocked unprecedented capabilities, the high cost of serving large-scale models continues to be a significant barrier to widespread accessibility and rapid innovation. Compiler optimizations have long driven…

Meeting stringent Time-To-First-Token (TTFT) requirements is crucial for LLM applications. To improve efficiency, modern LLM serving systems adopt disaggregated architectures with diverse parallelisms, introducing complex multi-stage…

Dynamically scheduled high-level synthesis (HLS) achieves higher throughput than static HLS for codes with unpredictable memory accesses and control flow. However, excessive dataflow scheduling results in circuits that use more resources…

Recent advances with large language models (LLM) illustrate their diverse capabilities. We propose a novel algorithm, staged speculative decoding, to accelerate LLM inference in small-batch, on-device scenarios. We address the low…

In the realm of AI, large language models (LLMs) like GPT-4, central to the operation of AI agents, predominantly operate in the cloud, incurring high operational costs. With local-based small language models (SLMs) becoming more accurate,…

LLM inference serving typically scales out with a two-tier architecture: a cluster router distributes requests to multiple inference engines, each of which then in turn performs its own internal scheduling. However, this commonly used…

The deployment and scaling of large language models (LLMs) have become critical as they permeate various applications, demanding high-throughput and low-latency serving systems. Existing frameworks struggle to balance these requirements,…

Large Language Models (LLMs) increasingly rely on inference-time reasoning algorithms such as chain-of-thought and multi-branch reasoning to improve accuracy on complex tasks. These methods, however, substantially increase token usage and…

Federated fine-tuning of on-device large language models (LLMs) mitigates privacy concerns by preventing raw data sharing. However, the intensive computational and memory demands pose significant challenges for resource-constrained edge…

Test-time scaling (TTS) -- the dynamic allocation of compute during inference -- is a promising direction for improving reasoning in large language models (LLMs). However, a systematic comparison of well-known TTS strategies under identical…

In this paper, we consider a mixed-prompt scenario for a large language model (LLM) inference serving system that supports diverse applications with both short prompts and long prompts and heterogeneous SLOs for iteration time. To improve…

Split learning emerges as a promising paradigm for collaborative distributed model training, akin to federated learning, by partitioning neural networks between clients and a server without raw data exchange. However, sequential split…

Many shared-memory parallel irregular applications, such as sparse linear algebra and graph algorithms, depend on efficient loop scheduling (LS) in a fork-join manner despite that the work per loop iteration can greatly vary depending on…

High-Level Synthesis (HLS) improves IC development productivity by enabling hardware design from C-like languages. However, strict coding constraints and design-specific optimizations limit its widespread adoption. While recent efforts…

While Large Language Models (LLMs) have achieved remarkable success in various fields, the efficiency of training and inference remains a major challenge. To address this issue, we propose SUBLLM, short for Subsampling-Upsampling-Bypass…