The open source wave is rushing forward. With the characteristics of equality, openness, collaboration, and sharing, it optimizes the software production model, improves development efficiency, and helps enterprises reduce costs and increase efficiency. Open source has become the source of global information technology innovation, leading the development of global technological innovation.
Over the years, Intel has been practicing the concept of open source and actively promoting the construction of a global open source community. It has participated in more than 700 foundations and standard formulations around the world, has nearly 70,000 global patent assets, and has created more than 450 software tools and application solutions for developers. Intel's influence in the open source field continues to expand, and it has always been one of the important contributors in the open source community.
This year, Intel China also established an Open Source Technology Committee, dedicated to co-building China’s open source software ecosystem. For more information on Intel technology, come to GOSIMConference to reveal it to you!
On September 26, the GOSIM Conference, hosted by GOSIM Open Source Innovation Center and co-organized by CSDN, will be held at the Shanghai Transnational Sourcing Convention and Exhibition Center. It will set up five major technology forums on artificial intelligence, Rust language, mobile and Internet, automobiles and robots, and metaverse and games. Inviting top technology creators and leaders at home and abroad to share cutting-edge insights and best practices in the open source field, and provide everyone with a high-quality technology feast. (GOSIM Conference will be held concurrently with KubeCon + CloudNativeCon + Open Source Summit China 2023)
As a strategic sponsor of the GOSIM Conference, Intel will make a major appearance at this event. Senior personnel and technical experts from Intel will deliver a number of important speeches to share Intel's latest technological progress and innovation results.
09:00-10:00 Keynote Speech | Main Forum
Lian Liping
Guest profile: Intel's senior R&D director of open source cloud software. She leads the development of the company's cloud native software strategy and engineering team and is responsible for driving Intel's contribution to cloud computing software and cooperation with the open source community. Currently, her team is active in multiple software projects and working groups at CNCF. She has been engaged in the research and development of underlying system software for the past two decades, leading global development teams and making significant contributions in areas such as UEFI, Linux, HTML5, Android and cloud computing. In 2001, she established Intel's first Linux development team in China and developed it into the Intel China Open Source Technology Center.
15:15-15:45 Artificial Intelligence Forum
Hu Yongwen
Guest profile: Graduated from KTH Royal Institute of Technology in Sweden, Intel artificial intelligence framework engineer, focusing on the development of deep learning and large language model acceleration framework. Not only has she shared teachings related to LLM many times, she is also one of the core team members of the BigDL project.
Sharing topic: "BigDL-LLM: A low-bit LLM library that can run on Intel CPU and GPU"
Topic introduction: BigDL-LLM is an open source library designed for Intel laptops and GPUs to run LLM. It can use low-bit optimization (INT3/INT4/NF4/INT5/INT8) to achieve very low latency. BigDL-LLM is built on various technologies (such as llama.cpp, GPTQ, bitsandbytes, QLoRa, etc.) and optimized on Intel CPU/GPU. By using BigDL-LLM, users can easily build and deploy LLM applications on Intel laptops using standard PyTorch APIs such as Hugging face Transformers and LangChain. It is worth mentioning that we have successfully verified and optimized a variety of models on BigDL-LLM, such as LLaMA/LLaMA2, ChatGLM/ChatGLM2, MPT, Falcon, MOSS, Baichuan, Tongyi Qianwen, Dolly/Dolly-v2, RedPajama , StarCoder and Whisper, etc.
13:45-14:15 Automobile and Robot Forum
Mike Roberts
Guest Profile : Research Scientist, Intel Intelligent Systems Lab, working with Vladlen Koltun. I'm interested in using realistic synthetic data in computer vision. Previously, I was a research scientist at Apple, where I led the development of the Hypersim dataset.
Sharing topic: "Spear: A realistic simulator for embodied intelligence research"
Topic Introduction : Interactive simulators are becoming powerful tools for training embodied artificial intelligence agents, but existing simulators are limited in content diversity, physical interactivity, and visual fidelity. This issue discusses our ongoing efforts to address these limitations. Specifically, I will discuss a new open source simulator we are developing called "SPEAR - A Simulator for Photorealistic Embodied AI Research". To develop SPEAR, we worked extensively with a team of professional artists to build multiple highly realistic virtual indoor environments containing thousands of unique objects that can be individually manipulated. The environment is implemented as an Unreal Engine asset, providing a high-level Python interface for interacting with robots in the environment. I'll provide an overview of Spear's current capabilities, demonstrate a sample application, and discuss the roadmap for the upcoming v0.4.0 release and beyond. Spear is available online at https://github.com/isl-org/spear.
We sincerely invite you who love open source to work with Intel to gather open source innovation and see a new future of open source!
