https://mlsys-ucsd.org//events We are a group of faculty, researchers, and students pushing the frontiers of systems for machine learning (ML) and artificial intelligence (AI) and their applications. en-us address@yoursite.com (UCSD-MLsys) address@yoursite.com (UCSD-MLsys) Thu, 06 Mar 2025 18:30:00 GMT https://mlsys-ucsd.org//events/seminar_2025_0306 https://mlsys-ucsd.org//events/seminar_2025_0306 This talk presents the EAGLE series, a groundbreaking approach to accelerating large language model inference without compromising output quality. Instead of traditional token-level processing, EAGLE operates at the structured feature level and incorporates sampling results to reduce uncertainty. The technology has gained significant industry adoption, with integration into major frameworks including vLLM, SGLang, TensorRT-LLM, and several others from AWS and Intel. Thu, 06 Mar 2025 18:30:00 GMT address@yoursite.com (UCSD-MLsys) MLSys Seminar https://mlsys-ucsd.org//events/seminar_2025_0220 https://mlsys-ucsd.org//events/seminar_2025_0220 The LLM360 project advances AI through open-source foundation models and datasets. This talk explores key initiatives including K2, the most capable fully open-source language model, and TxT360, examining the true meaning of open source while proposing new approaches to academic and industry collaboration in open-source AI. Thu, 20 Feb 2025 09:00:00 GMT address@yoursite.com (UCSD-MLsys) MLSys Seminar https://mlsys-ucsd.org//events/seminar_2025_0206 https://mlsys-ucsd.org//events/seminar_2025_0206 In this talk, we will discuss the lessons learned in building an efficient large language model deployment system for both server and edge settings. We will cover general techniques in machine learning compilation and system support for efficient structure generation. We will also discuss the future opportunities in system co-design for cloud-edge model deployments. Thu, 06 Feb 2025 09:00:00 GMT address@yoursite.com (UCSD-MLsys) MLSys Seminar https://mlsys-ucsd.org//events/sigsoft_award_2024 https://mlsys-ucsd.org//events/sigsoft_award_2024 Our ISSTA'24 paper "Multi-modal Learning for WebAssembly Reverse Engineering" received an ACM SIGSOFT Distinguished Paper Award. Thu, 12 Sep 2024 00:00:00 GMT address@yoursite.com (UCSD-MLsys) news https://mlsys-ucsd.org//events/mlsys_rising_star_2024 https://mlsys-ucsd.org//events/mlsys_rising_star_2024 Congratulations to MLsys group student Hanxian Huang on being selected as a 2024 MLCommons Rising Star. She was among the 41 junior researchers selected from over 170 applicants globally. The MLCommons Rising Stars are selected based on their excellence in Machine Learning (ML) and Systems research and stand out for their current and future contributions and potential. Mon, 01 Jul 2024 00:00:00 GMT address@yoursite.com (UCSD-MLsys) news https://mlsys-ucsd.org//events/seminar_5_9 https://mlsys-ucsd.org//events/seminar_5_9 Numerous domain-specific accelerators have been developed recently to address the growing computational needs of machine learning, and the success of these DSAs hinges on effective ML compilers like Google's XLA, which enhances ML performance on various hardware and supports multiple frameworks, and is further advanced through collaborative development in OpenXLA. Thu, 09 May 2024 17:00:00 GMT address@yoursite.com (UCSD-MLsys) MLSys Seminar https://mlsys-ucsd.org//events/seminar_4_30 https://mlsys-ucsd.org//events/seminar_4_30 PyTorch 2 leverages new technologies like TorchDynamo and TorchInductor to significantly enhance training and inference speeds without compromising its ease of use, flexibility, and Pythonic environment. TorchDynamo optimizes unmodified PyTorch code at the Python bytecode level, while TorchInductor translates programs for efficient execution on GPUs and CPUs, maintaining the dynamism inherent in PyTorch and allowing for easy user customization. Tue, 30 Apr 2024 17:00:00 GMT address@yoursite.com (UCSD-MLsys) MLSys Seminar https://mlsys-ucsd.org//events/seminar_4_29 https://mlsys-ucsd.org//events/seminar_4_29 With the ubiquitous use-cases of modern LLMs, the deployment scale of these models is unforeseen. This has led to a large-scale datacenter expansion with GPUs, currently running into an energy wall worldwide. This talk will focus on the properties of generative LLMs that can be used to make the deployment of these models more power-efficient. The talk will also introduce POLCA and Splitwise, two techniques to reduce the power consumption for the LLM serving. Mon, 29 Apr 2024 12:00:00 GMT address@yoursite.com (UCSD-MLsys) MLSys Seminar https://mlsys-ucsd.org//events/mlsys_launch https://mlsys-ucsd.org//events/mlsys_launch 🎉 MLSys @UCSD is launched! Mon, 15 Apr 2024 00:00:00 GMT address@yoursite.com (UCSD-MLsys) news